Introduction
Glaucoma is a progressive optic neuropathy in which irreversible loss of the visual field is produced, and it remains one of the leading causes of blindness worldwide.1 Elevated IOP is regarded as its most important risk factor, and the lowering of IOP continues to be relied upon as the mainstay of treatment.2,3 Most agents currently in use act either by suppression of aqueous humor production or by an increase in uveoscleral outflow.4
In primary open-angle glaucoma, IOP is raised largely because outflow resistance is increased along the conventional trabecular meshwork–Schlemm canal pathway. As aging and oxidative stress proceed, trabecular meshwork cells are rendered dysfunctional; transforming growth factor-β2 is upregulated, and Rho/Rho-kinase (ROCK) signaling is activated, by which the formation of cross-linked actin networks, the accumulation of extracellular matrix, and the stiffening of trabecular tissue are promoted.5,6 Outflow resistance and IOP are raised by these changes, and a cycle of glaucomatous optic nerve damage is in turn sustained. When ROCK is inhibited, the stiffening and the associated cytoskeletal rearrangement are reversed, and aqueous outflow is restored.7
Netarsudil (Rhopressa®), distributed in Korea by Santen Pharmaceutical Co., Ltd. (Osaka, Japan), is a ROCK inhibitor that was first evaluated in clinical trials in the mid-2010s and was approved by the US Food and Drug Administration in December 2017.8 It is a prodrug that is converted by corneal esterases to its active metabolite, netarsudil-M1, and both ROCK1 and ROCK2 are inhibited by it.9 In preclinical work, aqueous outflow was increased and the effective filtration area of the trabecular meshwork was expanded by netarsudil-M1,10 and in a randomized study in humans episcleral venous pressure fell from 7.9 to 7.2 mmHg, a reduction of about 10%,11 an effect also demonstrated experimentally in rabbits.12 Because episcleral venous pressure is not readily reduced by the other classes of agent, a rationale for further IOP reduction is provided by this effect even when the baseline IOP is already low.
The efficacy and safety of netarsudil 0.02% have been established in several international studies,13–15 chiefly in Western populations, and in the Japanese Phase 3 trial (J-ROCKET), in which netarsudil was compared with ripasudil 0.4%, an IOP reduction of about 4.65 mmHg was obtained with netarsudil monotherapy at week 4.16
The relevance of these findings to Korean practice is not straightforward, for three reasons.
First, the difference between East Asian and Western eyes bears on glaucoma chiefly through the pressure at which the disease occurs. In East Asian populations primary open-angle glaucoma arises predominantly at untreated pressures that lie within the statistically normal range,17,18 whereas the trials that established netarsudil enrolled Western patients selected for elevated pressure, with mean baselines above 20 mmHg.13–15 An agent’s effect expressed in millimetres of mercury is not independent of the baseline IOP, so a reduction obtained in a high-tension trial population cannot be assumed to reappear in a population whose baseline IOP is several millimetres lower. In East Asian practice the clinical question is correspondingly often whether any further reduction can be obtained at all, rather than how large that reduction will be.
Second, the disease mix differs even within East Asia. In the Japanese Tajimi Study about 92% of patients with primary open-angle glaucoma had an untreated IOP of 21 mmHg or lower,17 whereas in the Korean Namil Study the corresponding proportion was about 77% (a prevalence of 2.7% at 21 mmHg or lower against a total POAG prevalence of 3.5%).18 The balance of normal-tension to high-tension disease, and with it the pressure from which any further reduction must be obtained, is therefore not the same in the two countries.
Third, the setting in which the drug is used differs. The trials that define what is known were conducted as monotherapy in selected trial populations over periods of 4 weeks to 12 months, whereas in Korea, because netarsudil is reimbursed only for patients with previous ocular hypotensive treatment, it is generally added to or substituted into a regimen that already contains several ocular hypotensive classes. Ripasudil, the comparator in J-ROCKET, is moreover not marketed in Korea, so netarsudil is the first ROCK inhibitor that Korean clinicians and patients encounter.
Most large analyses have been drawn from Western and Japanese populations. Netarsudil was made reimbursable in Korea in October 2024 and was launched in November 2024, so that a full year of use has only recently become possible; to our knowledge, Korean patients have not previously been followed for 12 months across the range of settings met in daily practice. Our 12-month experience was reviewed, and attention was directed both to how efficacy varied by glaucoma subtype and by treatment context, and to how many patients were able to remain on the drug for a year.
Materials and Methods
Study Design and Participants
The study was carried out in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of Kim’s Eye Hospital (approval no. 2025-08-012). Because the design was retrospective, the requirement for informed consent was waived.
Data were obtained by review of the medical records of individual patients.
Consecutive patients in whom netarsudil 0.02% was newly started at our center between November 1, 2024 and February 11, 2025 were eligible. A minimum of 3 months of follow-up was required for entry, so that eyes seen once and not returning were not counted; each eye was then followed for as long as it remained under observation, up to 12 months. The 12-month designation of the study therefore refers to the maximum planned observation period of the cohort and not to a period completed by every eye, and the number of eyes contributing IOP at each visit is given with every result.
An eye was excluded if glaucoma surgery or laser treatment had been performed at any time, if IOP could not be obtained by Goldmann applanation tonometry, if a medication capable of raising IOP was being used, or if angle-closure, neovascular, pseudoexfoliative, pigmentary, uveitic, steroid-induced, or any other secondary open-angle glaucoma was present. The agents considered under this criterion were corticosteroids by any route of administration, and antidepressants and sulfonamide-derived drugs, which have been reported to provoke ciliary body effusion with secondary angle closure. The analysis was therefore confined to primary open-angle glaucoma and to normal-tension glaucoma, which was treated throughout as a subtype of primary open-angle glaucoma and analyzed as a separate group.
Open angles were confirmed by gonioscopy in every patient.
Diagnosis and Classification
Both primary open-angle glaucoma and normal-tension glaucoma were diagnosed by a glaucoma specialist and required an open angle on gonioscopy, a glaucomatous appearance of the optic disc (neuroretinal rim thinning, notching, or a retinal nerve fiber layer defect), and a reproducible visual field defect corresponding to the disc change on standard automated perimetry. The two were separated by the IOP recorded at the initial diagnostic visit, before any ocular hypotensive treatment had been started: an eye whose untreated IOP was 21 mmHg or lower was classified as normal-tension glaucoma, and an eye whose untreated IOP exceeded 21 mmHg as primary open-angle glaucoma. This untreated value was obtained by Goldmann applanation tonometry on a single occasion during office hours.
Because the study was based on records generated in routine care, repeated treatment-free measurements, diurnal tension curves, and 24-hour monitoring were not performed.
Eyes were also grouped by treatment context. In the add-on group netarsudil was added to the existing regimen and no other change was made to it at that visit. In the switch group one agent already in use was withdrawn and netarsudil was started in its place at the same visit; no washout period was interposed, so that baseline IOP in this group was the last value recorded while the agent about to be replaced was still in use.
Under the Korean national health insurance criteria in force throughout the study period, netarsudil was reimbursed only in patients with a documented history of previous ocular hypotensive treatment, a criterion that refers to treatment history rather than to concurrent use; no eye could therefore receive netarsudil as first-line therapy, and no treatment-naive eye was included.
Measurements
All IOP measurements were made by Goldmann applanation tonometry mounted on a slit lamp, after topical anesthesia and instillation of fluorescein. Each patient was examined throughout by the ophthalmologist responsible for their care, so that no patient’s sequence of measurements was interrupted by a change of examiner. Because the examiner was constant within a patient, any systematic difference between examiners is absorbed by the patient-level random intercept and cannot bias the within-patient change in IOP.
The examiners were not masked, since the measurements were made in the course of routine care. Because visits were scheduled for clinical convenience, the time of day at which IOP was measured was not standardized and varied both between patients and between visits of the same patient. The tonometers were checked for calibration error once a month in accordance with the manufacturer’s instructions, and any instrument found outside tolerance was serviced before further use.
At each visit, adverse events were sought by history, slit-lamp examination, and fundus examination. The systemic medications taken by each patient were recorded in the medical record and were reviewed for agents known to produce corneal epithelial deposits.
Discontinuation and Persistence
Persistence on netarsudil was estimated by the Kaplan–Meier method. A documented discontinuation of the drug was taken as the event, and an eye still receiving netarsudil at its last recorded visit was censored at that visit. The add-on and switch groups were compared with the log-rank test.
The medical record of every eye without a 12-month measurement was reviewed specifically to determine whether netarsudil had been discontinued and, if so, the date and the reason. Discontinuation was ascertained in this way for every eye in the cohort, and no eye was censored for loss to follow-up on the basis of a missing measurement alone. Reasons were classified as an ocular adverse event, inadequate pressure response or subsequent surgery, poor adherence, or withdrawal because the pressure was judged to be adequately controlled without the drug.
Responder Analysis
Two responder definitions were applied, each separately. An eye was counted as a 20% responder when its IOP at the timepoint under consideration was at least 20% lower than its own baseline value, and, in a separate analysis, as a 30% responder when the reduction from its own baseline was at least 30%. Each definition was applied first at the 12-month visit and then, in an additional analysis, to the best response recorded at any follow-up visit.
The proportions of eyes reaching an absolute IOP of 18 mmHg or lower and of 15 mmHg or lower at 12 months were determined in the same way.
These two absolute levels were pre-specified because they correspond to the targets most often adopted in practice for the two halves of this cohort. In the Advanced Glaucoma Intervention Study, eyes whose IOP was below 18 mmHg at every visit had a mean change in visual field defect score close to zero over six years, although this was an association observed within the trial rather than a randomized comparison of pressure targets;19 in the Early Manifest Glaucoma Trial the mean IOP achieved in the treated arm was about 15.5 mmHg.3 For normal-tension glaucoma the Collaborative Normal-Tension Glaucoma Study used a 30% reduction from an already normal baseline,20 which in eyes such as ours corresponds to a pressure in the mid-teens. Since mean baseline IOP in the cohort was 17.3 mmHg and 47 of the 131 eyes had normal-tension glaucoma, these two levels spanned the clinically meaningful range for this population.
No formal target was assigned prospectively to individual patients, and the thresholds were applied uniformly for descriptive purposes only.
Statistical Analysis
IOP was analyzed with a linear mixed-effects model in which visit was entered as a categorical fixed effect with baseline as the reference (IOP ~ visit), so that no functional form was imposed on the time course. The random part of the primary model comprised a single intercept for each patient, and the residuals were taken to be independent with constant variance; at the patient level this is equivalent to a compound-symmetric covariance among all observations belonging to that patient, whether they arise from the two eyes at one visit or from one eye at different visits. Adjusted mean changes from baseline were obtained at each visit as contrasts of estimated marginal means with Satterthwaite degrees of freedom.
All available observations were used, so that eyes with an incomplete visit sequence contributed to every visit they attended, under a missing-at-random assumption.
Because a single patient-level intercept constrains the correlation between fellow eyes and the correlation between repeated visits of one eye to be equal, two further models were fitted to test that constraint. In the first, a second random intercept for eye nested within patient was added, by which the two correlations are free to differ. In the second, a first-order autoregressive residual correlation across visits was added within each eye.
Both are reported in Supplementary Table S1.
Whether the change in IOP over time differed by glaucoma subtype or by treatment context was tested with models containing a group-by-time interaction; the interaction was assessed by a likelihood-ratio test, and the between-group difference in the 12-month change was estimated as a difference-in-differences contrast. Because the subtype composition of the add-on and switch groups differed, the treatment-context comparison was repeated with subtype and its interaction with time included as covariates. Relative reductions were analyzed on the same mixed-model framework applied to the percentage change from each eye’s own baseline (percentage change ~ visit + (1 | patient)), with group and its interaction with visit added for between-group contrasts.
Proportions are reported with exact binomial confidence intervals. Analyses were performed in R 4.6.1 (R Foundation for Statistical Computing, Vienna, Austria) with the lme4, lmerTest, emmeans, nlme, and survival packages, and P < 0.05 was taken as significant. No correction was made for multiple comparisons.
Results
Cohort
Eighty-six patients (131 eyes) were analyzed. The mean age was 66.4 ± 12.0 years (range, 32–89 years), and 56 men and 30 women were included. POAG was present in 55 patients (84 eyes) and NTG in 31 patients (47 eyes).
Netarsudil was given as add-on therapy in 106 eyes and as switch therapy in 25 eyes (Table 1). IOP was available in 131 eyes at 1 and 3 months, in 107 eyes at 6 months, and in 91 eyes at 12 months.Table 1Baseline Demographic and Clinical Characteristics of the Study PopulationVariableValuePatients, n86Eyes analyzed, n131Age, years (mean ± SD, range)66.4 ± 12.0 (32–89)Sex (M:F), n56: 30POAG, n patients (eyes)55 (84)NTG, n patients (eyes)31 (47)Add-on group, eyes106Switch group, eyes25Concomitant classes, add-on (median, range)3 (1–5)Baseline IOP, mmHg (mean ± SD)17.34 ± 4.06
In the add-on group, netarsudil was added to the existing regimen and no other change was made to it at that visit. The median number of concomitant classes was 3 (range, 1–5); a beta-blocker was being used in 85 eyes (80.2%), a prostaglandin analog in 84 (79.2%), a carbonic anhydrase inhibitor in 76 (71.7%), and an alpha-2 agonist in 54 (50.9%) (Supplementary Table S2).
Intraocular Pressure
Mean baseline IOP was 17.34 ± 4.06 mmHg. The intraclass correlation coefficient from the mixed model was 0.66, by which it was confirmed that the within-patient correlation was substantial and had to be taken into account. The adjusted mean change from baseline was −3.00 mmHg at 1 month, −3.11 mmHg at 3 months, −3.28 mmHg at 6 months, and −3.05 mmHg (95% CI, −3.80 to −2.30) at 12 months, and each was significant (all P < 0.001) (Table 2 and Figure 1).
Expressed as a percentage of each eye’s own baseline and estimated from the same mixed model, the reduction was 17.2% (95% CI, 14.0 to 20.4) at 1 month, 17.3% (95% CI, 14.1 to 20.5) at 3 months, 17.7% (95% CI, 14.3 to 21.1) at 6 months, and 16.2% (95% CI, 12.7 to 19.7) at 12 months. Figure 1Observed mean intraocular pressure during 12 months of netarsudil treatment in the whole cohort. Error bars denote 95% confidence intervals.
The number of eyes contributing at each visit is shown below the axis (131, 131, 131, 107, and 91). Significance vs baseline (all P < 0.001 at 1, 3, 6, and 12 months) is based on the linear mixed-effects model.A line graph showing intraocular pressure over months after netarsudil initiation.Table 2Adjusted Change in Intraocular Pressure from Baseline (Linear Mixed-Effects Model)GroupBaseline IOPΔ 1 moΔ 3 moΔ 6 moΔ 12 moOverall (n = 131)17.34 ± 4.06−3.00 (−3.67, −2.33)a−3.11 (−3.78, −2.45)a−3.28 (−3.99, −2.56)a−3.05 (−3.80, −2.30)aPOAG (n = 84)18.76 ± 4.07−3.58 (−4.49, −2.68)a−3.65 (−4.56, −2.75)a−4.03 (−4.98, −3.08)a−3.73 (−4.75, −2.72)aNTG (n = 47)14.81 ± 2.54−1.96 (−2.82, −1.10)a−2.15 (−3.01, −1.29)a−1.82 (−2.78, −0.87)a−1.80 (−2.77, −0.82)aAdd-on (n = 106)17.59 ± 4.18−3.15 (−3.89, −2.41)a−3.25 (−3.99, −2.52)a−3.47 (−4.26, −2.69)a−3.40 (−4.23, −2.57)aSwitch (n = 25)16.28 ± 3.37−2.36 (−3.90, −0.82)a−2.52 (−4.06, −0.98)a−2.38 (−4.10, −0.65)b−1.36 (−3.20, +0.48); P = 0.208Notes: Values are adjusted mean changes in mmHg with 95% confidence intervals. aP < 0.001; bP = 0.003. ICC = 0.66.
Eyes contributing IOP: 131 at 1 and 3 months, 107 at 6 months, 91 at 12 months.Abbreviations: CI, confidence interval; ICC, intraclass correlation coefficient; IOP, intraocular pressure.
In the nested model the variance components were 8.96 mmHg2 between patients, 0.40 mmHg2 between fellow eyes within a patient, and 4.62 mmHg2 for the residual, giving a correlation of 0.641 between fellow eyes examined at the same visit and 0.670 between visits of the same eye. Because these two values were almost identical, the compound-symmetric assumption of the primary model was supported, and the nested model did not fit significantly better (likelihood-ratio test, P = 0.100). The adjusted 12-month change was unaltered at −3.05 mmHg (95% CI, −3.78 to −2.31), and the autoregressive model, in which the estimated within-eye autocorrelation was 0.30 (95% CI, 0.11 to 0.47), gave −3.06 mmHg (95% CI, −3.82 to −2.30) (Supplementary Table S1).
Glaucoma Subtype
When the two diagnoses were examined separately, the adjusted change in POAG (baseline 18.76 mmHg) was −3.58 mmHg at 1 month, −4.03 mmHg at 6 months, and −3.73 mmHg (95% CI, −4.75 to −2.72) at 12 months. In NTG (baseline 14.81 mmHg) it was −1.96 mmHg at 1 month, −1.82 mmHg at 6 months, and −1.80 mmHg (95% CI, −2.77 to −0.82) at 12 months; all were significant (all P < 0.001) (Table 2 and 3 and Figure 2). Although mean baseline IOP in the NTG group was under 15 mmHg, a further reduction was maintained through 12 months.
The reduction differed significantly between subtypes (subtype-by-time interaction, P = 0.002); at 12 months it was greater in POAG than in NTG by 1.94 mmHg (95% CI, 0.70 to 3.17; P = 0.002). Figure 2Observed mean intraocular pressure during 12 months of netarsudil treatment, by glaucoma subtype (POAG vs NTG). Error bars denote 95% confidence intervals, and the number of eyes contributing at each visit is shown below the axis.
The reduction differed significantly between subtypes (subtype-by-time interaction, P = 0.002).A line graph showing intraocular pressure over months after netarsudil initiation for POAG and NTG.Table 3Observed Intraocular Pressure Over 12 Months According to Glaucoma SubtypeGroupBaseline1 mo3 mo6 mo12 moPOAG18.76 (17.88–19.65) n = 8415.18 (14.33–16.03) n = 8415.11 (14.34–15.87) n = 8414.73 (13.78–15.68) n = 7315.05 (13.99–16.11) n = 59NTG14.81 (14.06–15.55) n = 4712.85 (11.98–13.72) n = 4712.66 (11.89–13.42) n = 4712.47 (11.35–13.59) n = 3412.50 (11.50–13.50) n = 32Relative reduction, overall (model)—17.2% (14.0–20.4)17.3% (14.1–20.5)17.7% (14.3–21.1)16.2% (12.7–19.7)Relative reduction, POAG (model)—19.3% (15.3–23.2)19.1% (15.1–23.0)20.6% (16.5–24.7)19.0% (14.6–23.3)Relative reduction, NTG (model)—13.5% (8.2–18.7)14.0% (8.8–19.3)12.1% (6.4–17.8)11.2% (5.4–17.1)Notes: IOP values are observed means (95% CI) in mmHg with the number of eyes contributing. Relative reductions are estimated marginal means (95% CI) from a mixed-effects model of the percentage change from each eye’s own baseline, so that the between-group differences quoted in the text are the differences between the values shown; those differences are computed from unrounded estimates and may therefore differ from the rounded values by 0.1 percentage point. All timepoints P < 0.001 vs baseline.Abbreviations: NTG, normal-tension glaucoma; POAG, primary open-angle glaucoma.
Percentage reductions from each eye’s own baseline were analyzed in the same mixed-model framework. At 12 months the estimated reduction was 19.0% (95% CI, 14.6 to 23.3) in POAG and 11.2% (95% CI, 5.4 to 17.1) in NTG, a difference of 7.7 percentage points (95% CI, 0.5 to 15.0; P = 0.037); at 6 months the corresponding estimates were 20.6% and 12.1% (difference 8.5 percentage points; 95% CI, 1.5 to 15.5; P = 0.018). The relative reduction therefore favored POAG in the same direction as the absolute reduction, and the smaller absolute change in NTG is not accounted for by its lower baseline alone.
Treatment Context
By treatment context, the reduction in the add-on group (baseline 17.59 mmHg) was significant at every timepoint and reached −3.40 mmHg (95% CI, −4.23 to −2.57; P < 0.001) at 12 months, an estimated relative reduction of 17.7% (95% CI, 13.9 to 21.5). In the switch group (baseline 16.28 mmHg) the reductions at 1, 3, and 6 months were significant, but the 12-month change was smaller, −1.36 mmHg (95% CI, −3.20 to +0.48), and did not reach significance (P = 0.208); the corresponding estimate was 9.2% (95% CI, 1.5 to 16.9), 8.5 percentage points less than with add-on therapy (95% CI, 0.2 to 16.8; P = 0.045) (Table 2). As with the absolute change, this difference did not persist once subtype was accounted for (7.4 percentage points; 95% CI, −1.0 to 15.9; P = 0.086).
The reason for switching was retrievable from the record in all 25 eyes. Netarsudil replaced an existing agent because IOP remained above target in 18 eyes (72.0%), of which 12 also showed documented structural or functional progression, and because the previous agent was not tolerated in 7 eyes (28.0%). No eye was switched merely to simplify the regimen.
The single class most often replaced was a prostaglandin analog, including latanoprostene bunod (11 eyes, 44.0%), followed by an alpha-2 agonist (7 eyes, 28.0%), a carbonic anhydrase inhibitor (5 eyes, 20.0%), and a beta-blocker (2 eyes, 8.0%) (Supplementary Table S3). After the switch, the median number of concomitant classes retained was 3 (range, 0–4), and no further change was made to the regimen during follow-up. In 5 eyes of 3 patients the agent withdrawn was the only one in use, so netarsudil was continued alone in those eyes; all had nevertheless received previous topical therapy, and none was treatment-naive.
The overall regimen-by-time interaction was not significant (P = 0.156). The two groups were also not comparable in composition: normal-tension glaucoma accounted for 14 of the 25 switched eyes (56.0%) but for only 33 of the 106 eyes treated by addition (31.1%; Fisher P = 0.035), and within each subtype the baseline pressures of the two groups were closely similar. When the model was adjusted for subtype and its interaction with time, the 12-month difference between the two contexts was 1.61 mmHg (95% CI, −0.01 to 3.22) and was not significant (P = 0.051); the corresponding unadjusted difference-in-differences contrast was 2.04 mmHg (95% CI, 0.45 to 3.63; P = 0.012).
The apparent advantage of add-on therapy is therefore substantially attributable to the greater proportion of normal-tension eyes among those switched, and this comparison is presented as exploratory.
Adverse Events
At least one recognizable ocular finding was recorded in 93 eyes (71.0%) of 62 patients (72.1%). The most frequent finding was corneal verticillata, seen in 54 eyes (41.2%) of 37 patients (43.0%), followed by conjunctival hyperemia (43 eyes, 32.8%), increased lacrimation (29 eyes, 22.1%), itching (21 eyes, 16.0%), blurred or reduced vision (17 eyes, 13.0%), periocular swelling (6 eyes, 4.6%), and foreign-body sensation (6 eyes, 4.6%) (Table 4). No systemic adverse event was observed.Table 4Ocular Adverse Events During 12 Months of Netarsudil Treatment (131 Eyes, 86 Patients)Adverse EventEyes, n (%)Patients, n (%)Corneal verticillata54 (41.2)37 (43.0)Conjunctival hyperemia43 (32.8)29 (33.7)Increased lacrimation29 (22.1)19 (22.1)Itching21 (16.0)13 (15.1)Blurred or reduced vision17 (13.0)10 (11.6)Periocular swelling6 (4.6)3 (3.5)Foreign-body sensation6 (4.6)4 (4.7)≥ 1 adverse event93 (71.0)62 (72.1)Discontinuation due to ocular adverse events40 (30.5)—
The visit at which verticillata was first recognized was recorded in all 54 affected eyes. None was detected at the 1-month visit; 30 eyes were first affected at 3 months, 2 at 4 months, and 22 at 6 months, giving a median onset of 3 months (range, 3–6 months). Verticillata was asymptomatic in every case and vision was not affected by it.
Discontinuation and Persistence
Netarsudil was discontinued within 12 months in 44 of 131 eyes (33.6%). The reason was an ocular adverse event in 40 eyes (30.5% of the cohort), poor adherence in 2, an inadequate pressure response leading to glaucoma surgery in 1, and withdrawal because the pressure was judged adequately controlled without the drug in 1 (Table 5). At the patient level, 28 of the 86 patients (32.6%) stopped netarsudil in at least one eye, 25 of them (29.1%) because of an ocular adverse event, and 10 of those 25 (11.6% of the cohort) had stopped within the first 3 months.
Discontinuation occurred at 3 months in 18 eyes, at 4 months in 4, at 5 months in 2, at 6 months in 16, at 9 months in 2, and at 12 months in 2. Forty of the 44 discontinuations (91%) had therefore occurred by 6 months. Four of the 44 eyes attended the 12-month visit despite having stopped the drug — two that stopped at 6 months for poor adherence and two that stopped at the 12-month visit itself — which is why 40 rather than 44 eyes lack a 12-month IOP measurement.Table 5Treatment Response, Discontinuation, and Persistence During 12 Months of Netarsudil TherapyMetricValue (95% CI)≥ 20% reduction at any follow-up visit (n = 131)64.9% (56.1–73.0)≥ 30% reduction at any follow-up visit (n = 131)37.4% (29.1–46.3)≥ 20% reduction at 12 months (n = 91)42.9% (32.5–53.7)≥ 30% reduction at 12 months (n = 91)18.7% (11.3–28.2)Target ≤ 18 mmHg at 12 months (n = 91)87.9% (79.4–93.8)Target ≤ 15 mmHg at 12 months (n = 91)68.1% (57.5–77.5)Discontinuation within 12 months (n = 131)33.6% (25.6–42.4); 44 eyes – Ocular adverse events30.5% (22.8–39.2); 40 eyes – Poor adherence1.5% (0.2–5.4); 2 eyes – Inadequate pressure response leading to surgery0.8% (0.0–4.2); 1 eye – Withdrawn as pressure adequately controlled0.8% (0.0–4.2); 1 eyeKaplan–Meier persistence, 3 months86.3% (80.6–92.4)Kaplan–Meier persistence, 6 months69.5% (62.0–77.8)Kaplan–Meier persistence, 12 months66.4% (58.8–75.0)Notes: Apart from the column headings, bold is used only for the aggregate discontinuation rate; the four rows immediately below it are its mutually exclusive component reasons and sum to that value (44 eyes; 33.6%).
Reasons for discontinuation were taken from the medical record. Corneal verticillata was present in 18 of the 40 eyes discontinued for an ocular adverse event but was recorded as the reason for stopping in only 2. Persistence did not differ significantly between the add-on (69.8%) and switch (52.0%) groups (log-rank P = 0.075) or between POAG (67.9%) and NTG (63.8%) (log-rank P = 0.483).
Kaplan–Meier persistence on netarsudil was 86.3% (95% CI, 80.6 to 92.4) at 3 months, 69.5% (95% CI, 62.0 to 77.8) at 6 months, and 66.4% (95% CI, 58.8 to 75.0) at 12 months (Figure 3). Persistence did not differ significantly between the add-on and switch groups (69.8% vs 52.0%; log-rank P = 0.075) or between POAG and NTG (67.9% vs 63.8%; log-rank P = 0.483). Figure 3Kaplan–Meier estimates of persistence on netarsudil during 12 months of treatment.
Discontinuation was ascertained from the medical record for every eye in the cohort, so that no eye was censored for loss to follow-up on the basis of a missing measurement alone. Shading denotes the 95% confidence interval and the number of eyes at risk is shown below the axis; the number at risk at 12 months (89) is slightly smaller than the 91 eyes with a 12-month IOP measurement, because two eyes discontinued at the 12-month visit itself. Persistence did not differ significantly between the add-on and switch groups (log-rank P = 0.075).A Kaplan–Meier step curve showing persistence on netarsudil declining from 100 percent to 66.4 percent at 12 months.
Attrition
The 91 eyes in which a 12-month measurement was available and the 40 in which it was not did not differ in baseline IOP (17.25 vs 17.55 mmHg; P = 0.677), age (67.4 vs 66.0 years; P = 0.480), diagnosis (P = 0.953), treatment context (P = 0.368), or the percentage reduction already achieved at 3 months (17.5% vs 14.9%; P = 0.338). In a completer analysis restricted to the 91 eyes with a 12-month measurement, the adjusted 12-month change was −3.10 mmHg (95% CI, −3.93 to −2.27), almost identical to the primary estimate.
What separated the two groups was tolerability rather than pressure. Every one of the 40 eyes without a 12-month measurement had stopped netarsudil, and in 38 of them the reason recorded was an ocular adverse event. We do not present the frequency of adverse events in the two groups as an independent comparison, because for eyes that left the study the adverse event and the reason for stopping are the same entry in the record; the informative comparisons are the five above, all of which are null.
Treatment Response
Among the 91 eyes still measured at 12 months, a reduction of at least 20% from baseline was present in 39 eyes (42.9%; 95% CI, 32.5 to 53.7) and of at least 30% in 17 (18.7%; 95% CI, 11.3 to 28.2); a target IOP of 18 mmHg or lower was reached in 80 eyes (87.9%; 95% CI, 79.4 to 93.8) and of 15 mmHg or lower in 62 (68.1%; 95% CI, 57.5 to 77.5) (Figure 4 and Table 5). When any follow-up visit was considered, a reduction of at least 20% was reached at some point in 85 of the 131 eyes (64.9%; 95% CI, 56.1 to 73.0) and of at least 30% in 49 (37.4%; 95% CI, 29.1 to 46.3). Figure 4Responder analysis at 12 months of netarsudil treatment, restricted to the 91 eyes still measured at that visit.
Every eye was already receiving at least one other ocular hypotensive agent when netarsudil was started, so the baseline against which response is measured is an on-treatment pressure; in the switch group it was the value recorded while the agent about to be replaced was still in use. The first two bars show the proportions with an IOP reduction of at least 20% and at least 30% from each eye’s own baseline; the last two show the proportions reaching a target IOP of 18 mmHg or lower and of 15 mmHg or lower. Error bars denote exact binomial 95% confidence intervals, and the numerator and denominator are printed above each bar.Bar graph of 12-month outcomes for intraocular pressure reduction and target pressure.
Discussion
In this single-center retrospective series, Korean patients with primary open-angle glaucoma, including normal-tension glaucoma, were followed for up to 12 months after netarsudil 0.02% was started. Two findings are of principal interest, and they point in opposite directions. The pressure-lowering effect was significant and durable, was obtained in normal-tension as well as in high-tension disease, and was largely unchanged over the year.
Yet one-third of eyes had stopped the drug by 12 months, and in nine of every ten of those the reason was an ocular surface symptom rather than a failure to lower pressure.
Efficacy in Context
Our cohort began from a mean IOP of 17.34 mmHg, several millimetres below the populations in which netarsudil was registered: baseline pressures ranged from 20.4 to 23.4 mmHg across the diurnal timepoints of ROCKET-1 and ROCKET-2,13 from 20.7 to 22.4 mmHg in ROCKET-4,14 and the mean diurnal baseline in J-ROCKET was 20.48 mmHg.16 More than a third of our eyes had normal-tension glaucoma, and every eye had been receiving at least one other agent when netarsudil was started; in all but 5 of the 131 eyes at least one concomitant agent was continued alongside it. Neither the reduction in millimetres of mercury nor the percentage derived from it is therefore directly comparable with those trials. They asked how far an elevated pressure could be brought down; the question here is whether a further reduction can be obtained in eyes that are already treated and already at or near the statistically normal range.
On that basis, a reduction of about 3 mmHg was obtained and sustained for 12 months in a population starting from 17.3 mmHg, in which the therapeutic margin is narrow and conventional agents had already been applied. In the pooled analysis of the ROCKET 1–4 trials, once-daily netarsudil 0.02% was noninferior to twice-daily timolol,15 and in J-ROCKET IOP fell by about 4.65 mmHg (approximately 22%) at week 4 from a baseline of 20.48 mmHg.16 Those figures describe what the drug achieves as monotherapy against an elevated pressure, and they are not the benchmark against which an adjunctive effect in a low-baseline population should be judged.
A comparable pattern was seen in the multicenter real-world study reported by Zaman et al, in which IOP was lowered by 4.3 to 4.5 mmHg (20.5%–20.9%) when netarsudil was added, whereas only a 0.4 mmHg (1.7%) reduction was produced by switching, both measured at week 12.21 In our series a reduction of −3.40 mmHg was likewise maintained through 12 months in the add-on group, whereas the change in the switch group was smaller and did not reach significance at 12 months. That difference warrants caution. The switch group was small (25 eyes), contained a majority of normal-tension eyes, and the difference lost significance once subtype was adjusted for.
A comparable dependence on baseline pressure was reported by Villegas and Lee in 62 eyes already receiving maximally tolerated medical therapy: a larger mean reduction was obtained in eyes starting from 21 mmHg or above (−5.05 mmHg) than in eyes starting from 20 mmHg or below (−2.80 mmHg), whereas a pre-specified target was reached in 69% of the latter and in only 17% of the former.22 Durable adjunctive efficacy over a comparable period has also been reported in refractory glaucoma.23 The contrast indicates that absolute reduction and target attainment do not move together, and in a low-baseline population the second is the more informative measure. In our series a target of 18 mmHg or lower was reached in 87.9% of the 91 eyes with a 12-month measurement, and of 15 mmHg or lower in 68.1%.
The documented reasons for switching bear on this comparison. Switching was undertaken predominantly to obtain further pressure reduction — IOP was above target in 18 of the 25 switched eyes (72.0%), and 12 of those 18 also showed documented progression — and not to reduce treatment burden. That the switched eyes nonetheless began from a lower absolute pressure than the eyes treated by addition reflects the predominance of normal-tension glaucoma among them rather than better pre-existing control, since within each subtype the two groups started from closely similar pressures.
Normal-Tension Glaucoma
The significant reduction maintained in NTG through 12 months (adjusted −1.80 mmHg; P < 0.001) is of clinical interest, because a lowering of IOP has been shown to slow glaucomatous progression even in eyes whose pressure is already within the normal range; in the Collaborative Normal-Tension Glaucoma Study, a reduction of approximately 30% lowered the rate of disease progression.20 By lowering episcleral venous pressure, IOP can be reduced by netarsudil even in the low range that is difficult to reach with other agents.11,12 Normal-tension glaucoma is particularly common in East Asian populations, including Korea.17,18 In such patients, in whom the therapeutic margin is narrow, the treatment options may be widened by this property.
The relative reduction in NTG was nevertheless only 11.2%, against 19.0% in POAG, so the smaller absolute change in NTG is not fully explained by its lower baseline. Because netarsudil was generally started in eyes that had not reached their target pressure, part of the reduction seen in this low-baseline subgroup may also reflect regression to the mean.
Tolerability as the Limiting Factor
The most consequential finding of this study is that netarsudil was discontinued within 12 months in a third of eyes, and that ocular adverse events accounted for 40 of the 44 discontinuations. Persistence at 12 months was 66.4%, and 40 of the 44 discontinuations (91%) had already occurred by 6 months. The closest real-world comparison is a retrospective series from a single glaucoma clinic in San Jose, California, in which 90% of the patients identified as Vietnamese.
Among the 235 eyes of 159 patients analyzed there, discontinuation because of an ocular adverse event occurred in 13.1% by 6 months, with conjunctival hyperemia the leading reason.24 At the same timepoint the corresponding figure here was 28.2% (37 of 131 eyes). The two proportions are not, however, formed on the same denominator. Entry to that series required 3 months of netarsudil therapy, and a further 116 patients who did not reach it were excluded, 49 of them because of side effects, so the eyes lost earliest to intolerance are absent from its denominator.
Entry here required 3 months of follow-up rather than of therapy, so an eye that stopped the drug but remained under observation was retained; in the event no discontinuation in this cohort was recorded before the 3-month visit. Part of the distance between 13.1% and 28.2% is therefore a difference in how the two cohorts were assembled, and the comparison should not be read as a simple excess.
That report is nevertheless one of the few real-world analyses of netarsudil in a largely Asian population and, as in the present study, a mixed-effects model with a subject-level random intercept was used so that the correlation between fellow eyes could be allowed for. Its 6-month reduction in the open-angle group was 4.2 mmHg, obtained after IOP had been adjusted by 3.8 mmHg for each change of medication made after netarsudil was started; without that correction the same figure was 3.8 mmHg, and the mean percentage reduction across its three angle groups was 19.5%. The adjusted reduction at 6 months here, −3.28 mmHg, corresponding to an estimated relative reduction for the cohort as a whole of 17.7% (95% CI, 14.3 to 21.1), is of similar magnitude.
Our rate is, by contrast, close to that of the registration programme itself: discontinuation for an ocular adverse event occurred in 11.6% of our patients (10 of 86) over the first 3 months, against 10% in ROCKET-1 and 12% in ROCKET-2 over the same period,13 and in 29.1% (25 of 86) over 12 months, against 28% (71 of 251) in ROCKET-2.25 The attrition seen here therefore reproduces in routine Korean practice what the pivotal trials found, and is not peculiar to this cohort or to our method of ascertainment.
Several factors are likely to contribute to the difference that remains from the San Jose series. Ascertainment is the most probable: adverse events were sought actively at every scheduled visit by history and slit-lamp examination rather than recorded on spontaneous report, and the outcome of every eye was verified against the medical record. It is also possible that the ocular surface of these patients was already compromised, since the median eye in the add-on group was receiving three other topical classes, most of them preserved, before netarsudil was added.
These explanations do not alter the practical implication. Whether the benefit could be realized depended on continuation rather than on pressure response: one eye was discontinued for an inadequate pressure response, against 40 for an ocular adverse event. The pressure results at 12 months are necessarily drawn from the eyes that remained on treatment and therefore describe a population selected for tolerance of the drug.
Counselling about ocular surface symptoms before starting netarsudil, and review within the first 3 to 6 months when most discontinuations occur, appear warranted.
Corneal Verticillata
Corneal verticillata was the most frequent finding, at 41.2%, a rate higher than the 20.9% reported in the pooled ROCKET analysis.15 In the 4-week J-ROCKET study verticillata was rare (1.6%, 2 of 122).16 The contrast is largely a matter of observation time. In our series the finding was first recognized at a median of 3 months, and in no eye was it detected at the 1-month visit, so at the timepoint at which J-ROCKET reported, the frequency in our own cohort would likewise have been zero. Ascertainment is likely to account for a further part of the difference from the pooled ROCKET figure: because netarsudil-associated verticillata was already known to us as an expected finding, the cornea was examined for it deliberately at the slit lamp at every scheduled visit, whereas an asymptomatic finding is recorded far less completely when adverse events are captured by spontaneous report.
Our figure therefore represents the frequency obtained under active surveillance rather than evidence of greater biological susceptibility in Korean eyes.
Corneal verticillata is not specific to netarsudil. The same whorl-like epithelial deposits are seen with cationic amphiphilic drugs such as amiodarone, chloroquine and hydroxychloroquine, and tamoxifen, and the use of these systemic agents should therefore be considered when verticillata is found in a patient by whom netarsudil is being received. The systemic medications of every patient in this cohort were recorded, and none was taking any of these agents, so a concomitant drug cause of the deposits can be excluded.
The verticillata observed here was asymptomatic in all our patients, vision was not affected by it, and it has been reported to resolve or stabilize once the drug is stopped, over a mean of about 341 days from the last dose in the ROCKET follow-up.25 Although verticillata was present in 18 of the 40 eyes discontinued for an ocular adverse event, it was recorded as the reason for stopping in only 2 of them; conjunctival hyperemia, increased lacrimation and itching were the symptoms that more often ended treatment.
Limitations
This study was retrospective, was carried out at a single center, and lacked a control group, so selection bias cannot be ruled out. The effect of netarsudil alone is hard to isolate, given the range of concomitant medications the patients were taking. Because netarsudil was started in eyes that had failed to reach target IOP on previous therapy, part of the observed reduction may be attributable to regression to the mean.
The number of eyes contributing IOP fell from 131 at baseline to 91 at 12 months. The eyes that were lost did not differ from those retained in baseline IOP, age, diagnosis, treatment context, or the pressure reduction they had already achieved at 3 months, and a completer analysis gave a 12-month change almost identical to the primary estimate, which provides no evidence that the 12-month efficacy estimate was materially inflated by attrition. Attrition was, however, strongly associated with the recording of an ocular adverse event, so the 12-month results describe a population that had tolerated the drug for a year.
The distinction between primary open-angle glaucoma and normal-tension glaucoma rested on a single untreated Goldmann measurement made at the initial diagnostic visit. Telemetric monitoring after nonpenetrating glaucoma surgery has shown that short-term IOP fluctuation is moderate while long-term fluctuation is large, so that irregular measurements are insufficient to describe it.26 The diurnal pattern of an individual eye is, moreover, not reproducible from day to day: in treated primary open-angle glaucoma the between-visit agreement of the change in IOP across the day was poor, with intraclass correlations of 0.08 to 0.38 in right eyes and 0.11 to 0.36 in left eyes, and single-day measurement characterized short-term variation poorly.27 One daytime reading therefore cannot exclude pressures above 21 mmHg occurring outside the moment at which it was taken. Some eyes classified here as normal-tension glaucoma may therefore have had unrecorded elevations, and the subtype assignment should be read as the pragmatic one used in daily practice rather than as a definitive phenotypic classification.
This limitation applies to every analysis stratified by subtype, including the smaller reduction observed in normal-tension glaucoma and the adjustment for subtype in the comparison of add-on with switch therapy.
Because baseline IOP in the switch group was measured on the medication about to be withdrawn, the change reported for that group confounds the loss of the withdrawn agent with the gain from netarsudil and cannot be read as the hypotensive effect of netarsudil itself. The absence of a washout period is unavoidable in routine care but limits the interpretation of the switch analysis accordingly. The switch group was also small and was selected on clinical grounds, so the comparison with add-on therapy is open to confounding by indication and is presented as exploratory.
Because visits were scheduled for clinical convenience, IOP was not measured at a fixed time of day, so part of the variation between visits will reflect diurnal fluctuation rather than treatment effect. Because each patient was examined throughout by the same ophthalmologist, differences between examiners do not affect the within-patient change, but the examiners were not masked and observer bias cannot be excluded. Multiple comparisons were made across timepoints and subgroups without correction, and borderline findings such as those in the switch group should therefore be read with caution.
Adverse events, verticillata in particular, may be subject to detection bias, although in this study that bias operates towards a higher rather than a lower estimate. Finally, structural and functional measures such as visual field mean deviation and central corneal thickness were not collected systematically, and severity-adjusted analyses could therefore not be performed. Prospective multicenter studies in which structural and functional endpoints and a standardized ocular surface assessment are included are needed.
Conclusions
In Korean patients with primary open-angle glaucoma, including those with normal-tension glaucoma, all of whom were already receiving ocular hypotensive treatment, adding netarsudil 0.02% to the existing regimen or substituting it for one of its components produced a further reduction of about 3 mmHg that was sustained over 12 months in the eyes that continued it. This was obtained in a population whose mean baseline IOP was 17.3 mmHg and more than a third of whom had normal-tension glaucoma — a setting in which further pressure lowering is often difficult to achieve with conventional agents, and one in which the millimetres gained are less informative than whether the target was reached. Whether that benefit could be realized depended, however, on tolerability rather than on efficacy: a third of eyes stopped the drug within a year, almost always because of ocular surface symptoms, and most did so within the first 6 months.
These findings are drawn from a retrospective, single-center series without a control group, and a 12-month IOP measurement was available in 91 of the 131 eyes; they should therefore be regarded as supportive rather than confirmatory. They nevertheless suggest that netarsudil is best introduced with explicit counselling about ocular surface symptoms and with early review, and that its real-world value in East Asian practice will depend as much on tolerability as on its pressure-lowering capacity.