Netarsudil as an Adjunctive Therapy: Efficacy and Factors Contributing to a Favorable IOP-Lowering Effect

Purpose The aim of the study is to assess netarsudil's intraocular pressure (IOP)-lowering potential when prescribed as an adjunctive agent, to examine the effect of baseline IOP on patients' response to netarsudil, and to explore patients' characteristics predictive of pronounced responses to netarsudil. Methods This is a single-center, multiprovider retrospective cohort study set at Massachusetts Eye and Ear. Patients with a diagnosis of glaucoma or ocular hypertension on netarsudil and at least one other hypotensive agent for glaucoma who had at least one month of follow-up were included. Patients with additional procedures or glaucoma medication changes were excluded. The main outcome measures were IOP reduction, Kaplan–Meier survival analyses, netarsudil responder type, and complication rates. Results 236 eyes of 236 patients were included. The mean baseline IOP was 19.06 ± 4.6 mmHg on an average of 4 ocular hypotensive medications. 196 (83.1%) patients experienced IOP reduction at the first follow-up visit of 2.84 ± 0.30 mmHg at 55.66 ± 51.89 days. IOP reduction at the second visit among these patients was 3.01 ± 0.44 mmHg at 133.24 ± 77.63 days. After starting netarsudil, 59% had a sustained response (median duration of 315 days), 25% had a robust response (>20% IOP reduction for at least 80% of visits), and 10% had a super response (>20% and >10 mmHg IOP reduction). Netarsudil was effective as an adjunctive therapy across all baseline IOP categories with greater relative IOP reduction in higher baseline IOP groups. Conclusions Netarsudil is an effective adjunctive glaucoma therapy. IOP reductions between 2 and 3 mmHg are typical, but a minority had more pronounced and sustained effects (>10 mmHg). Further analysis is needed to assess specific demographic and clinical factors predictive of these robust responses.

Te efcacy of netarsudil as adjunctive therapy, however, has only been recently explored in the literature. Te MERCURY trials demonstrated that the fxed-dose combination (FDC) of netarsudil (0.01% and 0.02%) and latanoprost (0.005%) lowered IOP more efectively than its individual components [4,5]. Similarly, a few recent studies have demonstrated netarsudil's efcacy as adjunctive therapy after a treatment duration of 1-6 months [6][7][8]. For example, Prager et al. noted a modest IOP reduction of 2.2 mmHg (p < 0.001) in patients on two or more ocular hypotensive medications, including netarsudil with 16.2% of eyes showing a sustained ≥20% IOP reduction during the frst two follow-up visits. However, the magnitude of response to netarsudil addition remains unexplored. If known, these data may help providers anticipate the degree of IOP drop according to the patient's demographic and clinical data.
In this study, we characterized netarsudil's IOP-lowering efect when used as a second, third, fourth, ffth, and sixth agent and characterized the response duration in a survival analysis. We also uniquely identifed patient cohorts with a pronounced and sustained response to netarsudil, whom we designated as "Robust long-term responders" (>20% IOP reduction for at least 80% of visits) and "Super responders" (>20% and >10 mmHg IOP reduction). Such a signifcant IOP reduction has not been extensively reported to our knowledge in any clinical trials or case reports. Finally, we explored the characteristics associated with a favorable IOPlowering response as well as reasons for netarsudil's discontinuation.

Study Design.
Tis is a retrospective cohort study of patients with a diagnosis of glaucoma or ocular hypertension on netarsudil and at least one other hypotensive agent for glaucoma at Massachusetts Eye and Ear (MEE). Approval was obtained from the Mass General Brigham Institutional Review Board, and data collection abided by the Declaration of Helsinki and the Health and Portability and Accountability Act. Medical records of patients who were prescribed netarsudil by a provider at MEE between April, 2018, and August, 2020, were identifed and reviewed. Records were included for analysis if they met the following criteria: (1) the presence of at least 2 IOP recordings prior to netarsudil initiation and (2) at least 1 month of follow-up without additional procedures or glaucoma medication changes. For patients treated with netarsudil in both eyes, the right eye was included in our study. Patients were included in the analysis until they experienced one of the following censoring events: netarsudil discontinuation, a change in the number of adjunctive hypotensive medications, intraocular surgery, or laser procedure. Self-reported medication adherence was recorded.

Outcome Measures.
Te primary outcome measures were intraocular pressure (IOP) reduction (mmHg) from baseline and percent IOP reduction from baseline. Baseline IOP was calculated by averaging the 2 most recent IOP measurements prior to starting netarsudil. IOP measurements were performed with Goldmann applanation tonometry. IOP reduction following netarsudil initiation was assessed at various time intervals which were selected based on a statistical function that evenly distributes ofce visits across the fve chosen time intervals: 1 to 39 days, 40 to 74 days, 75 to 124 days, 125 to 299 days, and 300+ days.

Response Classifcation.
Te patients' response to netarsudil was defned according to the following designations (Figures 1 and 2). "Nonresponders" were patients who did not experience IOP reduction at the frst visit after netarsudil initiation. "Initial responders" were those with IOP reduction at the frst follow-up visit after netarsudil initiation who did not maintain IOP reduction from baseline for at least 80% of subsequent visits. We further divided responders who demonstrated sustained IOP response into three groups, which we termed "Long-term responders," "Robust long-term responders," and "Super responders," to highlight netarsudil's pronounced IOP-lowering efect in the latter groups. "Long-term responders" were patients with IOP reduction from baseline at the frst visit after netarsudil initiation and for at least 80% of subsequent visits. Patients characterized as "Robust long-term responders" were those with >20% IOP reduction from baseline after netarsudil initiation for at least 80% of visits. Lastly, "Super responders" were those with >20% IOP reduction from baseline after netarsudil initiation for at least 80% of visits who also experienced 10 mmHg or greater IOP reduction at any point while on netarsudil. IOP reduction of 10 mmHg was chosen based on its being twice the standard deviation of IOP reduction seen in the original clinical trials [2]. Given that our responder classifcations are time-dependent, the average number of follow-up visits for the total number of subjects and each responder type was as follows: 2.47 ± 2.42 overall, 1.95 ± 2.32 for nonresponders, 2.31 ± 2.44 for initial responders, 2.60 ± 2.39 for long-term responders, 2.59 ± 2.37 for robust long-term responders, and 2.92 ± 2.60 for super responders.

Statistical Analysis.
Primary analyses were unpaired ttests on IOP reduction from baseline at specifc time intervals or ofce visits. Secondary analyses included Kaplan-Meier survival analysis for failure to preserve >20% IOP reduction from baseline for two consecutive visits. Tertiary analyses were logistic regression model outcomes for the odds of a subject being classifed as a given responder type. Logistic regression models were ftted using baseline data to model the log odds of subjects belonging to various responder groups of interest. Stepwise selection in both directions utilizing Akaike's information criterion was performed on each fully saturated model to determine a fnal, "best" model for each outcome. Statistical analysis was performed in R software.

Results
Overall, 236 eyes of 236 patients were included for analysis, with a mean age of 71.6 ± 14.2 years (range: 10-98 years) with 0.85% of patients falling in the 10-18 years age range, 22.88% in the 19-64 years, and 76.27% in the 65-98 years; 53% were female, and 70% were Caucasian (Tables 1 and 2). Netarsudil was added to a glaucoma regimen as a second (8.5%, 20/236), third (17.4%, 41/236), fourth (33.1%, 78/236), ffth (32.6%, 77/236)), and sixth or later (8.5%, 20/236) medication (Table 2). Te medication classes for adjunctive glaucoma agents for the overall population are shown in Table 3. Te proportion of patients falling in each responder group were nonresponders ( (Tables 1 and 2). On average, netarsudil was prescribed as a part of a regimen of 4 agents. Baseline characteristics were similar across all 5 agent groups, with no statistical signifcance between-group diferences in characteristics other than the glaucoma stage and having prior surgery ( Table 2). When comparing responder groups, baseline characteristics were similar as well, except for the glaucoma stage (Table 1). Te baseline diferences in the glaucoma stage are expected from a statistical standpoint given that we are comparing 4 diferent glaucoma stage categories across 5 diferent responder groups with a small sample size in each cross-tabulation. Additionally, super responders were the only responder group with a signifcantly greater baseline IOP of 26.8 ± 7.1 mmHg (p < 0.001), resulting in a signifcantly greater representation in the "Very High" IOP baseline category, 9 subjects (53%, 9/17), compared to the other responder groups (p < 0.001) ( Table 1).     (17) 7 (18) 10 (15) 14 (19) 7 (17) 3 (18) (Table 4). A Tukey test was also conducted to determine if the mean IOP reduction from baseline signifcantly difered between each responder type. Each of the pairwise Tukey comparisons was signifcant (p < 0.0001 for each comparison), and the overall ANOVA was signifcant (p < 0.0001). Figures 3(a) and 3(b) show the stacked bar chart comparing the IOP reduction from baseline in the overall population and across the diferent responder types, respectively.

Kaplan-Meier Survival Analysis.
Kaplan-Meier survival analysis for failure to preserve >20% IOP reduction from baseline for two consecutive visits for the overall population is reported in Figure 4. Te median survival was 315 days (95% CI 264-429). Patients were censored after they had a censoring event (intraocular surgery or laser procedure, change in regimen, netarsudil discontinuation, or reaching the end of the study period).
Notably, patients who met the super responder criteria (17/236 (7.2%)) maintained >10 mmHg reduction for an average of 218.82 ± 285.59 days (7.19 ± 9.39 months). Failure to preserve the super responder status occurred when patients had an IOP reduction from a baseline of less than 10 mmHg at two consecutive ofce visits.
A Cox proportional hazards (PH) model analysis was conducted to investigate the efect of age under the same Kaplan-Meier survival criteria. Te Cox PH coefcient for age is 0.007 (p � 0.39), indicating that age was neither protective nor harmful to survival. We also stratifed age into three groups 10-18 years, 19-64, and 65-98 and recalculated the Cox PH coefcient for age as a noncontinuous variable and got a similar result (Supplementary Table 1).

IOP Reduction by Agent Number.
A Tukey test was conducted to determine whether there was an association between mean IOP reduction and baseline number of glaucoma medications. None of the pairwise Tukey comparisons were found to be signifcant. Te magnitude of IOP lowering, therefore, did not depend on the number of baseline medications (p ≥ 0.771 for all mean comparisons), and the overall ANOVA was not signifcant (p � 0.785). A stacked bar chart was generated to visually compare differences in IOP reduction for the number of glaucoma agents (Figure 3(c)).

IOP Reduction by Baseline IOP.
Likewise, a Tukey test was conducted to determine whether the mean IOP reduction depended on the patient's baseline IOP. When comparing means, the "Low," "Medium," and "High" baseline IOP categories were each signifcantly diferent from the "Very High" baseline IOP category. Te mean diference in reduction is −5.29 mmHg when comparing "Low" to "Very High," −4.42 mmHg for "Medium" to "Very High," and −3.38 mmHg for "High to "Very High" (p < 0.0001, p < 0.0001, and p � 0.008, respectively); this Percentage of subjects out of the total population who took medication that fell in the respective medication class. 2 Number of subjects that took a medication in the relevant medication class.    indicates that those subjects with a "Very High" baseline IOP had a signifcantly greater mean IOP reduction than any of the other baseline IOP groups. Te overall ANOVA was signifcant (p < 0.0001). Figure 3(d) shows the stacked bar chart comparing the IOP reduction from baseline across the diferent baseline IOP categories.

Logistic Regression Analysis.
Secondary analyses for the odds of a patient being classifed as a responder yielded a fnal, parsimonious model post-stepwise selection using AIC. Te fndings from fve diferent multiple logistic regression models, representing each responder type, are seen in Table 5. Te covariates that were statistically signifcant were seen across the robust long-term and super responder models. Of note, some odds ratios are presented as reciprocals of those found in Table 5 to allow for easier clinical interpretation. For the super responder model, it was noted that the odds of being a super responder was 76.92 (95% CI: 12.5-100) times that for subjects with a "Very High" baseline IOP categorization compared to "Medium" IOP and 4.02 (95% CI: 0.068-0.827) times that for subjects with a "Very high" baseline IOP categorization as compared to those with a "High" baseline IOP categorization (p < 0.001 and p � 0.027, respectively). It is important to note, however, the small sample size of patients in this responder group (n � 17) with only 1 patient having a "Medium" baseline IOP which reduces the certainty of the true efect size of this association. Finally, the odds of being a super responder was 3.127 (95% CI 0.984-11.078) for patients with a history of prior surgery compared to those without previous eye surgery with our results trending towards signifcance (p � 0.061).
Te odds of being a robust long-term responder was 9.353 (95% CI: 1.790-172.813) times that for subjects with a "Medium" baseline IOP categorization as compared to those with a "Very High" baseline IOP categorization, maintaining the other parameters as constant (p � 0.034). In addition, the odds of being a robust long-term responder was 2.03 (95% CI: 0.922-4.386) times for patients with a glaucoma type other than POAG compared to those with POAG with our results approaching signifcance (p � 0.074).

Adverse Efects of Netarsudil.
Te adverse efects experienced across all patients and by responder type are summarized in Table 6. Te only adverse efect that demonstrated statistical signifcance across responder types was tearing (p � 0.003), which was the highest in long-term responders.

Discussion
In our retrospective cohort study, we reported netarsudil's IOP-lowering efect as an adjunctive agent. Our study found an IOP-lowering efect when prescribed as adjunctive therapy with an average reduction from baseline of 2.84 ± 0.30 mmHg (14.34% ± 22.19%) at the frst follow-up visit and 3.01 ± 0.44 mmHg (14.94% ± 28.04%) at the second follow-up visit. Our results are consistent with those of previous studies exploring netarsudil's efcacy as a monotherapy and adjuvant therapy. Te ROCKET-1 monotherapy clinical trial reported IOP lowering of 3.3-5.0 mmHg (17-22%) at 3 months [2]. When assessing netarsudil's efect in patients on maximally tolerated medical therapy, Villegas et al. noted a mean IOP lowering of  Stepwise-selected "best" models created from the full model, which include all of the possible baseline covariates (seen in Table 1). Te automated procedure chooses the best model, which is the best combination of the aforementioned covariates. 2 CI cannot be estimated here since 0 super responders fall into the "low" baseline IOP category. 3 IOP baseline categories: low (IOP < 15 mmHg), medium (15 mmHg ≤ IOP < 20 mmHg), high (20 mmHg ≤ IOP < 25 mmHg), and very high (IOP ≥ 25 mmHg). −3.53 mmHg (−17%) at the frst follow-up visit which occurred at a mean of 4.8 ± 3.5 weeks. Another study analyzing netarsudil's adjunctive efect found a similar IOP reduction of 3.9 ± 4.6 mmHg (17.5 ± 6.0%) in a relatively small sample of 95 eyes [6]. Tat study, however, was limited by very short follow-up duration (mean duration on netarsudil: 53.7 days) and high variability in follow-up frequency for IOP measurements (data only reported up to three follow-up visits).
Our study included a longer follow-up period compared to the aforementioned studies and thereby ofered important insights when assessing the durability of netarsudil's efect. We found that patients experienced a sustained hypotensive efect lasting well beyond one year, with survival analysis demonstrating a 50% probability of preserving >20% IOP reduction at 400 days. Moreover, 59% (144/244) of all patients in the study experienced a sustained response (IOP reduction from baseline at >80% of ofce visits) to netarsudil. Explanations for the durability of netarsudil's IOPlowering efect include potential trabecular meshwork remodeling [4,5,[10][11][12] and its unique role in blocking profbrotic TGF-beta efects in glaucoma fltration surgery [13][14][15][16][17][18].
In addition, our study uniquely stratifed patients by the change in IOP into diferent responder types to demonstrate the range of drug efects. We identifed a cohort of patients (17/236, 7.2%), referred to in this study as "super responders," who demonstrated ≥10 mmHg IOP reduction from baseline as well as a sustained ≥20% IOP reduction at >80% of follow-up visits. Such a pronounced IOP reduction has not been previously reported to our knowledge in clinical trials or case reports. Mounting anecdotal accounts from providers with standout cases of pronounced netarsudil hypotensive efect warrant further exploration of baseline characteristics predictive of such a response. Our logistic regression analysis identifed that among the characteristics contributing to the super responders' 12.1 mmHg (45.7%) IOP reduction from baseline was having a "Very High" baseline IOP (IOP ≥ 25) and a history of prior eye surgery (p � 0.061) ( Table 5). Te logistic regression analysis further revealed that having a "Medium" baseline IOP and a glaucoma type other than POAG are potential contributors to netarsudil's efcacy in patients who were referred to in this study as "robust long-term responders," who experienced >20% reduction in IOP in >80% visits. Studies with larger sample sizes should further investigate the efect of the aforementioned variables on netarsudil's efcacy as well as explore mechanisms that potentially mediate these associations.
We also noted that the degree of IOP reduction was not afected by the baseline number of hypotensive agents (Figure 3(c)). Tis is consistent with a study by Prager et al., which reported no signifcant diferences in IOP reduction from baseline between patients on 1, 2, 3, or 4 glaucoma medications. Tis fnding provides additional support that netarsudil's adjunctive efcacy may stem from its unique mechanisms of action (decreased trabecular meshwork resistance, decreased aqueous humor formation, and decreased episcleral venous pressure) which complement that of other agents.
IOP inclusion criteria for this study were broad (mean baseline IOP: 19.06 ± 4.6). Netarsudil's adjunctive efcacy was apparent across all baseline IOP categories with greater relative IOP reduction in the "High" and "Very High" baseline IOP groups (38.3% and 53.84%) relative to the low and medium groups (24.53% and 28.3%). Tere are few prior studies that have explored the role of baseline IOP and its efect on netarsudil efcacy. In the FDA clinical trials for netarsudil and netarsudil/latanoprost FDC, Brubaker et al. found a greater proportion of patients with ≥20% IOP reduction in patients with baseline IOP ≥ 25 mmHg when compared to those with baseline IOP < 25 mmHg [19]. Similarly, previous studies report that the absolute hypotensive efects of timolol and latanoprost were related to baseline IOP [20,21]. Some of these studies, however, noted that netarsudil's IOP-lowering efcacy was unchanged across various baseline pressures [20]. Our fndings suggest that as an adjunctive treatment modality, netarsudil's IOP-lowering efect is greater in patients with higher baseline pressures (>25 mmHg).
Similar to prior studies, we found that netarsudil was well tolerated in most patients [2,22]. In terms of adverse efects, the most common complication studies have found is conjunctival hyperemia with incidence ranging between 30.4% and 60.6% [2,7]. Our study is consistent with these results as 38.1% of patients were reported to have conjunctival hyperemia (Table 6).
Our study has several limitations, including the absence of a control group. In addition, given that this study was not a clinical trial and instead was aimed at investigating netarsudil's efcacy in a real-world glaucoma practice, there was no wash-out period; therefore, the hypotensive efect cannot be entirely attributed to netarsudil. Furthermore, while the study did control for changes in the glaucoma regimen with its stringent inclusion criteria, it did not account for changes in the glaucoma regimen coinciding with netarsudil's initiation, such as the concurrent discontinuation of other glaucoma medications. Finally, there was no practical way to distinguish normal diurnal fuctuations of IOP from the IOP measured in the clinic as this study focused on the real-world use of netarsudil as an adjunctive medication and was thus subjected to the constraints of regular daily clinic hours which start at 7:00 am and end at 7:00 pm EST.

Conclusion
Our results support the use of netarsudil as an adjunctive treatment that can provide signifcant and sustained IOP lowering with once-daily dosing for patients across a broad baseline IOP range and number of baseline medications. Future studies should aim to create more explicit terminology that considers criteria related to netarsudil response over a longer follow-up period as well as further assesses specifc demographic and clinical factors predictive of these robust responses.

Disclosure
Te research was conducted as part of the employment of the authors.

Data Availability
Te data used to support the fndings of this study are available upon request; please contact sandy.samuel96@ gmail.com for access.