Purpose To analyse the clinical characteristics and treatment outcomes of choroidal neovascular membranes (CNVM) in paediatric subjects at three paediatric retina referral centres.
Methods Medical charts of patients aged 18 years or less with a diagnosis of CNVM were retrospectively reviewed. The demographic profile, laterality, presenting complaint, corrected vision, underlying pathology, fundus, fundus fluorescein angiogram and optical coherence tomogram (OCT) were analysed. CNVM type, frequency, treatment indications, recurrences and final visual acuity were noted.
Results There were a total of 35 subjects (43 eyes) with a mean age of 11.2 years. The CNVMs were mostly type 2 (90.0%), classic (90.9%), subfoveal (59.09%) and active (84.1%). Best vitelliform macular dystrophy was found to be the most common association (32.5%). Intravitreal injection of an anti-vascular endothelial growth factor (VEGF) agent was the initial therapy of choice in all. Eyes with CNVMs responsive to anti-VEGF alone required a mean of 2.11 injections. Patients with recurrent disease (21.21%) had an average of 1.14 episodes per eye. While 50% of recurrent CNVMs stabilised with repeat anti-VEGF treatment, the remaining patients required photodynamic therapy, laser or surgery.
Conclusion Paediatric CNVMs in this series differed from those in the adult population with regard to aetiology, OCT and angiographic characteristics, treatment response and rate of recurrence.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Though the incidence of choroidal neovascular membrane (CNVM) in the paediatric population is relatively small, especially when compared with that in the adult population, its impact in terms of number of blind years lived is tremendous.1 2 Limited published data on the aetiology, clinical characteristics, natural history, management strategy, outcome and recurrence rate make the management of paediatric CNVM challenging. The eyes of a growing child differ in many respects from that of an adult and hence the lessons learnt from adult CNVM treatment may not apply to the paediatric cohort in their entirety. Although randomised, controlled clinical studies with larger numbers would generate the best data for treatment efficacy, such trials are difficult and impractical to design and execute due to the rarity of paediatric patients with CNVM. Hence, data obtained from retrospective series guide treatment approaches. The present study describes the demographic profile, clinical characteristics and treatment outcomes of paediatric CNVMs in three paediatric retinal practices.
Materials and methods
This retrospective observational study was conducted at three paediatric retinal practices: (1) Department of Ophthalmology and Visual Sciences, University of Michigan, Ann Arbor, MI, USA; (2) Associated Retinal Consultants, Royal Oak, MI, USA; and (3) LV Prasad Eye Institute, Bhubaneswar, India. The study was in compliance with the principles of the Declaration of Helsinki. Medical charts were reviewed and all patients of age 18 years or younger with a diagnosis of CNVM were included. Diagnosis of CNVM was based on the clinical judgement of the paediatric retina specialist and was supported by fundus fluorescein angiogram (FFA) and optical coherence tomogram (OCT). The study period ranged from 2012 to 2014, from 2007 to 2014 and from 2007 to 2012 at study locations 1, 2 and 3 respectively. The charts with insufficient or missing data were excluded from the study. Data collected included age, gender, laterality, presenting complaint, aetiology of CNVM, presenting visual acuity, fundus picture, FFA and OCT findings. While the older OCT images belonged to time domain OCT, the newer images were from spectral domain OCT (Spectralis, Heidelberg, Germany) at study locations 1 and 2 and Cirrus at study location 3 (Carl Zeiss Meditec, Germany). We presumed the CNVM to be active if there were any one or more of the following findings: clinical evidence of exudate, fluid or haemorrhage, leakage on FFA, worsening subretinal or intraretinal fluid on OCT and increase in lesion bulk in the event of deteriorating visual acuity. CNVMs were classified as per the Macular Photocoagulation Study (MPS) protocol,3 as subfoveal (figure 1A, B), juxtafoveal (Figure 1C , D), extrafoveal (Figure 1E , F) and peripapillary (Figure 1G , H, for membranes continuous with the optic disc). In cases where FFA was not available, the CNVM was classified as per OCT and fundus photo. In cases where both FFA and OCT were not available, classification was determined based on fundus examination. A membrane localised predominantly (>90% of the lesion) in the subretinal space on OCT was classified as type 2, while the ones lying predominantly (>90% of the lesion) in the sub-retinal pigment epithelium (RPE) space were classified as type 1. Whenever possible, based on the pattern of leakage on FFA, the membrane was classified as classic (early hyperfluorescence), occult (late leakage or leakage from undetermined source) or minimally classic as per MPS protocol.3 The lesion was considered regressed if there was no dye leakage on FFA, no fluid on OCT and no subretinal fluid or haemorrhage clinically. The membrane was considered stable if the lesion characteristics and visual acuity remained unchanged for at least 6 months. The type and frequency of treatment, duration of follow-up and number of eyes that had recurrent CNVM despite treatment were analysed. We defined recurrence as the reappearance or worsening of the lesion activity after complete regression or stabilisation (as defined above). In the recurrence group, the average number of recurrences per eye, the type and frequency of treatments for the recurrences and best corrected visual acuity (BCVA) at complete regression or stabilisation or at last follow-up (for those eyes that were still under treatment) were noted. A subgroup analysis was done for those eyes that received anti-vascular endothelial growth factor (VEGF) injections only. The number of eyes that were treated on a pro re nata (PRN) basis or initial monthly loading doses was noted. Any differences in the rate of recurrences between these groups were also analysed. The mean change in visual acuity with treatment (for those that completed treatment until study endpoint or disease stabilisation) was compared for subfoveal, juxtafoveal, extrafoveal and peripapillary CNVM groups. Corrected visual acuity measured with Snellen chart was converted to logMar for statistical calculations.
Statistical analysis was done using the InStat statistical software V.3.0× for Windows (GraphPad Software, CA, USA). Descriptive statistics were used for demographic data. Kolmogorov-Smirnov test was used to determine parametric or non-parametric data. Wilcoxon-matched pair signed-rank test was used for comparison of preintervention and postintervention BCVA. Mann-Whitney U test was used to compare the improvement in BCVA between PRN and loading groups. Unpaired t-test is used to compare improvement in BCVA in subfoveal and peripapillary groups. Kruskal-Wallis test was used to compare improvement in BCVA in subfoveal, juxtafoveal, extrafoveal and peripapillary CNVM groups. Fisher’s exact test was used for comparing recurrence in PRN and loading groups. P value less than 0.05 was considered statistically significant.
The demographic profile and ocular associations are presented in table 1.
A total of 35 subjects (43 eyes) with a mean age of 11.2 years were included in the study (table 1). There were 27 subjects from USA (10 and 17 subjects at locations 1 and 2, respectively) and eight patients from study location 3 in India. The duration of follow-up ranged from 6 months to 2 years, from 6 months to 7 years and from 6 months to 5 years at locations 1, 2 and 3, respectively. The youngest patient was 1.3 years old. The CNVMs were more likely to be type 2 (90.0%), classic (90.9%), subfoveal (59.09%) and active (84.1%) (table 2) across all ethnic groups. There were a wide range of ocular conditions associated with CNVM, including retinal dystrophies, pathological myopia, postinflammatory, persistent fetal vasculature (PFV), familial exudative vitreoretinopathy (FEVR), optic nerve drusen, choroidal osteoma, optic nerve and chorioretinal coloboma, with Best vitelliform macular dystrophy being the most common association overall (32.55%) in all ethnic groups. Diagnosis of Best disease was confirmed by genetic testing in 40% of cases; other patients were diagnosed based on history, characteristic fundus appearance and electro-oculogram. The age of patients with idiopathic CNVM ranged from 4 to 17 years (mean 11.2 years). These patients underwent a detailed fundus evaluation and other investigations including OCT and FFA and when required before diagnosing the CNVM as idiopathic in all three locations.
CNVM was active in 36 of 43 eyes. Among those with active CNVM, 30 eyes underwent treatment (table 3). Five subjects (six eyes) could not be treated as they were lost to follow-up despite active CNVM at the initial visit. Intravitreal anti-VEGF agent was the initial treatment of choice for CNVM in paediatric patients in all the three centres. While bevacizumab was the most frequently used anti-VEGF at locations 1 and 3, ranibizumab, bevacizumab and pegaptanib accounted for 40%, 40% and 20%, respectively, of cases receiving anti-VEGF at location 2. Recurrent CNVM activity was treated first with an anti-VEGF agent at all locations.
The outcome of treatment with anti-VEGF agents for paediatric CNVMs is summarised in table 4. The eyes that regressed with anti-VEGF treatment alone required a mean of 2.11 injections and the gain in visual acuity (mean and median) was maximum in the peripapillary followed by subfoveal group (table 4). Among those treated for CNVM, 5 out of 14 (35.71%) among White American cases, and 2 out of 5 (40%) among Asian Indian cases and the African-American patient had recurrence. Out of three Hispanic cases, only one required treatment and did not experience any CNVM recurrence.
Six eyes (20%) were stabilised with one injection only and demonstrated no recurrence during the study period. Among the eyes that received intravitreal anti-VEGFs, eight had recurrences and these eyes had an average of 1.14 recurrences per eye. While 50% of these eyes stabilised with repeat anti-VEGF treatment, the rest required photodynamic therapy (PDT), thermal laser or submacular surgery. The rationale for treatment of recurrence CNVM was based on physician choice at each study location. There was a tendency for treating the majority of CNVMs on PRN basis. A small number of eyes received three monthly loading doses at treatment initiation. There was no significant difference in terms of gain in visual acuity or recurrent disease between the loading dose group and PRN group (table 4, P value 0.655). There were no significant ocular or systemic complications in the anti-VEGF group.
Peripapillary area (20.45%) was the second most common site of involvement (table 2). Five eyes in the peripapillary CNVM group had pre-existing pathology: optic nerve drusen, two eyes; PFV, one eye; Best vitelliform macular dystrophy, one eye; postinflammatory, one eye.
The majority of the PPCNVMs affected the foveal or juxtafoveal macula either through direct extension or indirectly by hard exudates, haemorrhage or retinal fluid. One eye with active PPCNVM and foveal involvement by subretinal haemorrhage improved spontaneously during follow-up and was managed conservatively by observation only. The CNVM regressed completely with improvement in vision from 20/50 at presentation to 20/20 at 8 months and remained stable until last follow-up.
Although the gain of 0.7752 and 0.4361 logMAR, respectively, in peripapillary and subfoveal groups after treatment was clinically significant, the mean gain in visual acuity using multiple comparisons for all CNVM subgroups did not reach statistical significance (table 4).
Paediatric CNVMs in our series differed from that in the adult population in terms of aetiology, OCT and angiographic characteristics, treatment response and the number of recurrences per eye. The ocular associations noted in this series were different from what has been reported in similar studies in adult patients as well as in the paediatric series.4–8 While two large series on paediatric CNVM report uveitis to be the most common association,4 5 the present study found that retinal dystrophies were the leading ocular association with CNVM in our centres. Best disease was the most common aetiology.9–11 Twelve patients (27.9%) did not have any identifiable ocular association. Among four eyes (9.30%) in the postinflammatory group, two had healed toxoplasma retinochoroiditis, one had choroiditis and one had idiopathic choroiditis. While recurrence is one of the major hurdles in the management of CNVMs in adult patients, the number of recurrences as a whole, as well as recurrences per eye, was much less frequent in the paediatric population.
Intravitreal anti-VEGF treatment was the initial treatment of choice for CNVM in paediatric patients in all three study centres. While bevacizumab was the most frequently used anti-VEGF agent at study locations 1 and 3, ranibizumab, bevacizumab and pegaptanib accounted for 40%, 40% and 20% of cases, respectively, receiving anti-VEGF treatment at study location 2. Ocular delivery of anti-VEGF agents has been shown to cause rapid and sustained systemic VEGF suppression in adult and paediatric patients.12 This has raised serious concerns, as VEGF is known to be vital in the development of major organs in children and the maintenance of normal organ function in adults.13 Given the difficulty of conducting prospective, randomised clinical trials in paediatric patients, retrospective analysis provides important information for long-term safety of anti-VEGF treatment in CNVM.
Children with CNVMs seem to respond well to anti-VEGF treatment and do not require prolonged treatment.4 Kozak et al showed that 60% of paediatric CNVM cases stabilised with a single intravitreal anti-VEGF injection while only 20% required repeat injections.4 We found that an average number of 2.11 anti-VEGF injections were needed for CNVM regression or stabilisation, which is significantly lower than what is seen commonly with CNVMs in the adult population. Injections on a PRN basis showed similar treatment results as monthly injections in our cohort. However, the patient numbers in each group were too small to draw any definitive conclusions on which treatment approach performs better in terms of post-treatment recurrence.
The majority of the PPCNVMs in the present series had a pre-existing ocular pathology, including optic nerve head drusen, PFV, uveitis and Best vitelliform macular dystrophy. The lesions were active in 55.5% of cases. Regardless of size and aetiology, the natural course of untreated PPCNVM has been reported to be variable, ranging from spontaneous involution to fulminant enlargement towards the fovea.14 In our series, one eye regressed spontaneously without affecting the vision; however, neither the fovea nor the juxtafoveal region was involved. The majority of the PPCNVMs had foveal or juxtafoveal involvement (seven out of nine eyes) either directly through the membrane (three eyes) or indirectly through fluid or haemorrhages (four eyes), causing vision loss and requiring treatment.
We had one patient with CNVM in an eye with PFV (previously published as a case report15). The lesion showed shrinkage in size and leakage in FFA 6 weeks after treatment with pegaptanib. At 14 weeks postinjection the lesion was inactive with fibrosis as seen on OCT (reduction in thickness and absence of fluid) and FFA (absence of any leakage). We also report CNVM in a 13-year-old patient with FEVR and high myopia (−14.75D sph x+5.00D cyl at 90° in right eye and −15.00D sph x+3.75D cyl at 105° in left eye), which has not been reported previously. Both FEVR and high myopia may contribute to the development of CNVM, but the contribution by the latter is more likely since myopia is a well-characterised risk factor for CNVM. Patient was treated with bevacizumab elsewhere prior to presentation and the lesion had minimal activity on initial evaluation. Lesion regressed completely after two more intravitreal anti-VEGF injections at 3-month intervals with one line drop in final visual acuity.
The patient who underwent submacular surgery had an excellent anatomical outcome but vision remained unchanged secondary to a large area of RPE atrophy (Figure 1H). Uemura and Thomas16 have reported good visual improvement after submacular surgery in a series of children with CNVM. The membrane was quite large in our case, and remained unchanged despite three intravitreal bevacizumab injections. Submacular surgery was done and the bulk of the membrane was removed. However, she required repeat intravitreal anti-VEGFs and PDT for residual membrane and finally regressed. She had persistent oedema at last follow-up.
There have been few reports on successful management of CNVM related to choroidal osteoma in children with both ranibizumab17 18 and bevacizumab.19 In our series, there was one patient with an active CNVM at the margin of the osteoma which was treated with three injections of intravitreal ranibizumab, followed by PDT for persistent activity as determined by FFA and OCT. A new CNVM at the edge of osteoma, away from the initial CNVM location, was noted by clinical examination and confirmed by FFA and OCT a year later. This was treated with a single dose of intravitreal ranibizumab. The subretinal haemorrhage and subretinal fluid disappeared and the lesion remained inactive (as determined by OCT) until last follow-up.
In summary, we present our series of CNVM in paediatric patients from three referral centres including a diverse patient population and extended follow-up data. The present series compares the visual outcome after anti-VEGF treatment with respect to the location of the disease, response to treatment and recurrence rates using PRN and loading dose anti-VEGF injection regimens. The retrospective nature of the series, limited visual acuity data in young children, unavailability of FFA and OCT findings in a small number of cases, different generations of OCT technology used in the studied subjects and potential referral bias for high number of patients with Best vitelliform macular dystrophy are the limitations of this study.
Best vitelliform macular dystrophy was the most common cause of CNVM in paediatric patients in our series. The visual gain was the highest in the peripapillary and subfoveal lesion groups. The overall recurrence rate in treated eyes and the number of recurrences were lower in paediatric patients compared with adult patients. The average number of injections needed for CNVM resolution was fewer compared with adult eyes. Intravitreal anti-VEGF treatment on a PRN basis after the first injection was observed to be as effective as monthly loading doses in terms of recurrences and gain in visual acuity at regression. Additional studies with larger sample sizes are needed to validate our results.
Contributors TRP: concept and design of the work; worked under the guidance of the corresponding author throughout in drafting the original to the revised version; acquisition, analysis and interpretation of data; preparation of tables and figures; compiling and checking the references; revision and approval of the original and revised versions. BJA: concept and design of the work; acquisition, analysis and interpretation of data; drafting the work; revision and approval of the original and revised versions. AMA, KAD: concept and design of the work; analysis and interpretation of data; drafting the work; revision and approval of the original and revised versions. YY, MS, RRM: assistance in the design of the work; acquisition, interpretation of data and statistical analysis; revision and approval of the original and revised versions. DA: assistance in the concept and design of the work; acquisition, analysis and interpretation of data; drafting the work; revision and approval of the original and revised versions. LPS: assistance in the design of the work; data acquisition, analysis and interpretation; drafting the work; preparation of tables and figures; revision and approval of the original and revised versions. MTT: concept and design of the work; analysis and interpretation of data; drafting the work; editing and revision to improve the intellectual content; approval of the original and revised versions. AC: concept and design of the work; critical analysis and interpretation of data; editing and revision; approval of the original and revised versions. CGB: initial concept and design of the work; guiding the first author throughout from initial draft to the final and revised version; assembled all the information and critical analysis of the draft prepared by the first author and all coauthors; edited and re-edited the drafts to increase the intellectual content; final correction and adjustment of tables and figures; correction and approval of the references; final approval of the original and revised versions. All authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Competing interests None declared.
Ethics approval Institutional Review Board of LV Prasad Eye Institute, Bhubaneswar, India, and Institutional Review Board of Kellogg Eye Center, University of Michigan.
Provenance and peer review Not commissioned; externally peer reviewed.
Presented at Annual conference, ARVO, Denver, Colorado, 2015; Annual conference, ARVO, Seattle, WA, 2016.