Descemet Stripping Only in Fuchs Endothelial Corneal Dystrophy: Results of a Randomized Clinical Trial of Topical Ripasudil and Directions for Future Innovation

Purpose

To review history of Descemet stripping only (DSO) in Fuchs endothelial corneal dystrophy, describe the results of a clinical trial of topical ripasudil after DSO (K-321-201 study), and discuss future directions.

Methods

A 1-year, phase 2, randomized, placebo-controlled multicenter clinical trial of two doses of K-321 (ripasudil) administered for 12 weeks after DSO surgery in Fuchs endothelial corneal dystrophy was performed. The primary endpoint, central corneal endothelial cell density (ECD) at 12 weeks after surgery, was determined by an independent reading center that was masked to study group assignment. Duration of corneal edema, need for medical or surgical rescue therapy, corneal thickness, and central ECD throughout the entire study period were also examined. Adverse events and exploratory endpoints were collected.

Results

Sixty-five subjects were enrolled (21 in the QID group, 22 in the BID, and in the placebo group). Over 95% of subjects completed the trial. The QID group had a higher central ECD 12 weeks after DSO than the placebo group (531 ± 312 cells/mm ² vs 228 ± 298 cells/mm ², P =.0065). Corneal edema cleared in 17/21 (81.0%) of the QID group at 12 weeks, compared with 2/22 (9.1%) of the placebo group ( P <.0001). Rescue was required in 2/21 (9.5%) subjects in the QID group and 6/22 (27.3%) subjects in the placebo group ( P =.0092). Adverse events were mild and did not lead to discontinuation of treatment.

Conclusions

Topical K-321 given QID improves DSO outcomes, as demonstrated by a higher ECD, more rapid resolution of corneal edema, and reduced failure rate. The medication was well-tolerated.

INTRODUCTION

F uchs endothelial corneal dystrophy (FECD) is an important corneal disease, affecting from 4% to 7% of people in the United States (US) and accounting for approximately one-third of corneal transplants each year. ^,^, Described over 100 years ago, FECD is a fascinating condition in which progressive bilateral endothelial dysfunction occurs in association with the formation of guttae (excrescences on Descemet membrane), beginning in the central cornea. As the disease progresses, corneal edema and reduced vision, especially in the morning, necessitate corneal transplantation, the mainstay of current treatment. Progress has been made in unraveling the pathophysiology of FECD, with pathways such as mitochondrial dysfunction, oxidative stress, unfolded protein response, apoptosis, RNA toxicity, ferroptosis and endothelial-mesenchymal transition purported to play a role, although the exact contribution of each remains unknown at present. ^,^,^,^,^,^,^,^,^, Generally phenotypically homogeneous, FECD is a genetically heterogenous condition, with multiple genes associated with the disease. ^,^,^, Up to three-quarters of FECD cases result from expansion of a CTG nucleotide repeat in the TCF4 gene, making it the most common genetic association. ^, Various medical therapies are under development for FECD, which may be used independently or in combination with some type of corneal surgery. ^,^,^,^,^, Cultured human corneal endothelial cells, already approved in Japan as a treatment for FECD, are currently being studied in clinical trials in the US. ^,^,^,

Surgical treatment of FECD has had a remarkable evolution since the 1990s, when full-thickness penetrating keratoplasty (PK) was the surgical treatment of choice. Now, most FECD patients requiring transplantation are treated by selective endothelial keratoplasty (EK), in which there is replacement of the endothelium and Descemet membrane with stroma (Descemet stripping EK) or without (Descemet membrane EK [DMEK]). EK was a major advance over PK. By eliminating corneal sutures, EK allowed preservation of normal corneal topography, resulting in faster recovery and better quality of vision. Additionally, EK has a lower rejection rate than PK. DMEK is now the most commonly performed surgical technique for FECD in the US.

The corneal endothelium, which is responsible for maintaining corneal deturgescence via energy-requiring ion pumps and a passive barrier function, has interesting characteristics that are relevant to FECD. Derived from neuroectoderm, the corneal endothelium does not typically divide after birth, although endothelial cells can be stimulated to divide in vitro , a finding that underlies the cultured cell transplantation therapies in development. When endothelial cells are damaged or become dysfunctional (eg, after intraocular surgery or in dystrophic conditions), the remaining cells enlarge and spread to reconstitute an intact endothelial layer. When migrating cells reach a neighboring cell, they stop spreading, so-called “contact-inhibition.” After endothelial damage, the central endothelial cell density (ECD) is reduced, and the cells show increased variation in size (polymegethism) and shape (pleomorphism) on imaging.

Approximately 20 years ago, evidence began to accumulate suggesting that, in fact, the corneal endothelium in FECD could “self-rejuvenate” after removal of or damage to the central corneal endothelium. Case reports demonstrated corneal clearance despite deliberate (interrupted EK) and inadvertent (following cataract surgery) removal of Descemet membrane and endothelium in healthy corneas and those with FECD. ^,^,^,^, Koizumi et al published a case report of corneal clearance in FECD after destruction of the endothelium with cryotherapy and topical application of a Rho-associated kinase (ROCK) inhibitor. Dirisamer et al reported corneal clearance despite EK graft dehiscence in FECD but not in pseudophakic bullous keratopathy, suggesting that the remaining peripheral Fuchs endothelium retained some capacity to promote corneal deturgescence.

Descemet stripping only (DSO), the surgical removal of the central endothelium in FECD without placement of a corneal graft, is not a new idea, having been attempted in the past as early as 1953, without adoption by the cornea community. Early reports of deliberate descemetorhexis showed inconsistent results with a high chance of failure or corneal scarring but added to the increasing body of evidence that the corneal endothelium in FECD retained some capacity for regeneration. ^,^,

The era of successful DSO began in 2014 when two surgeons, unknown to each other and located in different countries, independently performed a series of deliberate central descemetorhexis as a treatment for FECD. ^, Borkar and coworkers were the first to show promising outcomes, with corneal clearance in 10/13 eyes that underwent a 4 mm descemetorhexis with concurrent cataract surgery. Patients in this series demonstrated differences in the timing of corneal clearance after DSO, with some patients healing within a few weeks, most in a few months, and some in 4 to 6 months, suggesting individual variation in DSO response. It is still not clear what underlies this variation, which may be related to the genetic profile, the number of trinucleotide repeats present or an as-yet unidentified disease or patient characteristic.

The following year, Moloney et al showed outright success of DSO in 9 of 12 eyes, with two of the remaining eyes clearing after treatment with topical ripasudil, a ROCK inhibitor approved in Japan for the treatment of glaucoma. The final eye did not clear after treatment with a compounded ROCK inhibitor and required keratoplasty. Subsequent DSO series have demonstrated a higher success rate, with some reports of 100% clearance rate with a “peeling” technique that produces a smooth-edged descemetorhexis. ^,^,^,

Following initial skepticism among corneal specialists, more groups began to explore DSO as a surgical option for FECD. In an investigator-initiated, placebo-controlled study, Macsai and Shiloach demonstrated faster endothelial healing and higher ECD in DSO patients given adjuvant topical ripasudil. Huang and coauthors compared DSO to DMEK, showing fewer complications with DSO and equivalent visual acuity results, although the DSO patients took longer to reach their final vision. This group, and another, have recently published their 5-year outcomes, showing durability of DSO. ^,

Laboratory evidence supports the beneficial effects of ROCK inhibitors on healing of the corneal endothelium. In animal models, application of topical ROCK after endothelial injury, by either mechanical scraping or freezing, improved corneal ECD and sped healing. ^, In in vitro and ex vivo human endothelial models, ripasudil improved endothelial barrier and pump function.

Given the potential of DSO as a nontransplant treatment option for FECD and the mounting evidence regarding the beneficial effects of ROCK inhibitors on the corneal endothelium, Kowa Research Institute, Inc, the manufacturer of ripasudil, approached us about the feasibility of developing this compound to aid endothelial healing after DSO. Following a multiyear planning process, including a delay due to the COVID-19 pandemic, the K-321-201 study enrolled its first patient in June 2020. Herein, we describe the results of this trial and discuss remaining questions and future directions for DSO.

METHODS

study design

The K-321-201 trial was a double-masked, randomized, placebo-controlled, parallel-group, 12-week, phase 2 study whose goal was to investigate the safety and efficacy of topical K-321 ophthalmic solution (ripasudil 0.4%) after DSO in patients with FECD (Clinicaltrials.gov registration# NCT04250207). There were 38 study sites in 5 countries: the US (28 sites), Spain (4 sites), Germany (2 sites), Denmark (2 sites), and Australia (2 sites). The study took place from June 23, 2020, to June 27, 2022. Institutional Review Board (IRB) approval was granted for this study, either through a central IRB or a local IRB, as dictated by study site policy. Written informed consent was obtained from all subjects.

The primary outcome measure was the effect of topical K-321 on the central ECD 12 weeks after DSO, determined by a masked independent reading center, the Corneal Image Analysis Reading Center at Case Western Reserve University, and University Hospitals Eye Institute (CIARC). Secondary endpoints included the effect of topical K-321 on central ECD, corneal thickness, and corneal edema at each visit for 52 weeks following DSO. The time to intervention of medical or surgical rescue therapy was also examined. The effect of K-321 on visual acuity and morphologic parameters of the corneal endothelium, including coefficient of variation (CV) and hexagonality (HEX), as well as vision-related quality of life assessed with the Visual Function Questionnaire 25 (VFQ-25), was also explored. Safety and tolerability of K-321 were assessed at each visit.

The study consisted of a 12-week treatment period followed by a 2-week tapering phase, and an additional 38-week follow-up period. The study had 3 arms: placebo four times daily (QID; placebo group), K-321 twice daily with placebo twice daily (BID group), and K-321 four times daily (QID group).

Subjects were screened for eligibility more than 1 week but up to 4 weeks prior to DSO surgery. Inclusion and exclusion criteria are summarized in Table 1 . On postoperative day 1 (POD1), subjects were randomized using a 1:1:1 ratio once a postoperative photo confirmed that the descemetorhexis fell within the defined inclusion parameters. An interactive web response system was used to administer the randomization schedule, which was generated by an independent biostatistician using SAS software Version 9.4 or later (SAS Institute Inc).

TABLE 1

Study Inclusion and Exclusion Criteria.

Inclusion Criteria	Exclusion Criteria
1. Minimum 18 y of age at screening visit	1. Study eye with confluent guttae outside the stripped area
2. Diagnosis of FECD at screening visit	2. Advanced stromal edema in the study eye, characterized by widespread corneal haze of bullae
3. Study eye with confluent central guttae of ≤5 mm	3. Study eye with a central corneal thickness of ≥670 µm
4. Study eye with BCVA of ≤75 letters by ETDRS visual acuity testing	4. Study eye with a history of any intraocular surgery besides cataract surgery
5. Negative urine pregnancy test on day of surgery for patients of child-bearing potential and agreement to use effective contraception during the study period	5. Study eye that underwent cataract surgery within 90 d prior to study enrollment 6. Nonstudy eye with a history of any intraocular surgery within 30 d of study enrollment
6. Study eye with a descemetorhexis diameter of 4.5-5.5 mm, measure at the completion of surgery	7. Plan to have any other intraocular surgery in the study eye (except the descemetorhexis) during the study
	8. Plan to have surgical treatment of FECD or cataract in the nonstudy eye during the screening and treatment periods
	9. Presence of any other clinically significant ocular condition besides FECD or cataract that requires surgery or medication
	10. Presence of any known ocular condition (such as infection) that could interfere with the performance of descemetorhexis 11. Presence of diabetes with a hemoglobin A1c > 8.5%
	12. Previous use of any topical Rho kinase inhibitor
	13. History of hypersensitivity to any ophthalmic medication used for diagnosis or treatment
	14. Known hypersensitivity to any component of the study medication
	15. Use of any investigational medication within 30 d of the screening visit
	16. Positive urine test for drugs of abuse or alcohol at screening visit

BCVA = best corrected visual acuity; ETDRS = Early Treatment Diabetic Retinopathy Study; FECD = Fuchs endothelial corneal dystrophy.

Study medications were dispensed at the POD1 visit. The first dose was applied by study staff at POD1, and all subsequent doses were administered by the study subject. Used drug was retained by the subjects and collected by study staff at study visits to assess compliance. To minimize medication errors during the 12-week treatment phase in BID group, identical containers of study medication or placebo were color-coded differently for morning/night and mid-day/evening doses. To preserve masking, all subjects self-administered their assigned study agent four times daily from 2 sets of assigned containers, one to be used morning/night and the other to be used mid-day/evening. On each study visit day during the treatment period, patients were instructed not to apply any study drug until after all study measurements had been completed, due to the fact that ripasudil has been noted to change the morphology of endothelial cells for several hours following administration, which would make accurate measurement of the ECD difficult. After the 12-week treatment period, the drops were tapered to twice daily for 1 week and then once daily for 1 week. The study medication was stopped at week 14. Subjects were followed for 38 weeks after medication cessation, for a total of 52 weeks after the surgery.

surgical technique

An instruction video demonstrating the recommended DSO surgical technique was shared with all study surgeons (Supplemental Video 1). Following application of 5% povidone-iodine to the conjunctiva and 5 or 10% povidone-iodine to the eyelids and peri‑ocular area, ophthalmic anesthesia was given according to surgeon preference (topical or subtenon’s injection with or without intravenous sedation). No intracameral anesthesia was permitted. Gentian violet was used to mark the anterior corneal surface to delineate the area of guttae to be removed. A 2.5 to 2.75 mm incision was made according to surgeon preference. A cohesive viscoelastic was used to maintain the anterior chamber. A continuous curvilinear descemetorhexis (4.5-5.5 mm in diameter) encompassing the area of central confluent guttae was performed according to surgeon preference, with special attention paid to avoid leaving any tags of Descemet membrane. The viscoelastic was removed using irrigation and aspiration. A 10.0 nylon suture could be placed in the wound if the surgeon wished. The size of the completed descemetorhexis was measured with calipers, and a still photograph was taken. Postoperative medications, including topical prednisolone acetate 1% and a fluoroquinolone antibiotic of choice, were applied. No concomitant surgical procedures were performed in this trial. A video recording of the surgery was made.

endothelial cell analysis

Prior to the start of the trial, each site submitted certification images to CIARC for its specular microscope(s) and all staff performing study imaging. Imaging included 5 central, 3 nasal, and 3 temporal images. Nasal and temporal images were captured at the 10 and 2 o’clock midperipheral positions (2-4 mm from center, depending upon the machine used for imaging). Images from the study and fellow eye were taken at the screening visit and at the end of study (week 52) visit. The study eye was imaged at all visits except on the day of surgery and the POD1 exam. All endothelial cell images at each site were obtained using the same specular microscope, with efforts made to have all images obtained by the same imager.

Deidentified images were sent to CIARC for analysis through a secure portal. CIARC personnel were masked to study group assignment. Analyses included ECD, CV, and HEX, which were determined using previously described protocols. ^, Konan CellChek D analytics software (Konan Medical) utilizing the Konan Center and Flex-Center methods were used to analyze 1 to 4 frames and a preferred minimum of 5 contiguous cells per image, dependent on image cell size, image quality, pathology, and number of available endothelial cells. Two independent graders analyzed 5 central and 3 midperipheral nasal and temporal images for each time point, with adjudication by a third reader if the two readers’ ECD differed by 10% or more. The final cell density for each image was the average of that reported by the two independent readers when there was agreement (<10% difference in ECD). If adjudication was needed, the final cell density was the average of that reported by the adjudicator and any reader in agreement. The final cell density for each location was the average of the final ECD for each individual image for that location.

other study evaluations

Central corneal thickness (CCT) in both eyes was measured using the same ultrasonic pachymeter throughout the trial, with efforts made to have the same examiner do the measurements at each visit. Three measurements were taken, and the average recorded in microns. Corneal edema in the study eye was assessed by slit lamp examination by the study investigator and reported based on location (epithelial, stromal, or endothelial).

Best corrected visual acuity (BCVA) was measured through an undilated pupil by certified examiners in a certified Early Treatment of Diabetic Retinopathy Study (ETDRS) exam room. BCVA was recorded as an ETDRS letter score. Both eyes were measured at the screening visit and at the end of study (week 52) visit. BCVA was measured in the study eye at all visits except on the day of surgery and on POD1.

Assessment of visual function was done using an Interviewer Administered Format of National Eye Institute Visual Functioning Questionnaire-25 (VFQ-25, version 2000), given by the same examiner, when possible, at screening, and weeks 5, 12, 24, and 52.

safety assessments

At each study visit, subjects were asked about any changes in their medical health, including hospitalizations, accidents, eye surgeries, new medications, or changes in existing medications (both prescription and over-the-counter medications). Adverse events (AEs) were also identified from study data, including laboratory values or physical examination findings. The information collected included time of onset, duration of the event, date of resolution of the event, seriousness, severity, relatedness to study drug, any required treatment or evaluations, and outcome. Severity was assessed by the site investigator according to standard definitions (mild, moderate, severe). The causal relationship of the study agent to the AE was also assessed by the site investigator (unrelated or related). The Medical Dictionary for Regulatory Activities (MedDRA) was used to code all AEs.

Serious AEs, defined as events causing death, hospitalization, persistent or significant disability, or which were immediately life-threatening, were required to be reported to the sponsor within 24 hours, unless they occurred more than 30 days after the longer of end of treatment (week 12) or end of study (week 52) visits. Any Serious AE’s thought to be related to the study drug were reported to the sponsor no matter what the timing of the event. All AEs were followed to adequate resolution.

treatment interruption/discontinuation and study withdrawal

Interruption of treatment was defined as a temporary stoppage of the study drug, due to an AE or any other reason, which resumed during the treatment period. Early discontinuation of treatment was defined as a permanent stoppage of the study drug before completion of the week 12 visit. Early discontinuation of treatment was allowed if the site investigator judged it was medically imperative, eg, if the study subject did not tolerate the dosing regimen or if it was thought to jeopardize the subject’s health. Subjects who had early discontinuation of treatment were encouraged to undergo all scheduled study site visits and assessments. Where possible, study data were collected for any subjects who had early discontinuation of treatment. Early withdrawal from the study was defined as failing to complete the week 52 (end of study) visit. Data from subjects who withdrew from the study were included using the Last Observation Carried Forward method.

rescue treatment

Corneal transplantation (endothelial or PK) could be offered as rescue surgical procedure to subjects whose endothelial healing was not complete at the week 12 visit. Early surgical rescue was allowed if the investigator decided that the study subject required urgent therapy. Study treatment was discontinued immediately (without tapering) if rescue surgery was performed. Subjects who underwent rescue surgery continued with all planned study visits until the end of study (week 52) visit.

statistical analysis

Using simulation, a sample size of 20 patients in each group was calculated to detect a difference of 323 in the measurement in the corneal ECD at week 12 between K-321 0.4% QID/BID and placebo groups with a power of 83.3% for Step 1 and a power of 71.9% for Step 2 using a Wilcoxon rank sum test with a 0.050 two-sided significance level with closed testing procedures. ^,^,

Statistical analyses were performed using SAS software Version 9.4 or later. Continuous variables were summarized using the mean, the SD, median, Quartile 1, Quartile 3, minimum value, and maximum value. Categorical variables were summarized using frequency counts and percentages. All statistical tests were 2-sided and performed using a 0.05 significance level, leading to 95% (2-sided) CIs.

The Wilcoxon rank sum test was used to evaluate the primary endpoint, central ECD at week 12 after surgery. It was expected that subjects would have different levels of endothelial healing at week 12. Subjects with ungradable ECD at week 12 were considered to have less healing than subjects with gradable ECD and were therefore ranked lower. Subjects whose ECD was ungradable at week 12 were assigned a value of 0 if no cells were visible on specular microscopy and a value of 1 if the images were ungradable but cells were visible.

A closed 2-step ordered testing procedure was used to control overall type I error for the primary efficacy endpoint, central corneal ECD at week 12. In step 1, the K-321 0.4% QID group was statistically compared with the placebo group. If the Step 1 comparison was significant, Step 2 was performed to statistically cf K-321 0.4% BID group with the placebo group. If the statistical comparison made during Step 1 was not significant, testing was still performed to cf the BID group to placebo; however, all P values were considered nominal/descriptive only.

The differences between the groups in the secondary and other endpoints were evaluated using the Wilcoxon rank sum test for continuous variables, Pearson Chi-square test or Fisher’s exact test for binary variables, or log-rank test for time-to-event analyses. To confirm the treatment effect with adjustment for preoperative values, stratified Wilcoxon tests were performed in a posthoc manner for each stratification factor: ECD (gradable or ungradable), CCT (<550, 550 to <600, or ≥600 µ;m), and BCVA (<70 or ≥70 letters).

RESULTS

study population

Study investigators screened 100 potential subjects, 65 of whom met all inclusion criteria and were randomized on POD1 following completion of a DSO that met study guidelines. Participants were assigned as follows: placebo 22 subjects, K-321 BID 22 subjects, and K-321 QID 21 subjects. Demographic and baseline characteristics of study subjects are detailed in Table 2 .

TABLE 2

Demographics and Baseline Characteristics of Study Subjects.

	K-321 QID N = 21	K-321 BID N = 22	Placebo N = 22	Total N = 65
Mean age, SD (y)	64.7 ± 12.3	65.1 ± 9.7	65.3 ± 9.0	65.0 ± 10.2
Female, n (%)	16 (76.2)	13 (59.1)	20 (90.9)	49 (75.4)
Subjects with family history of FECD, n (%)	4 (19.0)	4 (18.2)	4 (18.2)	12 (18.5)
Mean central corneal ECD, SD (cells/mm ²), n (%)	1318.0 ± 805.8 7 (33.3)	1114.2 ± 588.0 6 (27.3)	930.0 ± 408.2 6 (27.3)	1131.1 ± 621.2 19 (29.2)
Subjects with ungradable/missing ECD	14 (66.7)	16 (72.7)	16 (72.7)	46 (70.8)
Mean corneal thickness, SD (µm)	590.7 ± 48.9	609.8 ± 48.9	622.9 ± 31.2	608.2 ± 45.0
Mean BCVA by ETDRS, SD (letters)	71.5 ± 5.4	65.2 ± 12.2	68.6 ± 6.0	68.4 ± 8.8

BCVA = best corrected visual acuity; BID = twice a day; ECD = endothelial cell density; ETDRS = Early Treatment Diabetic Retinopathy Study visual acuity chart; FECD = Fuchs endothelial corneal dystrophy; QID = four times a day.

The baseline characteristics of the study groups were not statistically different. The average age was 65 years (±10.2). Across the entire cohort, 75% of the subjects were female (range at individual sites, 59%-91%). Fewer than 20% of subjects in each group had a known family history of Fuchs. Most subjects (71%) had unmeasurable central ECD at baseline. Among those who did have measurable baseline central ECD, the average was 1131 cells/mm ² (range 930-1318 cells/mm ²). Average CCT was 608 microns (range 591-623 microns). Baseline vision was 68 ETDRS letters (range 65-72).

Overall, 62/65 (95.4%) subjects completed the week 52 visit. Three subjects withdrew from the study prior to its completion: two subjects in the placebo group (one rescue keratoplasty and one voluntary withdrawal) and one voluntary withdrawal in the K-321 BID group. Study treatment was completed by 63/65 (96.9%) subjects. Two subjects in the placebo group discontinued treatment prior to week 12 due to need for rescue keratoplasty but continued to attend study visits. There were no treatment discontinuations in the K-321 BID or K-321 QID groups.

primary endpoint

Central ECD following DSO

Immediately following descemetorhexis, the central ECD was defined as zero, as the central endothelium was removed during DSO surgery. At week 12, the mean central ECD in the placebo group was 228 ± 298 cells/mm ². The mean central ECD in the QID group was 531 ± 312 cells/mm ² ( P =.0065 compared to placebo) and 468 ± 322 cells/mm ² in the BID group ( P =.0344 compared with placebo) ( Table 3 ). Comparison of the central ECD of the QID group to that of the BID group at 12 weeks was not a prespecified endpoint. Posthoc analysis showed no statistically significant difference in central ECD between the QID and BID groups at the 12-week visit ( P =.7603). The central ECD increased for all treatment groups during the study ( Figure 1 ). At the end of the study (week 52), the QID group had a higher mean central ECD (751 ± 407 cells/mm ²) than the placebo group (432 ± 327 cells/mm ²) ( P =.0047). The week 52 mean central ECD for the BID group was 633 ± 313 cells/mm ², which did not reach statistical significance when compared with the central ECD of the placebo group.

TABLE 3

Central Endothelial Cell Density at Week 12.

	K-321 QID N = 21	K-321 BID N = 22	Placebo N = 22
Mean, SD (cells/mm ²)	530.9 ± 312.5	468.0 ± 322.4	228.1 ± 297.9
Median [Q1, Q3] (cells/mm ²)	561.2 [408, 701.6]	545.7 [323.8, 638.4]	1.0 [0.0, 482.6]
P value ^a for difference to Placebo P value ^a for difference to K-321 BID	.0065 .7603	.0344	–

BID = twice a day; ECD = endothelial cell density; Q1 = the first quartile; Q3 = the third quartile; QID = four times a day.

Stratified Wilcoxon tests using ECD (gradable or ungradable), CCT (<550, 550 to <600, or ≥600 µ;m), and BCVA (<70 or ≥70 letters) yielded similar findings (Supplemental Table 1). Statistically significant differences favoring QID over placebo were observed in all tests, while a statistically significant difference for BID vs placebo was found only in the test adjusted by preoperative ECD. Although not statistically significant, a numerically greater trend was seen in QID compared to BID.

Central ECD adjudication rate across the entire study was 42%. At study screening, 29.9% of images required adjudication, which increased to ≥50% for postoperative visits from 2 to 12 weeks.

secondary endpoints

Time to clearance of corneal edema

Corneal edema was present on clinical exam in all study subjects at each visit, prior to 3 weeks after surgery. At week 3, 8/21 (38.1%) subjects in the K-321 QID group achieved clearance of corneal edema. In contrast, none of the 22 subjects in the K-321 BID group, and 1/22 (4.5%) of the subjects in the placebo group had resolution of corneal edema at week 3 ( Figure 2 ). At week 12, 17/21 (81%) of subjects in the QID group ( P <.0001 compared to placebo) and 12/22 (54.5%) in the BID group ( P =.0028 compared to placebo) had complete resolution of corneal edema, compared with 2/22 (9.1%) of subjects in the placebo group ( Table 4 ). In the following visits, there was a gradual improvement in the number of subjects without corneal edema in each cohort. K-321 QID was significantly better than placebo at most visits prior to week 38, when these groups became equal in this measure. K-321 BID was significantly better than placebo in most visits until week 16 (Supplemental Figure 1).

TABLE 4

Subjects Achieving Complete Clearance of Corneal Edema at Week 12.

	K-321 QID N = 21	K-321 BID N = 22	Placebo N = 22
Complete clearance of corneal edema at week 12
Achieved	17 (81.0%)	12 (54.5%)	2 (9.1%)
Not achieved	4 (19.0%)	10 (45.5%)	20 (90.9%)
Difference to placebo
Difference in percentage point	71.9%	45.5%	–
95% confidence interval	[42.9, 88.9]	[18.4, 68.3]	–
P value ^a	<.0001	.0028	–

BID = twice a day; QID = four times a day.

A total of 18/21 (85.7%) subjects in the K-321 QID group, 16/22 (72.7%) subjects in the K-321 BID group, and 14/22 (63.6%) subjects in the placebo group achieved no corneal edema of the study eye during the trial, with median duration of 35, 51.5, and 267 days after surgery, respectively. The K-321 QID group was statistically better than placebo in this measure ( P =.0021).

Time to recovery of baseline pachymetry

A total of 18/21 (85.7%) subjects in K-321 QID group, 19/22 (86.4%) subjects in K-321 BID group, and 13/22 (59.1%) subjects in the placebo group had return of CCT of the study eye to less than or equal to baseline corneal thickness, with a median time of 50, 50, and 165 days, respectively. A statistically significant difference was found in time to recovery of baseline pachymetry in the K-321 QID ( P =.0259) and K-321 BID group ( P =.0068) compared with the placebo group. The time course of changes in pachymetry after DSO is detailed in Supplemental Figure 2.

other endpoints

Need for and timing of rescue treatment

Six of 22 (27.3%) subjects in the placebo group required rescue keratoplasty at an average of 117 days after surgery (range 45-252 days). Three of 22 (13.6%) subjects in the K-321 BID group required rescue keratoplasty, at an average of 234 days after surgery (range 135-364 days). An additional subject in the K-321 BID group required medical rescue (treatment with topical netarsudil) at 114 days after surgery. In total, 4/22 (18.2%) subjects in the BID group required rescue therapy. Two of 21 (9.5%) subjects in the K-321 QID group had rescue keratoplasty, one at 174 days after surgery and one at 357 days after surgery (average 265 days) ( Figure 3 ).

Only gold members can continue reading. Log In or Register to continue

Stay updated, free articles. Join our Telegram channel

Tags: American Journal of Ophthalmology Volume 290

Sep 19, 2026 | Posted by drzezo in OPHTHALMOLOGY | Comments Off