Clinical study data

The medical records of 202 patients (207 eyes) with mCNV who received anti-VEGF therapy for the first time in the hospital were reviewed retrospectively. One hundred and fifty-six eyes were excluded: 72 had a follow-up period of less than 12 months, 66 had missing or incomplete critical follow-up data, 8 had image quality less than three leading to blurred angiography, 4 had other internal surgery, 3 had retinal detachment during the follow-up period and 3 had suspicious age-related macular degeneration. Further research was conducted on 51 eyes from 46 patients. Among all the cases, 3 had hypertension, no diabetes, 14 had cataract surgery before the observation period and all the ethnic groups were Han people.

Table 1 displays the demographic information of the 51 eyes. The average age was 53.88 ± 14.01 years old, with a refractive error of − 12.46 ± 3.12 D. The median baseline BCVA was 0.70 (0.40–1.30) logMAR, and the follow-up duration was 27 (20–41) months. 33 (64.7%) eyes had been diagnosed with subfoveal CNV while 18 (35.3%) with parafoveal CNV. The percentage of “Medusa”, “Seafan” and “indistinct network” were 22 (43.1%), 8 (15.7%), and 21 (41.2%) at baseline. The median baseline CNV area was 0.361 (0.180–0.897) mm2, and the median baseline CNV flow area was 0.299 (0.136–0.646) mm2. 16 (31.4%) eyes had lacquer cracks, 45 (88.2%) eyes had posterior staphyloma, 14 (27.5%) eyes had retinoschisis and 17 (33.3%) eyes had DCV. The median hyperreflective area height was 210 (170–286) µm, the median CFT was 300 (219–381) µm and the median SFCT was 54 (26–72) µm.

Table 1 Baseline characteristics of eyes with myopic choroidal neovascularizationClinical demographic comparisons between mCNV with and without neovascular signal of mCNV around PSV

The presence of PSV was observed in all eyes (51 eyes). Neovascular signal of CNV around PSV was found in 20 eyes (39.2%), and not in 31 eyes (60.8%). A total of three eyes showed differences during observation, and agreement was reached through further observation and discussion. The baseline data and the treatment situation were compared in Table 2.

Table 2 Clinical demographic comparisons between mCNVs based on the presence of neovascular signal around PSV

The occurrence rate of DCV in the group with neovascular signal around PSV was higher than that in the group without neovascular signal around PSV (50.0% vs. 22.6%, P = 0.043). The height of hyperreflective area between two groups were not significantly different [249.5 (185.25–318.5) μm vs. 197 (150–268) µm, P = 0.080)]. There was no significant difference between the other baseline data and the treatment situation.

Higher rate of recurrence occurred in eyes with neovascular signal around PSV

Twenty of 51 eyes (39.2%) experienced at least one recurrence throughout the follow-up period, of which 10 eyes (50.0%) had a reactivation of original CNV, 4 eyes (20.0%) had further enlargement of the original CNV, and 6 eyes (30.0%) had new CNV formation. For all patients, the recurrence rate within one year was 15.7%. The mean number of anti-VEGF injections was 3.06 times (median, 3.00 times). The overall mean interval between first recurrence was estimated to be 34.43 months (median, 44.00 months, range, 4.00–47.00 months) by Kaplan–Meier survival analysis.

Within one year after treatment, 6 eyes (30.0%) in the group with neovascular signal around PSV had at least one recurrence, and 2 eyes (6.5%) in group without neovascular signal around PSV had at least one recurrence. Higher rate of recurrence occurred in eyes with neovascular signal around PSV (P = 0.045) in the first year.

Kaplan–Meier survival analysis showed that the average interval of first recurrence was 28.68 months (median interval 36.00 months, range, 4.00–47.00 months), and the average interval of first recurrence was 39.55 months (median recurrence interval could not be predicted, range 11.00–20.00 months). The recurrence rate in the group with neovascular signal around PSV was significantly higher than that in the group without neovascular signal around PSV (Fig. 3, Log Rank χ2 = 14.712, P = 0.030). The one-year recurrence rates of the two groups were 30.0% and 6.5% and the two-year recurrence rates were 46.2% and 25.6%, respectively.

Fig. 3

Kaplan–Meier survival curve on recurrence of CNV based on the neovascular signal around PSV during follow-up. The Kaplan–Meier survival curve shows the survival rate each time the event of recurrence in CNV with neovascular signal around PSV group (red line) and CNV without neovascular signal around PSV group (blue line). Log-rank P = 0.030. CNV, choroidal neovascularization; PSV, perforating scleral vessel

When we conducted survival analysis separately based on drugs, the recurrence probability in the group with neovascular signal around PSV was significantly higher than that in the group without neovascular signal around PSV (Log Rank χ2 = 7.332, P = 0.007) when treated with conbercept. However, there was no difference in the recurrence probability (Log Rank χ2 = 0.364, P = 0.547) when treated with ranibizumab.

The risk factors for recurrence were analyzed by Cox proportional hazard regression model (Table 3). Univariate Cox regression analysis was used to analyze the presence or absence of neovascular signal around PSV, baseline BCVA, baseline CNV area, presence or absence of lacquer crack, type of anti-VEGF drugs and SFCT. In univariate analysis, the presence of neovascular signal around PSV and SFCT were associated with the increased risk of CNV recurrence, while baseline BCVA, baseline CNV area, presence or absence of lacquer crack and type of anti-VEGF drugs were not significantly correlated with the risk of CNV recurrence. In the multivariate analysis, the presence of neovascular signal around PSV [P = 0.026, HR: 2.904, 95% confidence interval (CI): 1.134–7.436] and SFCT (P = 0.034, HR: 0.368, 95% CI: 0.146–0.928) were still significant. The presence of neovascular signal around PSV and SFCT ≤ 50 μm are risk factors for CNV recurrence.

Table 3 The risk factors with recurrence of mCNV

Additional analysis revealed that 15 recurrent eyes with neovascular signal around PSV, 7 (53.8%) eyes were reactivated by the original CNV, 3 (23.1%) eyes had an enlargement of the original CNV, and 3 (23.1%) eyes developed a new CNV that developed away from the original CNV (Table 4). In 7 recurrent eyes without neovascular signal around PSV, 3 (42.9%) eyes were reactivated by the original CNV, 1 (14.3%) eye had an enlargement of the original CNV, and 3 (42.9%) eyes developed a new CNV that developed away from the original CNV. The distribution of recurrence types did not differ (P = 0.698). When observing the neovascular signal in CNV which had recurrences, 9 (69.2%) eyes were found PSV around the recurrent CNV in the presence of neovascular signal around PSV group, while 1 (14.3%) eye in the absence of neovascular signal around PSV group was found PSV around the recurrent CNV. However, there was no significant difference between two groups (P = 0.057).

Table 4 Case series of recurred mCNV patients with neovascular signal around PSVChanges of BCVA, hyperreflective area height, CNV area and CNV flow area in eyes with neovascularization around PSV

The changes of BCVA, hyperreflective area height, CNV area and CNV flow area were compared between the two groups through repeated measures analysis of variance (Additional file 1: Table S1). Since the original data is not normally distributed, all the data are converted into normally distributed data by open square root processing.

For BCVA, the main effect of follow-up time was statistically significant (F = 10.507, P < 0.001, η2p = 0.477), BCVA was significantly improved. The main effect of the group was statistically significant (F = 5.418, P = 0.024, η2p = 0.100), BCVA of the group with neovascular signal around PSV was worse (Fig. 4a).

Fig. 4

Changes of BCVA and CNV flow area change with time based on the neovascular signal around PSV after anti-VEGF. The x-axis stands for the serial times, y-axis stands for normal distribution data of logMAR BCVA and CNV flow area after converting to the open square root. The influence of groups on CNV flow area was significantly different over time by ANONA (Pinteraction = 0.027). BCVA, best-corrected visual acuity; CNV, choroidal neovascularization; PSV, perforating scleral vessel; anti-VEGF, anti-vascular endothelial growth factor

For hyperreflective area height, the main effect of follow-up time was statistically significant (F = 12.174, P < 0.001, η2p = 0.555). The main effect was not statistically significant (F = 0.093, P = 0.761, η2p = 0.002).

For CNV area, the main effect of follow-up time was statistically significant (F = 4.422, P = 0.009, η2p = 0.555), and the area of CNV was significantly reduced. The main effect was not statistically significant (F = 0.044, P = 0.836, η2p = 0.002).

For CNV flow area, the main effect of follow-up time was statistically significant (F =  7.711, P < 0.001, η2p = 0.562). The CNV flow area was significantly reduced. The main effect was not statistically significant (F = 0.029, P = 0.866, η2p = 0.001). The influence of groups on CNV flow area was significantly different over time (F = 3.326, P = 0.027, η2p = 0.357) (Fig. 4b). Further simple effect analysis showed that there was no significant difference in simple effect of CNV flow area between the two groups at baseline, 1 month, 3 months, 6 months and 12 months (P = 0.479, 0.308, 0.859, 0.613, 0.825, respectively). Multiple comparative findings: in the group with neovascular signal around PSV, the CNV flow area at 1 month, 3 months and 12 months were significantly decreased compared with baseline (P < 0.001, P = 0.001, P = 0.012, respectively), while there was no significant difference at 6 months (P = 0.074). In the group without neovascular signal around PSV, the CNV flow area was significantly decreased compared with baseline at three months (P = 0.043), while there was no significant difference in the other groups (P = 1.000, P = 0.163, P = 0.134, respectively). In summary, the CNV flow area of the group with neovascular signal around PSV decreased most significantly at one month and then increased, while the CNV flow area of the group without neovascular signal around PSV showed a decreasing trend, but there was no significant difference.