Predictive value of uninterrupted dry macula duration for discontinuation of anti-VEGF therapy in diabetic macular edema

Abstract

Successful discontinuation of intravitreal anti-vascular endothelial growth factor (VEGF) injections for diabetic macular edema (DME) without recurrence is rarely achieved. This investigator-initiated retrospective study examined whether maintaining uninterrupted dry macula, without even brief recurrences, using anti-VEGF injections is correlated with subsequent treatment discontinuation. Data were collected from 15 Japanese DME patients who completed two years of brolucizumab or aflibercept treatment in the KESTREL phase III trials and were followed for ≥ 3 years. Treatment discontinuation was defined as no anti-VEGF injections for ≥ 3 years without DME recurrence after KESTREL completion. Dry macula was defined using central macular thickness (CMT) thresholds between 270 and 320 μm. A moderate correlation between the duration of uninterrupted dry macula and treatment discontinuation was observed at a CMT threshold of ≤ 290 μm (r = 0.567, p = 0.028). No such correlation was observed for the cumulative dry duration that included brief recurrences. Firth’s logistic regression indicated that maintaining a dry macula at ≤ 280 μm for 7.34 months (95% CI: 3.62–11.07) was associated with a 50% probability of treatment discontinuation. Maintaining long-term uninterrupted dry macula with anti-VEGF agents may be associated with treatment discontinuation in DME. Confirmation in larger prospective studies would be warranted.

Introduction

Intravitreal anti-vascular endothelial growth factor (VEGF) injections are the first-line therapy for diabetic macular edema (DME). To prevent DME recurrences that leads to subsequent vision loss, long-term repeated injections are required. However, the burden of frequent hospital visits and associated costs may lead to poor treatment adherence, particularly in patients with multiple diabetes-related comorbidities^1,2. In real-world clinical settings, DME patients who require care from multiple other medical specialists tend to miss more ophthalmology appointments for anti-VEGF injections than those with neovascular age-related macular degeneration (nAMD), another condition that also requires long-term anti-VEGF therapy³. Consequently, effective strategies for discontinuing anti-VEGF therapy without causing recurrences remain an unmet need for both patients with DME and their physicians.

Successful discontinuation of anti-VEGF therapy without DME recurrence has rarely been reported, although some patients can eventually discontinue treatment, typically after long-term therapy^4–6. In the Endurance study, 60 patients from the VISTA trial were followed for 2 additional years. The mean number of aflibercept injections was 4.5 in the first additional year, with 30% of patients requiring no injections. By the end of the second year, the mean number of injections increased to 7.7, including 25% of patients who did not require any injections throughout the entire two-year period⁷. In a retrospective analysis conducted over two years at three National Health Service (NHS) clinics in the United Kingdom involving 1,606 eyes with DME, the proportion of cases with a treatment-free interval of more than 180 days was 27.8%⁸. Similarly, a retrospective study from the United States involving 1,702 DME patients reported that 30% discontinued anti-VEGF therapy between months 36 and 60 following the initial 24-month treatment period⁹.

Recently introduced new-generation anti-VEGF agents such as brolucizumab, faricimab, and aflibercept high-dose (HD) offer extended dosing intervals and superior drying efficacy for DME compared to conventional agents including ranibizumab and aflibercept^10–14. Although the number of cases was small, a recent case series demonstrated a high rate of successful treatment discontinuation with one of these new-generation anti-VEGF agents, brolucizumab. In this case series, 3 out of 4 patients with DME were able to discontinue injections and maintain a dry macula for over 3 years after completing 2 years of intensive brolucizumab therapy in the phase III KITE study. In contrast, the only patient treated with conventional anti-VEGF agent, aflibercept, who received 5 monthly intravitreal injections followed by fixed bimonthly dosing for 2 years until the end of the phase III trial, experienced a recurrence of DME only 5 months after the discontinuation of aflibercept¹⁵.

The KESTREL(CRTH258B2301)/KITE(CRTH258B2302)phase III studies demonstrated that brolucizumab has a superior ability to achieve and sustain a dry macula compared with the conventional anti-VEGF agent, aflibercept^16–18. The authors suggested that the greater drying potential of new-generation anti-VEGF agents may enable treatment discontinuation in patients with DME, underscoring the need for further investigations¹⁵.

To verify this hypothesis, we collected clinical data from Japanese patients with DME who participated in the KESTREL trial^16,17 at authors’ institutions, along with subsequent follow-up data for longer than 3 years. Using these data, we investigated the correlation between the duration of uninterrupted dry macula during the two-year KESTREL phase III trial and subsequent successful discontinuation of anti-VEGF injections for more than 3 years without DME recurrence.

Results

Successful discontinuation of anti-VEGF treatment following two years of intensive intravitreal injections during the KESTREL study

Of the 376 participants in the KESTREL study, 19 Japanese patients were treated at the authors’ institutions, where informed consent for the secondary use of clinical data was obtained at the time of enrollment.

Clinical data were obtained through an opt-out consent process. Of the 19 patients, 15 were followed up for at least 3 years after the conclusion of the KESTREL study. Among these 15 patients, 5 had been treated with aflibercept, receiving 5 monthly loading injections followed by fixed bimonthly injections for 2 years. The remaining 10 patients had been treated with brolucizumab, receiving 5 loading doses every 6 weeks, followed by injections at 2-month or 3-month intervals, depending on their disease activity.

In the brolucizumab group, 7 of 10 patients exhibited no recurrence of DME and required no additional anti-VEGF injections for three years or longer following the completion of the KESTREL study. These cases were therefore classified as successful treatment discontinuations. In contrast, only 1 of 5 patients in the aflibercept group demonstrated a similarly sustained absence of recurrence and did not require further treatment during the same three-year follow-up period.

Among the 7 brolucizumab discontinuation cases, the total number of injections over two years was 13 in three patients, 12 in two, and 11 in two. The single discontinuation case in the aflibercept group received 15 injections over 2 years.

Comparison of central macular thickness between treatment-discontinued and treatment-continued patients, Fig. 1.

Fig. 1.

Comparison of central macular thickness between treatment-discontinued and treatment-continued patients. Throughout the 2-year KESTREL study, which included 28 visits, the treatment discontinuation group consistently showed thinner central macular thickness (CMT) than the treatment continue, generally remaining below the 320 μm edema threshold. However, the difference did not reach statistical significance, possibly due to the limited sample size.

The baseline CMT was 566.9 ± 39.9 μm (average ± SE) in the treatment-continued group and 444.8 ± 28.3 μm in the treatment-discontinued group. This difference persisted after a single injection (356.7 ± 23.9 μm vs. 316.8 ± 24.9 μm) and throughout the 2-year KESTREL study, with the treatment-discontinued group consistently exhibiting thinner CMT, generally remaining below the 320 μm edema threshold. However, the difference did not reach statistical significance, possibly due to the limited sample size.

Correlation between CMT changes and discontinuation of anti-VEGF injections

Early factors, including baseline CMT (p = 0.057), the reduction in CMT after the initial injection of brolucizumab (p = 0.371), and the CMT at the subsequent visit following the initial injection (p = 0.164), were not significantly correlated with discontinuation of anti-VEGF therapy after completion of 2 years of brolucizumab treatment in the KESTREL trial.

On the other hand, the uninterrupted duration of dry macula, defined as CMT ≤ 280 μm, was correlated with subsequent discontinuation of anti-VEGF injections for at least 3 years after completion of KESTREL study (Correlation Coefficient = 0.647, p = 0.009). The estimated duration of uninterrupted dry macula that allow discontinuation of anti-VEGF injections for 50% of participants was 7.34 months (3.62–11.07). We also assessed the total duration of dry macula, allowing periodic recurrence of macular edema, but found no correlation with discontinuation of anti-VEGF injections (Correlation Coefficient = 0.447, p = 0.095).

Correlation between the duration of uninterrupted dry macula during 2 years of intensive intravitreal anti-VEGF therapy and subsequent treatment discontinuation

Macular edema thresholds were defined in 10 μm decrements from 320 to 270 μm based on SD-OCT–measured CMT, classifying eyes as having a dry macula if at or below each threshold, and as having macular edema if it was above. As shown in Table 1, a moderate correlation was observed between the duration of uninterrupted dry macula and successful anti-VEGF discontinuation at thresholds of 290 μm or thinner.

Table 1.

Correlation between the duration of uninterrupted dry macula during 2 years of intensive intravitreal Brolucizumab or Aflibercept therapy and subsequent treatment discontinuation.

Dry Macula Definition (CMT threshold)	Correlation coefficient (r)	p-value	correlation
≤ 320 μm	0.324	0.173	Not correlated
≤ 310 μm	0.331	0.145	Not correlated
≤ 300 μm	0.508	0.053	Not correlated
≤ 290 μm	0.567	0.028	Correlated
≤ 280 μm	0.647	0.009	Correlated
≤ 270 μm	0.625	0.017	Correlated

Macular edema thresholds ranging from 320 to 270 μm, in 10 μm increments, were defined based on SD-OCT–measured central macular thickness (CMT). Eyes with CMT values below each threshold were classified as having a dry macula. A moderate correlation between the uninterrupted duration of dry macula and successful discontinuation of anti-VEGF therapy was observed at thresholds of 290 μm or less.

The estimated durations of uninterrupted dry macula at which 50% of DME patients could successfully discontinue anti-VEGF injections were determined using Firth’s logistic regression analysis and are summarized in Table 3.

Table 3.

Analysis of the required duration of uninterrupted dry macula for successful anti-VEGF discontinuation.

Dry Macula definition (CMT threshold)	Dry macula duration and injection discontinuation	Dry Macula Duration at Which 50% of Cases may discontinued Injections (95% confidence interval)
≤ 320 μm	Not correlated	NA
≤ 310 μm	Not correlated	NA
≤ 300 μm	Not correlated	NA
≤ 290 μm	Correlated	9.62 months (3.69–15.55)
≤ 280 μm	Correlated	7.34 months (3.62–11.07)
≤ 270 μm	Correlated	5.61 months (1.66–9.57)

Estimated durations of uninterrupted dry macula required for 50% of DME patients to successfully discontinue anti-VEGF therapy were calculated using Firth’s logistic regression. Only the presence or absence of a correlation is shown in Table 3; correlation coefficients and p-values are provided in Table 1.

At the 300 μm threshold, the correlation coefficient was 0.51 (p = 0.053), narrowly missing statistical significance.

Correlation between the total duration of dry macula, interrupted by brief DME recurrences, and subsequent treatment discontinuation

In contrast, the total duration of all dry macula periods, intermittently interrupted by brief recurrences of DME, showed no correlation with subsequent discontinuation of anti-VEGF injections. This finding suggests that even brief episodes of macular edema recurrence may diminish the correlation (Table 2). These results underscore the importance of maintaining a continuously dry macula—without even brief recurrences of DME—through intravitreal anti-VEGF therapy to facilitate successful treatment discontinuation.

Table 2.

Correlation between the total duration of dry macula, interrupted by brief diabetic macular edema (DME) recurrences, and subsequent discontinuation of anti-vascular endothelial growth factor (VEGF) treatment.

Dry Macula Definition (CMT threshold)	Correlation Coefficient	p-value	correlation
Definition of dry macula with CMT
≤ 320 μm	0.225	0.273	not correlated
≤ 310 μm	0.213	0.193	not correlated
≤ 300 μm	0.343	0.211	not correlated
≤ 290 μm	0.382	0.159	not correlated
≤ 280 μm	0.447	0.095	not correlated
≤ 270 μm	0.401	0.155	not correlated

The total duration of all dry macula periods, when intermittently interrupted by brief DME recurrences, showed no correlation with subsequent anti-VEGF injection discontinuation, suggesting that even brief recurrences may disrupt this correlation.

Analysis of the required duration of uninterrupted dry macula for successful discontinuation of anti-VEGF therapy

The estimated durations of uninterrupted dry macula at which 50% of DME patients could successfully discontinue anti-VEGF injections were determined using Firth’s logistic regression analysis. These results are presented in the column titled ‘Dry Macula Duration at Which 50% of Cases may discontinued Injections’ in Table 3. Only the presence or absence of a correlation between uninterrupted dry macula duration and injection discontinuation is shown here; the correlation coefficient and p-value are provided in Table 1.

Representative CMT changes in patients who discontinue or continue anti-VEGF therapy

The threshold for DME was set at 280 μm and red squares indicate a dry macula below this threshold. Blue arrows indicated the timing of anti-VEGF injections during the induction phase, and black arrows injections during maintenance phase.

(Case 1) CMT changes during the phase III KESTREL study in a patient who successfully discontinued aflibercept after study completion, Fig. 2.

Fig. 2.

CMT changes during the phase III KESTREL study in a patient who successfully discontinued aflibercept injections after study completion. At baseline, the CMT was 343 μm and remained above 280 μm even after five monthly injections. It first dropped below 280 μm at month 13 and largely remained below this level throughout the second year of the KESTREL trial. The total duration of dry macula was 11 months (indicated by the red squares) with an uninterrupted duration of 10 months, exceeding the 7.34-month threshold for treatment discontinuation shown in Table 3. Representative OCT images at each time point are also presented.

At baseline, the CMT was 343 μm and remained above 280 μm even after the 5 monthly aflibercept injections during the induction phase. It first dropped below the threshold of 280 μm at month 13 and remained below this level throughout the second year of the KESTREL trial. The total duration of dry macula was 11 months (indicated by red squares), and the uninterrupted dry macula duration was 10 months, which exceeded the 7.34-month threshold at which 50% of cases may discontinue injections, as described in Table 3.

(Case2) CMT changes during the KESTREL study in a patient who was unable to discontinue aflibercept injections following study completion, Fig. 3.

Fig. 3.

CMT changes during the KESTREL study in a patient who was unable to discontinue aflibercept injections following study completion. At baseline, the CMT was 710 μm and decreased to below 280 μm after two monthly injections, remaining low until month 7. A saw-tooth pattern was observed, with CMT exceeding 280 μm every two months until month 21. The total dry macula duration was 16 months (indicated by red squares), but the uninterrupted dry period lasted only 6 months—shorter than the 7.34-month threshold for injection discontinuation described in Table 3. To visualize the changes, representative OCT images from each time point are presented.

At baseline, the CMT was 710 μm and decreased to below 280 μm just after 2 monthly aflibercept injections during the induction-phase, and remained this threshold until month 7. Thereafter, the CMT took a characteristic saw-tooth pattern observed in aflibercept treated eyes²⁰. The CMT exceeded the threshold every 2 months until month 21. As a result, although the total duration of dry macula was 16 months (indicated by the red squares), the longest uninterrupted dry macula duration was limited to 6months, which was shorter than the 7.4 month threshold at which 50% of cases may discontinue injections, as described in Table 3.

(Case 3) CMT changes during the phase III KESTREL study in a patient who successfully discontinued brolucizumab injections after study completion, Fig. 4.

Fig. 4.

CMT changes during the phase III KESTREL study in a patient who successfully discontinued brolucizumab injections after study completion. The baseline CMT was 494 μm and dropped below 280 μm after two monthly injections, remaining low until month 24 except at month 19 (295 μm). The total dry macula duration was 22 months (indicated by the red squares), with 9 months uninterrupted—exceeding the 7.34-month threshold for discontinuation shown in Table 3. To visualize the changes, representative OCT images from each time point are presented.

At baseline, the CMT was 494 μm and decreased to below the threshold of 280 μm just after 2 monthly brolucizumab injections, and remained below the threshold till month 24, except for month 19 when the CMT was 295 μm. The total duration of dry macula was 22 months (indicated by the red squares), and the uninterrupted dry macula duration was 9 months, which exceeded the 7.34 month threshold at which 50% of cases may discontinue injections, as described in the Table 3.

(Case 4) CMT changes during the KESTREL study in a patient who was unable to discontinue brolucizumab injections following study completion, Fig. 5.

Fig. 5.

CMT changes during the KESTREL study in a patient who was unable to discontinue brolucizumab injections following study completion. The baseline CMT was 639 μm and dropped below 280 μm after three monthly injections, remaining low during the induction phase, which consisted of five injections at 6-week intervals. During the bimonthly maintenance phase, the CMT fluctuated noticeably, repeatedly rising above and falling below the 280 μm threshold. The total dry macula duration was 16 months (indicated by the red squares), but the uninterrupted dry period lasted only 7 months—below the 7.4-month threshold shown in Table 3. To visualize the changes, representative OCT images from each time point are presented.

At baseline, the CMT was 639 μm and decreased to below the threshold of 280 μm after 3 monthly brolucizumab injections. It remained below this threshold during induction phase which consisted of 5 brolucizumab injections at 6-week intervals. CMT fluctuated in the maintenance phase while brolucizumab was injected bimonthly. The total duration of dry macula was 16 months (indicated by the red squares), and the uninterrupted dry macula duration was 7 months, which was shorter than the 7.4 month threshold at which 50% of cases may discontinued injections, as described in Table 3.

Discussion

VEGF was first identified in studies of tumor angiogenesis, a process that supplies oxygen and nutrients to rapidly growing malignant tumors. Because tumor growth often outpaces the development of new blood vessels, plasma leakage from both neovascularization and surrounding existing vasculature serves as a critical source of oxygen and nutrients that supports rapid tumor expansion^21–23.

Diabetes mellitus is a systemic vascular occlusive disease that causes complications such as stroke and myocardial infarction. Capillary occlusion in the diabetic retina leads to ischemia of retinal neurons, triggering hypoxia-inducible factor (HIF) and VEGF expression, resulting in increased plasma leakage from the remaining patent vessels. While this leaked plasma may support the survival of ischemic retinal cells by supplying oxygen and nutrients, the leaked plasma also flows into the central macula, resulting in DME and subsequent vision loss^24,25. By inhibiting VEGF activity, anti-VEGF agents suppress vascular leakage, which in turn resolves macular edema and improves visual acuity. However, as they are not curative but symptomatic treatments, long-term, repeated intravitreal injections are often required.

In this investigator-initiated clinical study, we aimed to test the hypothesis that maintaining an uninterrupted dry macula without even brief recurrences of DME through intravitreal anti-VEGF injections may enable successful discontinuation of anti-VEGF therapy. This hypothesis was based on a preliminary study, a small case series of only 4 patients, which reported that 3 of 4 experienced no recurrence of DME for over three years after completing the two-year course of brolucizumab treatment in the KITE phase III trial¹⁵. These results suggested that brolucizumab may potentially allow for a higher discontinuation rate of anti-VEGF agents than the 20–30% discontinuation rates reported in previous studies using conventional anti-VEGF agents^7–9,26. We conducted an investigator-initiated study at 4 Japanese institutions affiliated with the authors, all of which had participated in the KESTREL trials. Of the 19 patients enrolled in the phase III trials at these sites, follow-up data beyond 3 years after trial completion were available for 15 patients. Among the 10 patients who had received brolucizumab for two years in the KESTREL trial, 7 successfully discontinued anti-VEGF therapy without DME recurrence for more than 3 years. Although limited by small sample size, both the present study (7 of 10) and the earlier case series (3 of 4)¹⁵ suggest that approximately 70% of patients experienced no DME recurrence for over 3 years following completing the phase III KESTREL/KITE trial with brolucizumab.

To identify factors associated with the successful discontinuation of anti-VEGF injections in DME patients without recurrence, we investigated whether the duration of an uninterrupted dry macula achieved during the 2-year KESTREL trial was correlated with successful discontinuation of anti-VEGF therapy after the conclusion of the trial. When dry macula was defined as a central macular thickness (CMT) of 290 μm or less, a correlation was observed between the duration of uninterrupted dry macula and successful discontinuation of anti-VEGF therapy for more than 3 years following the completion of the 2-year phase III trial.

To evaluate whether brief recurrences of DME affect the likelihood of treatment discontinuation, we analyzed the association between the cumulative duration of dry macula—defined as the total time during which the CMT remained below the threshold, even if interrupted by brief recurrences of DME—and the eventual discontinuation of anti-VEGF therapy. This total dry macula duration showed no correlation even under the most stringent criterion of CMT ≤ 270 μm. As noted above, different from the duration of uninterrupted dry macula, the total duration of dry macula permits inclusion of brief intermittent recurrences of DME. These results suggest the possibility that, in the context of considering discontinuation of anti-VEGF therapy, treatment strategies that help prevent even occasional recurrence of macular edema could be more desirable. Extending treatment intervals while maintaining CMT at or below 290 μm with anti-VEGF injections—even in the absence of clinically evident visual acuity decline—may represent a more effective strategy for achieving treatment discontinuation.

Finally, we investigated the required duration of uninterrupted dry macula for successful discontinuation of anti-VEGF therapy using Firth’s logistic regression analysis, which has been shown to produce reliable results even with small sample sizes¹⁹. Based on CMT thresholds, uninterrupted dry macula durations without recurrence of macular edema required to achieve a 50% probability of anti-VEGF discontinuation were as follows: 9.62 months (95% CI: 3.69–15.55) for CMT ≤ 290 μm, 7.34 months (95% CI: 3.62–11.07) for CMT ≤ 280 μm, and 5.61 months (95% CI: 1.66–9.57) for CMT ≤ 270 μm. Although definitive conclusions cannot be drawn due to the small sample size, our findings suggest that if CMT can be maintained at ≤ 280 μm for approximately 7 consecutive months, around 50% of patients may be able to discontinue anti-VEGF therapy. In the KESTREL phase III study, the proportion of patients with a CMT of less than 280 μm was 63.5% in the brolucizumab group and 51.9% in the aflibercept group at Week 100, suggesting greater efficacy of brolucizumab in achieving a dry macula which may enable subsequent discontinuation of anti-VEGF injections¹⁶.

Based on these data, changes in CMT of four representative cases were illustrated using a DME threshold of 280 μm (Figs. 2, 3, 4 and 5). Case 1 that achieved discontinuation of aflibercept did not achieve a CMT below 280 μm during the first year of treatment. However, in the second year, the CMT dropped below 280 μm and remained there for an uninterrupted dry macula duration of 10 months, which exceeded the 7.34-month threshold at which 50% of cases may discontinue injections (Fig. 2).

The case that failed to achieve discontinuation with aflibercept exhibited a saw-tooth pattern of CMT, which is a characteristic fluctuation commonly observed in eyes treated with aflibercept²⁰. As a result, although the total duration of dry macula (CMT < 280 μm) was 16 months, the uninterrupted dry macula duration was limited to 6 months, which was shorter than the 7.34-month threshold at which 50% of cases could discontinue injections (Case 2). The case that achieved discontinuation of brolucizumab reached a CMT below 280 μm after just 2 injections, and it remained under 280 μm threshold except for 2 visits. The total duration of dry macula was 22 months, with an uninterrupted dry macula duration was 9 months, which is longer than the 7.34-month threshold at which 50% of cases could discontinue injections (Case 3).

In the KESTREL and KITE trials, a CST of ≥ 320 μm on OCT was used as an inclusion criterion to confirm the presence of DME^16–18. Even in the case that did not achieve brolucizumab discontinuation, CMT dropped below 320 μm after only 2 injections and remained below this threshold at all but two visits over the 2-year study period. When the dry macular threshold was set at 280 μm, the total duration of dry macula was 16 months and the uninterrupted dry macula duration was 7 months, which was shorter than the 7.34 month threshold at which 50% of cases may discontinued injections.

The achievement rate of a CMT < 280 μm was higher (63.5%) than that observed with fixed bimonthly aflibercept injections (51.9%) at week 100 in the KESTREL and KITE studies¹⁶. In addition, the impact of achieving a CMT below 280 μm on the rate of dry macula achievement has been validated through OCT findings. The achievement rate of a dry macula—defined as the proportion of patients with neither intraretinal fluid nor subretinal fluid—was higher (59.2%) than that observed with fixed bimonthly aflibercept injections (43.1%) at week 100¹⁶. We considered that this high dry macula achievement rate could be associated with a higher rate of anti-VEGF treatment discontinuation with brolucizumab.

In diabetic retinopathy, ischemic retinal cells resulting from diabetes-induced vascular occlusion upregulate VEGF expression, which promotes neovascularization and plasma leakage as compensatory pathways to restore oxygen supply^24,25. This excessive plasma leakage is a primary cause of DME. Anti-VEGF therapy effectively suppresses this excessive leakage and improve visual acuity. Furthermore, the resulting sustained dry macula over a prolonged period may lead to selective elimination of ischemic retinal cells that overexpress VEGF. Importantly, this process removes only VEGF-overproducing ischemic cells while preserving the adjacent functional retinal tissue that is adequately perfused. Normal non-ischemic retinal cells receive adequate oxygen through patent retinal vessels and therefore do not require VEGF-induced plasma leakage as a compensatory source of oxygen supply^24,25. Successful discontinuation of anti-VEGF agents could be associated with reduced VEGF overproduction by ischemic retinal cells. As a possible explanation for how sustained dry macula may contribute to reduce VEGF production by ischemic retinal cells, we would like to propose the following hypothesis.

In this study, uninterrupted dry macula induced by brolucizumab, lasting 6 to 10 months, was associated with successful discontinuation of anti-VEGF injections. We considered that prolonged resolution of macular edema may decrease the oxygen supply to ischemic cells, as the edema fluid largely consists of leaked plasma. If oxygen deprivation persists for a sufficient period, it is conceivable that ischemic cells may undergo ischemic necrosis, which could in turn lead to cessation of VEGF overproduction.

In conclusion, our results suggest that achieving and maintaining a continuously dry macula—particularly at a CMT ≤ 290 μm—for approximately 7 to 10 months may increase the likelihood of successfully discontinuing anti-VEGF therapy in patients with DME. These findings support the use of long-acting, new-generation anti-VEGF agents, such as brolucizumab, over conventional anti-VEGF agents in proactive T&E regimens with the goal of achieving successful discontinuation of anti-VEGF injections in patients with DME.

This study had several limitations. The primary limitations were the small sample size and the retrospective design. The limited number of cases remains a key limitation of this study. Still, statistical significance was attained using point-biserial correlation coefficient and Firth’s logistic regression, both of which are well-established techniques for producing reliable results in studies with small sample sizes. The analysis was conducted on only 15 patients who met the following two criteria: (1) they had provided informed consent for the secondary use of their data at the time of enrollment in the clinical trial, and (2) they continued to be followed at the trial-participating institutions for more than three years after the conclusion of the phase III KESTREL trial. Although increasing the sample size is inherently difficult in this setting, further studies are warranted to validate these findings.

Methods

Participants

This investigator-initiated clinical study, independent of industry sponsorship, involved four Japanese institutions that had participated in the phase III brolucizumab trial for DME, known as the KESTREL trial. The study aimed to investigate the post-trial clinical course of DME patients and to examine the correlation between the duration of uninterrupted dry macula during the KESTREL trial and the subsequent discontinuation of anti-VEGF therapy after trial completion.

In this retrospective study, the analyzed data were derived from the institutions and patients who had been enrolled in the phase III KESTREL trial according to the following two criteria:

[Criterion 1] Institutions where informed consent for the secondary use of the data obtained during phase III brolucizumab KESTREL trial had been obtained at the time of enrollment. These clinical data included intravitreal injection schedules of brolucizumab or aflibercept, visual acuity, central macular thickness (CMT), and the presence of intraretinal and subretinal fluid at each visit. The same parameters were collected during the follow-up period of more than 3 years after the completion of the KESTREL trial.

[Criterion 2] Institutions that had at least one KESTREL participant who was followed for 3 years or longer after completion of the trial.

Consequently, four institutions with 19 KESTREL participants were included in the present analysis, among whom 15 were followed for 3 years or longer.

Definition of successful anti-VEGF treatment discontinuation

Successful treatment discontinuation was defined as the absence of anti-VEGF injections for three or more years following the completion of the KESTREL phase III trial of brolucizumab for DME. This definition was consistent with that used in the prior case series from the KITE phase III trial, in which 3 out of 4 eyes successfully discontinued anti-VEGF therapy after trial completion¹⁵.

Definition of dry macula

Dry macula was defined as a CMT below a predefined threshold. The standard cutoff for macular edema in the KESTREL/KITE study was 320 μm. For further analysis, more stringent thresholds were also applied in 10-µm decrements from 320 μm to 270 μm in this study. Eyes with CMT measurements below each threshold at a given time point were considered to have a dry macula according to that specific threshold.

Definition of uninterrupted dry macula duration

Uninterrupted dry macula duration was defined as the continuous period during which CMT consistently remained below a given threshold (270–320 μm) at every monthly assessment during the 2-year KESTREL trial.

Definition of total dry macula duration

The total dry macula duration was defined as the cumulative period during which CMT remained below predefined thresholds, allowing brief recurrences of DME, during the 2-year KESTREL trials.

Correlation between CMT changes and discontinuation of anti-VEGF injections

We evaluated the correlation between successful discontinuation of anti-VEGF injections and the pattern of macular edema resolution, as assessed by CMT changes.

As potential early predictive factors, we evaluated baseline CMT, the reduction in CMT after the initial injection of brolucizumab or aflibercept, and the CMT at the subsequent visit following the initial injection. As a late factor, we assessed the duration of dry macula, defined as CMT < 280 μm. The threshold of 280 μm was adopted based on findings from the KESTREL study, in which a greater proportion of brolucizumab-treated eyes achieved a CSFT < 280 μm compared with aflibercept-treated eyes. The correlation was assessed using the point-biserial correlation coefficient.

Correlation between the dry macula duration and subsequent discontinuation of anti-VEGF injections

We investigated the correlation between the longest uninterrupted period of dry macula during the KESTREL study and the subsequent discontinuation of anti-VEGF injections during the follow-up period. To assess the impact of transient DME recurrences on subsequent treatment discontinuation, we also examined the correlation between the total duration of dry macula—defined as the cumulative length of distinct dry periods interrupted by brief recurrences of DME—and the discontinuation of anti-VEGF therapy. By comparing the influence of the total dry macula duration with that of the longest single continuous dry period, we aimed to clarify how short-term recurrences of DME affect the potential for successful treatment discontinuation. The correlation was assessed using the point-biserial correlation coefficient.

Estimating the required duration of a dry macula for successful discontinuation of anti-VEGF injections

Using Firth’s logistic regression¹⁹, we estimated the number of months of continuously uninterrupted dry macula required for 50% of patients to successfully discontinue anti-VEGF therapy under each predefined CMT thresholds between 320 and 270 μm. The 95% confidence interval (CI) for the threshold estimate was derived using the Wald method based on the variance-covariance matrix of the regression coefficients.

All statistical analyses were performed using R software version 4.5.0 (R Foundation for Statistical Computing, Vienna, Austria).

Ethical considerations

This study was conducted in accordance with the tenets of the Declaration of Helsinki, and was approved by the Institutional Review Board of Shinshu University (clinical research approval number: 6349). Owing to the retrospective study design and the use of an opt-out consent process, the requirement for written informed consent was waived by the Institutional Review Board.

Author contributions

Conception and design (T.M.) Analysis (T.M., TH, S.I., K.T., M.S.) Writing the article (T.M., T.H.) Critical revision of the article (TH, S.I., K.T., M.S.) Final approval of the article (T.M., TH, S.I., K.T., M.S.) Data collection (T.M., TH, S.I., K.T., M.S.) Literature search (T.M., T.H.).

Funding

None.

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

Toshinori Murata reports personal fees from Novartis Pharma K.K., Bayer, Canon, Kowa Pharmaceutical, Senju Pharmaceutical, Chugai Pharmaceutical, Kyowa Kirin and Carl Zeiss Meditec Takao Hirano reports personal fees from Novartis Pharma K.K., Bayer, Canon, Kowa Pharmaceutical, Senju Pharmaceutical, Chugai Pharmaceutical and Carl Zeiss Meditec. Shinji Ideta reports no competing interestsKoji Tanaka reports personal fees Novartis Pharma K.K., Bayer, Santen Pharmaceutical, Kowa Pharmaceutical, Senju Pharmaceutical, Chugai Pharmaceutica, Sumitomo Phama.Masahiko Shimura reports from Novartis Pharma K.K., Bayer, Kowa Pharmaceutical, Senju Pharmaceutical, Chugai Pharmaceutical, Kyowa Kirin, Boehrnger Ingelheim, and MSD.