- Email: [email protected]
The incidence of neovascular membranes and visual field defects from optic nerve head drusen in children
The incidence of neovascular membranes and visual field defects from optic nerve head drusen in children Jared E. Duncan, MD,a Sharon F. Freedman, MD,b and Mays A. El-Dairi, MDb PURPOSE METHODS
RESULTS
CONCLUSIONS
To evaluate the incidence of visual field defects and choroidal neovascular membranes (CNVM) in a cohort of pediatric patients with optic nerve head drusen (ONHD). The medical records of children with ONHD seen at a single center from January 2012 to July 2014 were retrospectively reviewed to identify patients who had a dilated fundus examination, fundus photography, spectral domain optical coherence tomography (SDOCT)/enhanced-depth imaging (EDI) of the optic nerve head (ONH), SITA fast 24-2 Humphrey visual field (HVF) testing, lumbar puncture, and ocular ultrasound. A masked neuro-ophthalmologist analyzed fundus photographs, OCT, and fields. Retinal nerve fiber layer (RNFL) data were compared to age-matched controls. A total of 52 children (98 eyes) were included. Mean age was 10.8 3.3 years. Of these, 42 patients had visual fields (57 eyes deemed reliable), and 19 eyes had documented visual field deficits (8 were reproducible across $1 sitting [frequency 14%]). After correction of plotting errors (40 eyes), RNFL thickness was 111.9 17.9 mm. CNVM were present in 24 of 98 eyes (24.5%), with 21 of 24 located nasally (87.5%). Neither RNFL thinning nor identification of ONHD on fundus photography correlated with the presence of visual field defects. Visual field defects due to ONHD can be reliably identified in children. In eyes of children with ONHD, RNFL protocol is frequently unreliable and may overestimate RNFL thickness. EDI scans through the ONH revealed peripapillary CNVM in nearly a quarter of the patients. Further longitudinal studies looking at the progression of CNVM and visual field deficits are warranted. ( J AAPOS 2016;20:44-48)
O
ptic nerve head drusen (ONHD) are calcified deposits formed by a combination of disturbances in axonal metabolism and a small scleral canal.1 These deposits are of clinical interest as an important cause of pseudopapilledema. Various modalities are employed to help differentiate pseudopapilledema and real papilledema, including B-scan ultrasonography,2 autoflourescence,3 and optical coherence tomography (OCT).4 Although ruling out papilledema is critical, the visual sequelae of ONHD are themselves a significant cause of morbidity. Visual field deficits from ONHD are well-described in adults.5,6 More recently, visual field deficits were described in a small cohort of pediatric patients with ONHD.7 In addition to visual field deficits, several important hemorrhagic complications result from ONHD. ONHD
Author affiliations: aOphthalmology Associates of Greater Annapolis, Arnold, Maryland; b Duke Eye Center, Durham, North Carolina Presented in part as a poster at the 40th Annual Meeting of the American Association for Pediatric Ophthalmology and Strabismus, Palm Springs, California, April 2-6, 2014. Submitted August 4, 2015. Revision accepted October 29, 2015. Correspondence: Mays A. El-Dairi, MD, Duke Eye Center, Durham, NC, 27710 (email: [email protected]). Copyright Ó 2016 by the American Association for Pediatric Ophthalmology and Strabismus. 1091-8531/$36.00 http://dx.doi.org/10.1016/j.jaapos.2015.10.013
44
can be a rare cause of choroidal neovascularization membranes (CNVM) in children.8 With the advent of spectral domain OCT (SD-OCT) and enhanced-depth imaging (EDI), the diagnosis of ONHD and the importance of associated vascular sequelae continues to evolve.9,10 There are inherent difficulties in documenting OHNDrelated visual field defects and hemorrhagic complications in children, which has likely resulted in their being underreported in the scientific literature. This retrospective study documents the incidence of visual field defects and CNVM in a large cohort of children with pseudopapilledema due to ONHD.
Subjects and Methods This retrospective review was approved by the Duke University Health System Institutional Review Board and conformed to the requirements of the US Health Insurance Portability and Accountability Act of 1996. The medical records of patients\18 years of age at presentation who were referred to the pediatric neuro-ophthalmology clinic to rule out papilledema reviewed to identify those found to have pseudopapilledema due to ONHD. The absence of papilledema was confirmed by the presence of ONHD on SD-OCT (Spectralis, Heidelberg, Germany), the documentation of spontaneous venous pulsations during funduscopic examination, and a lack of change in optic nerve appearance over a 6-month period. In
Journal of AAPOS
Volume 20 Number 1 / February 2016 especially suspicious cases, neuroimaging, including cranial magnetic resonance imaging (MRI) or computed tomography (CT), and sometimes lumbar puncture were used to rule out potential causes of true papilledema. The presence of ONHD on SDOCT images was verified by a pediatric neuro-ophthalmologist (MAE) with extensive experience and training in SD-OCT interpretation. All patients underwent a complete neuro-ophthalmic examination, including best-corrected visual acuity, color vision, confrontation visual fields, cycloplegic refraction, slit lamp examination, ocular motility testing, and dilated fundus examination. As part of their neurologic work-up, all patients were asked about symptoms of raised intracranial pressure, including headache, intracranial noise, transient vision loss, and binocular diplopia. When able, ancillary testing in the form of automated visual field testing (Humphrey Visual Field Analyzer 24-2, SITA-fast protocol, Carl Zeiss Meditec, Dublin, CA), fundus photography, and Bscan ultrasonography were also performed.
OCT Protocol For the peripapillary retinal nerve fiber layer (RNFL) scan, a 20 scan was centered around the optic nerve head (ONH). The circle projected would measure 3.41 mm in a 23 mm eye. The Spectralis software then created a quantitative analysis of the peripapillary RNFL. To generate the ONH map a B-scan pattern of 73 high-resolution EDI scans was centered over the ONH over a section of 4.2 mm 4.2 mm.
Data Gathering and Statistical Analysis Baseline data was gathered on all patients, including sex, age, race, and laterality of ONHD. The use of neuroimaging, B-scan ultrasonography, and/or lumbar puncture in excluding a diagnosis of papilledema was also recorded. A masked neuro-ophthalmologist analyzed all visual fields for reliability and the presence of clinically significant visual-field defects. Visual fields with acceptable reliability were those that demonstrated both a false negative or false positive rate of \20% and a fixation loss \50% (classic recommendations in adult studies for reliability have been \33% fixation losses or \33% false positive and false negative rate); we were more generous given that this was a pediatric population.11 Visual field deficits were deemed clinically significant if reproducible across more than one visit (when available more than once). In those cases where visual field deficits were present in a subject who sat for only one visual field, the defect was recorded but was not marked as clinically significant. SD-OCT images were evaluated by a masked neuroophthalmologist for the presence of ONHD and the presence of peripapillary CNVM. When present CNVM were characterized as either nasal or temporal. Average RNFL thickness as well as sectoral RNFL thickness values were recorded for both eyes in every subject. Minor plotting errors were corrected in a masked fashion by one of the authors (MAE) using Spectralis review software. In cases where plotting errors were not amenable to correction the RNFL data was excluded from analysis. However, these scans were still analyzed for the presence of ONHD and CNVM. The presence of plotting errors, both correctable and noncorrectable, was recorded.
Journal of AAPOS
Duncan, Freedman, and El-Dairi
45
Table 1. Demographic and ancillary testing data Characteristic Patients with pseudopapilledema due to ONHD Visual acuity, range Average age on initial presentation, years Bilateral Female sex Received neuroimaging (CT or MRI) Received lumbar puncture Visible drusen on fundus photography B-scan ultrasonography diagnostic of ONHD
No. (%) 52 20/15-20/25 10.8 47 (90.4) 29 (55.8) 31 (59.6) 15 (28.8) 16 eyes (13/50 patients; 26%) 52 eyes (27/31 patients, 87.1%)
CT, computed tomography; MRI, magnetic resonance imaging; ONHD, optic nerve head drusen. Normative, age-matched OCT data was taken from another prospective, ongoing study and were used as controls for comparison. The same masked neuro-ophthalmologist reviewed all fundus photographs for the presence of visible ONHD or any signs indicating true significant papilledema (vessel obscuration, peripapillary hemorrhages, pallor). A t test was performed for one eye of each patient (right eye unless ONHD was only seen in the left eye, in which case the latter became the study eye) comparing RNFL in each quadrant and average total RNFL to the presence or absence of visual field deficits, likelihood of neuroimaging, presence of CNVM, likelihood of RNFL plotting errors, and visibility of ONHD on fundus photographs. Continuous variables are reported as mean with standard deviation.
Results A total of 52 children (average age, 10.8 3.3 years) were enrolled (98 eyes, 90.4% bilateral). Race distribution was as follows: 44 white, 5 Hispanic, 2 Asian, 1 African American (Table 1). Automated visual field data was available for 42 patients, of whom 28 of were deemed reliable in the right eye (average age, 12.3 2.6 years) and 29 reliable in the left eye (average age, 12.3 2.6 years). Average ages for those children who did not have reliable visual fields in the right eye and left eye were 10.5 2.4 years and 9.0 2.6 years, respectively. Of the 57 total eyes that initially had reliable visual fields, 19 (33%) had documented visual field defects, 12 with enlarged blind spots, 6 with arcuate scotoma, and 1 with a nasal step (e-Supplement 1, available at jaapos.org). Only 8 of these eyes had repeatable visual field deficits across more than one visual field session (average age, 11.6 3.0 years), 8 were lost to follow-up or had no subsequent visual field, and 3 cases of visual field defects were not repeatable. Plotting errors on OCT RNFL automatic segmentation were present in 40 eyes of 31 patients (41% of total eyes). These plotting errors were subsequently correctable in 34 eyes (e-Supplement 2, available at jaapos.org). The 6
46
Duncan, Freedman, and El-Dairi
Volume 20 Number 1 / February 2016
FIG 1. Stereodisc photograph and corresponding SD-OCT of a patient with pseudopapilledema due to ONHD. Note the large choroidal neovascular membrane (CNVM) that is clinically evident on fundus photography and confirmed using SD-OCT.
FIG 2. Fundus photograph and corresponding SD-OCT of a patient with pseudopapilledema due to ONHD. In this patient there is no obvious CNVM on clinical examination but a temporal CNVM is apparent on SD-OCT.
remaining eyes that had uncorrectable plotting errors were excluded from RNFL analysis but were still analyzed for the presence of CNVM. The average RNFL thickness after correction of plotting errors was 111.9 17.9 mm; in controls, 103.4 mm 10.0 mm (P 5 0.04). Patients with ONHD and RNFL \80 mm (2 standard deviations below the mean RNFL for normal eyes) were not more likely to have visual field defects (P 5 0.22). There was no statistical difference between the RNFL thickness in each sector and the presence or absence of visual field defects (P . 0.17 prior to Bonferroni correction for all quadrants/average). CNVM were present on SD-OCT images in 24 of 98 eyes (25%), in 21 of which they were nasally located; in 3 of which, temporally (Figures 1 and 2). These eyes were not more likely to have visual field deficits or significant deviations in RNFL thickness after correction of plotting errors (P 5 0.2). The presence of visible ONHD on fundus photography was not associated with a higher likelihood of detecting CNVM or visual field deficits (P . 0.13 for all). Obtaining neuroimaging (by referring physician or neuroophthalmologist) was not statistically associated with RNFL thickness, presence or absence of CNVM, RNFL plotting errors, or presence of a visual field defect (P . 0.13 for all variables). Only presence of ONHD on fundus photography was correlated with less likelihood of obtaining neuroimaging (likelihood ratio, 3.9; P 5 0.048).
Discussion Children with pseudopapilledema due to ONHD can have significant visual field deficits and CNVM. Central vision is
unlikely to be affected. To our knowledge, this is the first study to evaluate the incidence of CNVM in pediatric patients with ONHD. Literature describing reliable visual field deficits in children with ONHD is also scarce.7 Patients are often referred to subspecialty clinics to rule out dangerous causes of optic nerve head elevation, most notably optic disk edema.12 Work-up of these patients is challenging, as evidenced by the rate of neurologic workup in our patient population: 60% of our patients had neuroimaging and/or lumbar puncture, the majority of whom were imaged prior to referral. The purpose of this study was not to limit the work-up of ONHD; in fact, it has been shown that 50% of children with resolved papilledema due to idiopathic intracranial hypertension also had ONHD on OCT and ultrasound.13 Our intention was rather to emphasize the associated ophthalmic findings, which can become visually significant in these children. Visual field defects are a known sequelae of ONHD in both adults and children. The frequency of visual defects in adults has been established to be anywhere between 24% and 87%.1 Owing to the inherent difficulty in reproducing reliable visual fields in children,14 less is known about the causal relationship that ONHD have for inciting visual field loss in this population. Noval and colleagues described a small cohort of 15 children who had visual defects due to ONHD, the most common of which was a nasal defect (54%).7 However, their study did not report on the frequency of their findings; nor did they report how they assessed for visual field reliability. Using a strict cutoff for reliability, 67% of the children in our cohort were able to take reliable visual fields. Unfortunately, 8 of the 19 eyes that exhibited initial visual field defects
Journal of AAPOS
Volume 20 Number 1 / February 2016 were lost to follow-up. Nevertheless, we were still able to document significant visual field defects across more than one field in 8 of 57 cases, indicating a confirmed incidence of at least 14%. The types of visual defects we found were similar to that which has been previously reported for ONHD.1 The incidence of visual field defects in our study is less than the 51% described by Hoover and colleagues15 but closer to that found by Erkkila (11%).16 This seems to indicate that visual field defects due to ONHD may be less common in children than adults. The fact that pediatric ONHD are more frequently buried within the optic nerve head is well established.3 Because visual field defects less frequently manifest in eyes with buried ONHD,4 it stands to reason that the lack of superficial ONHD in children may explain the decreased frequency of visual field defects in our cohort. We were unable to demonstrate a relationship between visible ONHD and visual field deficits, but this could be limited by the small number of patients with visible ONHD. Many children who are referred to subspecialty clinics to “rule out papilledema” are frequently lost to follow-up once the diagnosis of “pseudopapilledema” is made. Although inherently less dangerous than causes of true optic disk edema (eg, brain tumor, inflammatory or infectious optic nerve infiltration, and increased intracranial pressure), our data support periodic follow-up in these children on the basis of visual field defects alone. Visual field loss from ONHD has been shown to evolve and worsen over time in both children17 and adults.18 Although the majority of CNVM in the present study were small and nasally located, 3 children were found to have evidence of temporal CNVM, but without central visual loss. This complication was likely under-recognized prior to the advent of SD-OCT. In their study using SDOCT, Lee and colleagues10 found that 7% of eyes in a cohort of 63 patients with ONHD had hemorrhagic complications. The higher prevalence (25%) found in our study may be a result of our using EDI, which allows for visualization of deeper buried drusen in association with subtle CNVM. Vision loss as a consequence of CNVM is a possibility. Baillif and colleagues19 reported a case of choroidal neovascularization from ONHD in a 5-year-old who presented with vision loss and subretinal fluid. He was treated with intravitreal anti-VEGF therapy, and vision was restored to baseline. Other studies have supported the efficacy of intravitreal anti-VEGF therapy,20 focal laser photocoagulation,21 and photodynamic therapy with verteporfin.22 Peripapillary CNVM may cause vision loss by either direct expansion of the membrane or by serous/hemorrhagic retinal detachment. Given this possibility of vision loss, we recommend more frequent follow-up of those children with ONHD, especially if they demonstrate evidence of CNVM on SD-OCT scans. We also found RNFL plotting errors in 41% of the SD-OCT scans, 15% of which were not correctable with
Journal of AAPOS
Duncan, Freedman, and El-Dairi
47
Spectralis review software. The increased rate of plotting errors is likely due to the presence of ONHD. Adequate characterization of RNFL thickness is important for quantifying damage to the RNFL in cases of ONHD. RNFL thinning, for example, has been attributed to ONHD and was correlated to the presence of visual field defects.23 Casado and colleagues24 suggest that serial RNFL and macular ganglion cell-inner plexiform layer measurements can be used to indicate early neuronal loss in ONHD. In addition, clinicians frequently use RNFL thickening on SD-OCT as a means of differentiating pseudopapilledema (as a result of ONHD) from true papilledema.25 We found that in cases of RNFL plotting errors on SD-OCT scans, the software tended to overestimate RNFL thickness. Given the frequency with which we detected plotting errors by analytical software, we suggest caution in RNFL interpretation unless the reader is adequately trained in identifying and correcting such errors. There are several limitations to this study, including the small number of patients, its retrospective nature, and the referral nature of the practice. The OCT data is lacking long-term longitudinal follow-up. This information would be important for attempting to describe the natural history of CNVM and for understanding their visual significance. In addition, we did not confirm the activity of CNVM with fluorescein angiography. Visual field data is also limited by the inherent unreliability of visual field testing in children. In addition, we ensured clinical significance by including only those fields that were reproducible in multiple visits. Finally, visual fields, fundus photographs, and SD-OCT images were reviewed by a single examiner, who, although masked to case identity, could have been a source of interpretive bias. In conclusion, we show that pediatric ONHD can cause significant visual field defects and CNVM. Use of SD-OCT is valuable for diagnosing ONHD, but as a result of frequent plotting errors, caution should be exercised when using RNFL data to make clinical decisions regarding the presence of optic disk edema or RNFL thinning. Further studies that look at the longitudinal progression of the CNVM in this cohort are warranted.
Literature Search The authors searched PubMed without date restriction in July 2015 using the following terms in combination: optic nerve head drusen, choroidal neovascular membrane, children, incidence. References 1. Auw-Haedrich C, Staubach F, Witschel H. Optic disc drusen. Surv Ophthalmol 2002;47:515-31. 2. Boldt HC, Byrne SF, DiBernardo C. Echographic evaluation of optic disc drusen. J Clin Neuroophthalmol 1991;11:85-91. 3. Gili P, Flores-Rodriguez P, Yanguela J, Herreros Fernandez ML. Using autoflourescence to detect optic nerve head drusen in children. J AAPOS 2013;17:568-71.
48
Duncan, Freedman, and El-Dairi
4. Kulkarni KM, Pasol J, Rosa PR, Lam BL. Differentiating mild papilledema and buried optic nerve head drusen using spectral domain optical coherence tomography. Ophthalmology 2014;121:959-63. 5. Katz BJ, Pomeranz HD. Visual field defects and retinal nerve fiber layer defects in eyes with buried optic nerve drusen. Am J Ophthalmol 2006;141:248-53. 6. Lee AG, Zimmerman MB. The rate of visual field loss in optic nerve head drusen. Am J Ophthalmol 2005;139:1062-6. 7. Noval S, Visa J, Contreras I. Visual field defects due to optic disk drusen in children. Graefes Arch Clin Exp Ophthalmol 2013;251:2445-50. 8. Gregory-Evans K, Rai P, Patterson J. Successful treatment of subretinal neovascularization with intravitreal ranibizumab in a child with optic nerve head drusen. J AAPOS 2009;21:1-4. 9. Merchant KY, Su D, Park SC, et al. Enhanced depth imaging optical coherence tomography of optic nerve head drusen. Ophthalmology 2013;120:1409-14. 10. Lee KM, Hwang JM, Woo SJ. Hemorrhagic complications of optic nerve head drusen on spectral domain optical coherence tomography. Retina 2014;34:1142-8. 11. Keltner JL, Johnson CA, Cello KE, et al., Optic Neuritis Study Group. Visual field profile of optic neuritis: a final follow-up report from the optic neuritis treatment trial from baseline through 15 years. Arch Ophthalmol 2010;128:330-37. 12. Kovarik JJ, Doshi PN, Collinge JE, Plager DA. Outcome of pediatric patients referred for papilledema. J AAPOS 2015;19:344-8. 13. Gospe SM III, Bhatti MT, El-Dairi MA. Anatomic and visual outcomes of pediatric idiopathic intracranial hypertension. Epub ahead of print, August 12, 2015. Br J Ophthalmol 2015. 14. Akar Y, Yilmaz A, Yucel I. Assessment of an effective visual field testing strategy for a normal pediatric population. Ophthalmologica 2008;222:329-33. 15. Hoover DL, Robb RM, Petersen RA. Optic disc drusen in children. J Pediatr Ophthalmol Strabismus 1998;25:191-5.
Volume 20 Number 1 / February 2016 16. Erkkila H. Clinical appearance of optic disc drusen in childhood. Graefes Arch Clin Exp Ophthalmol 1975;193:1-18. 17. Frisen L. Evolution of drusen of the optic nerve head over 23 years. Acta Ophthalmol 2008;86:111-12. 18. Mustonen E. Pseudopapilledema with and without verified optic disc drusen. A clinical analysis II: visual fields. Acta Ophthalmol 1983;61: 1057-66. 19. Baillif S, Nguyen E, Colleville-El Hayek A, Betis F. Long term followup after a single intravitreal ranibizumab injection for choroidal neovascularisation secondary to optic nerve head drusen in a 5-year-old child. Graefes Arch Clin Exp Ophthalmol 2013;251:1657-9. 20. Knape RM, Zavaleta EM, Clark CL 3rd, Khuddus N, Peden MC. Intravitreal bevacizumab treatment of bilateral peripapillary choroidal neovascularization from optic nerve head drusen. J AAPOS 2011; 15:87-90. 21. Delyfer MN, Rougier MB, Fourmaux E, Cousin P, Korobelnik JF. Laser photocoagulation for choroidal neovascular membrane associated with optic disc drusen. Acta Ophthalmol Scand 2004;82:236-8. 22. Silva R, Torrent T, Loureiro R, Travassos A, de Abreu JR. Bilateral CNV associated with optic nerve drusen treated with photodynamic therapy with verteperforin. Eur J Ophthalmol 2004;14:434-7. 23. Roh S, Noecker RJ, Schuman JS, Hedges TR 3rd, Weiter JJ, Mattox C. Effect of optic nerve head drusen on nerve fiber layer thickness. Ophthalmology 1998;105:878-85. 24. Casado A, Rebolleda G, Guerrero L, et al. Measurement of retinal nerve fiber layer and macular ganglion cell-inner plexiform layer with spectral-domain optical coherence tomography in patients with optic nerve head drusen. Graefes Arch Clin Exp Ophthalmol 2014;252:1653-60. 25. Fard M, Fakhree S, Abdi P, Hassanpoor N, Subramanian PS. Quantification of peripapillary total retinal volume in pseudopapilledema and mild papilledema using spectral-domain optical coherence tomography. Am J Ophthalmol 2014;158:136-43.
Seeing Is Believing So ophthalmologists hate blindness. It represents failure and displays our inability to help, to fix, to heal. And the most heart rending cases are those of blind children, who were perhaps unlucky enough to have had severe retinopathy of prematurity or congenital glaucoma. You can’t help but pity the parents who may have to support the child into their old age. These are the times when life seems the most unfair, and when we physicians feel the most helpless. —Andrew Lam, Saving Sight (Bokeelia, Fla.: Irie Books, 2013), 163.
Journal of AAPOS
RESULTS
CONCLUSIONS
To evaluate the incidence of visual field defects and choroidal neovascular membranes (CNVM) in a cohort of pediatric patients with optic nerve head drusen (ONHD). The medical records of children with ONHD seen at a single center from January 2012 to July 2014 were retrospectively reviewed to identify patients who had a dilated fundus examination, fundus photography, spectral domain optical coherence tomography (SDOCT)/enhanced-depth imaging (EDI) of the optic nerve head (ONH), SITA fast 24-2 Humphrey visual field (HVF) testing, lumbar puncture, and ocular ultrasound. A masked neuro-ophthalmologist analyzed fundus photographs, OCT, and fields. Retinal nerve fiber layer (RNFL) data were compared to age-matched controls. A total of 52 children (98 eyes) were included. Mean age was 10.8 3.3 years. Of these, 42 patients had visual fields (57 eyes deemed reliable), and 19 eyes had documented visual field deficits (8 were reproducible across $1 sitting [frequency 14%]). After correction of plotting errors (40 eyes), RNFL thickness was 111.9 17.9 mm. CNVM were present in 24 of 98 eyes (24.5%), with 21 of 24 located nasally (87.5%). Neither RNFL thinning nor identification of ONHD on fundus photography correlated with the presence of visual field defects. Visual field defects due to ONHD can be reliably identified in children. In eyes of children with ONHD, RNFL protocol is frequently unreliable and may overestimate RNFL thickness. EDI scans through the ONH revealed peripapillary CNVM in nearly a quarter of the patients. Further longitudinal studies looking at the progression of CNVM and visual field deficits are warranted. ( J AAPOS 2016;20:44-48)
O
ptic nerve head drusen (ONHD) are calcified deposits formed by a combination of disturbances in axonal metabolism and a small scleral canal.1 These deposits are of clinical interest as an important cause of pseudopapilledema. Various modalities are employed to help differentiate pseudopapilledema and real papilledema, including B-scan ultrasonography,2 autoflourescence,3 and optical coherence tomography (OCT).4 Although ruling out papilledema is critical, the visual sequelae of ONHD are themselves a significant cause of morbidity. Visual field deficits from ONHD are well-described in adults.5,6 More recently, visual field deficits were described in a small cohort of pediatric patients with ONHD.7 In addition to visual field deficits, several important hemorrhagic complications result from ONHD. ONHD
Author affiliations: aOphthalmology Associates of Greater Annapolis, Arnold, Maryland; b Duke Eye Center, Durham, North Carolina Presented in part as a poster at the 40th Annual Meeting of the American Association for Pediatric Ophthalmology and Strabismus, Palm Springs, California, April 2-6, 2014. Submitted August 4, 2015. Revision accepted October 29, 2015. Correspondence: Mays A. El-Dairi, MD, Duke Eye Center, Durham, NC, 27710 (email: [email protected]). Copyright Ó 2016 by the American Association for Pediatric Ophthalmology and Strabismus. 1091-8531/$36.00 http://dx.doi.org/10.1016/j.jaapos.2015.10.013
44
can be a rare cause of choroidal neovascularization membranes (CNVM) in children.8 With the advent of spectral domain OCT (SD-OCT) and enhanced-depth imaging (EDI), the diagnosis of ONHD and the importance of associated vascular sequelae continues to evolve.9,10 There are inherent difficulties in documenting OHNDrelated visual field defects and hemorrhagic complications in children, which has likely resulted in their being underreported in the scientific literature. This retrospective study documents the incidence of visual field defects and CNVM in a large cohort of children with pseudopapilledema due to ONHD.
Subjects and Methods This retrospective review was approved by the Duke University Health System Institutional Review Board and conformed to the requirements of the US Health Insurance Portability and Accountability Act of 1996. The medical records of patients\18 years of age at presentation who were referred to the pediatric neuro-ophthalmology clinic to rule out papilledema reviewed to identify those found to have pseudopapilledema due to ONHD. The absence of papilledema was confirmed by the presence of ONHD on SD-OCT (Spectralis, Heidelberg, Germany), the documentation of spontaneous venous pulsations during funduscopic examination, and a lack of change in optic nerve appearance over a 6-month period. In
Journal of AAPOS
Volume 20 Number 1 / February 2016 especially suspicious cases, neuroimaging, including cranial magnetic resonance imaging (MRI) or computed tomography (CT), and sometimes lumbar puncture were used to rule out potential causes of true papilledema. The presence of ONHD on SDOCT images was verified by a pediatric neuro-ophthalmologist (MAE) with extensive experience and training in SD-OCT interpretation. All patients underwent a complete neuro-ophthalmic examination, including best-corrected visual acuity, color vision, confrontation visual fields, cycloplegic refraction, slit lamp examination, ocular motility testing, and dilated fundus examination. As part of their neurologic work-up, all patients were asked about symptoms of raised intracranial pressure, including headache, intracranial noise, transient vision loss, and binocular diplopia. When able, ancillary testing in the form of automated visual field testing (Humphrey Visual Field Analyzer 24-2, SITA-fast protocol, Carl Zeiss Meditec, Dublin, CA), fundus photography, and Bscan ultrasonography were also performed.
OCT Protocol For the peripapillary retinal nerve fiber layer (RNFL) scan, a 20 scan was centered around the optic nerve head (ONH). The circle projected would measure 3.41 mm in a 23 mm eye. The Spectralis software then created a quantitative analysis of the peripapillary RNFL. To generate the ONH map a B-scan pattern of 73 high-resolution EDI scans was centered over the ONH over a section of 4.2 mm 4.2 mm.
Data Gathering and Statistical Analysis Baseline data was gathered on all patients, including sex, age, race, and laterality of ONHD. The use of neuroimaging, B-scan ultrasonography, and/or lumbar puncture in excluding a diagnosis of papilledema was also recorded. A masked neuro-ophthalmologist analyzed all visual fields for reliability and the presence of clinically significant visual-field defects. Visual fields with acceptable reliability were those that demonstrated both a false negative or false positive rate of \20% and a fixation loss \50% (classic recommendations in adult studies for reliability have been \33% fixation losses or \33% false positive and false negative rate); we were more generous given that this was a pediatric population.11 Visual field deficits were deemed clinically significant if reproducible across more than one visit (when available more than once). In those cases where visual field deficits were present in a subject who sat for only one visual field, the defect was recorded but was not marked as clinically significant. SD-OCT images were evaluated by a masked neuroophthalmologist for the presence of ONHD and the presence of peripapillary CNVM. When present CNVM were characterized as either nasal or temporal. Average RNFL thickness as well as sectoral RNFL thickness values were recorded for both eyes in every subject. Minor plotting errors were corrected in a masked fashion by one of the authors (MAE) using Spectralis review software. In cases where plotting errors were not amenable to correction the RNFL data was excluded from analysis. However, these scans were still analyzed for the presence of ONHD and CNVM. The presence of plotting errors, both correctable and noncorrectable, was recorded.
Journal of AAPOS
Duncan, Freedman, and El-Dairi
45
Table 1. Demographic and ancillary testing data Characteristic Patients with pseudopapilledema due to ONHD Visual acuity, range Average age on initial presentation, years Bilateral Female sex Received neuroimaging (CT or MRI) Received lumbar puncture Visible drusen on fundus photography B-scan ultrasonography diagnostic of ONHD
No. (%) 52 20/15-20/25 10.8 47 (90.4) 29 (55.8) 31 (59.6) 15 (28.8) 16 eyes (13/50 patients; 26%) 52 eyes (27/31 patients, 87.1%)
CT, computed tomography; MRI, magnetic resonance imaging; ONHD, optic nerve head drusen. Normative, age-matched OCT data was taken from another prospective, ongoing study and were used as controls for comparison. The same masked neuro-ophthalmologist reviewed all fundus photographs for the presence of visible ONHD or any signs indicating true significant papilledema (vessel obscuration, peripapillary hemorrhages, pallor). A t test was performed for one eye of each patient (right eye unless ONHD was only seen in the left eye, in which case the latter became the study eye) comparing RNFL in each quadrant and average total RNFL to the presence or absence of visual field deficits, likelihood of neuroimaging, presence of CNVM, likelihood of RNFL plotting errors, and visibility of ONHD on fundus photographs. Continuous variables are reported as mean with standard deviation.
Results A total of 52 children (average age, 10.8 3.3 years) were enrolled (98 eyes, 90.4% bilateral). Race distribution was as follows: 44 white, 5 Hispanic, 2 Asian, 1 African American (Table 1). Automated visual field data was available for 42 patients, of whom 28 of were deemed reliable in the right eye (average age, 12.3 2.6 years) and 29 reliable in the left eye (average age, 12.3 2.6 years). Average ages for those children who did not have reliable visual fields in the right eye and left eye were 10.5 2.4 years and 9.0 2.6 years, respectively. Of the 57 total eyes that initially had reliable visual fields, 19 (33%) had documented visual field defects, 12 with enlarged blind spots, 6 with arcuate scotoma, and 1 with a nasal step (e-Supplement 1, available at jaapos.org). Only 8 of these eyes had repeatable visual field deficits across more than one visual field session (average age, 11.6 3.0 years), 8 were lost to follow-up or had no subsequent visual field, and 3 cases of visual field defects were not repeatable. Plotting errors on OCT RNFL automatic segmentation were present in 40 eyes of 31 patients (41% of total eyes). These plotting errors were subsequently correctable in 34 eyes (e-Supplement 2, available at jaapos.org). The 6
46
Duncan, Freedman, and El-Dairi
Volume 20 Number 1 / February 2016
FIG 1. Stereodisc photograph and corresponding SD-OCT of a patient with pseudopapilledema due to ONHD. Note the large choroidal neovascular membrane (CNVM) that is clinically evident on fundus photography and confirmed using SD-OCT.
FIG 2. Fundus photograph and corresponding SD-OCT of a patient with pseudopapilledema due to ONHD. In this patient there is no obvious CNVM on clinical examination but a temporal CNVM is apparent on SD-OCT.
remaining eyes that had uncorrectable plotting errors were excluded from RNFL analysis but were still analyzed for the presence of CNVM. The average RNFL thickness after correction of plotting errors was 111.9 17.9 mm; in controls, 103.4 mm 10.0 mm (P 5 0.04). Patients with ONHD and RNFL \80 mm (2 standard deviations below the mean RNFL for normal eyes) were not more likely to have visual field defects (P 5 0.22). There was no statistical difference between the RNFL thickness in each sector and the presence or absence of visual field defects (P . 0.17 prior to Bonferroni correction for all quadrants/average). CNVM were present on SD-OCT images in 24 of 98 eyes (25%), in 21 of which they were nasally located; in 3 of which, temporally (Figures 1 and 2). These eyes were not more likely to have visual field deficits or significant deviations in RNFL thickness after correction of plotting errors (P 5 0.2). The presence of visible ONHD on fundus photography was not associated with a higher likelihood of detecting CNVM or visual field deficits (P . 0.13 for all). Obtaining neuroimaging (by referring physician or neuroophthalmologist) was not statistically associated with RNFL thickness, presence or absence of CNVM, RNFL plotting errors, or presence of a visual field defect (P . 0.13 for all variables). Only presence of ONHD on fundus photography was correlated with less likelihood of obtaining neuroimaging (likelihood ratio, 3.9; P 5 0.048).
Discussion Children with pseudopapilledema due to ONHD can have significant visual field deficits and CNVM. Central vision is
unlikely to be affected. To our knowledge, this is the first study to evaluate the incidence of CNVM in pediatric patients with ONHD. Literature describing reliable visual field deficits in children with ONHD is also scarce.7 Patients are often referred to subspecialty clinics to rule out dangerous causes of optic nerve head elevation, most notably optic disk edema.12 Work-up of these patients is challenging, as evidenced by the rate of neurologic workup in our patient population: 60% of our patients had neuroimaging and/or lumbar puncture, the majority of whom were imaged prior to referral. The purpose of this study was not to limit the work-up of ONHD; in fact, it has been shown that 50% of children with resolved papilledema due to idiopathic intracranial hypertension also had ONHD on OCT and ultrasound.13 Our intention was rather to emphasize the associated ophthalmic findings, which can become visually significant in these children. Visual field defects are a known sequelae of ONHD in both adults and children. The frequency of visual defects in adults has been established to be anywhere between 24% and 87%.1 Owing to the inherent difficulty in reproducing reliable visual fields in children,14 less is known about the causal relationship that ONHD have for inciting visual field loss in this population. Noval and colleagues described a small cohort of 15 children who had visual defects due to ONHD, the most common of which was a nasal defect (54%).7 However, their study did not report on the frequency of their findings; nor did they report how they assessed for visual field reliability. Using a strict cutoff for reliability, 67% of the children in our cohort were able to take reliable visual fields. Unfortunately, 8 of the 19 eyes that exhibited initial visual field defects
Journal of AAPOS
Volume 20 Number 1 / February 2016 were lost to follow-up. Nevertheless, we were still able to document significant visual field defects across more than one field in 8 of 57 cases, indicating a confirmed incidence of at least 14%. The types of visual defects we found were similar to that which has been previously reported for ONHD.1 The incidence of visual field defects in our study is less than the 51% described by Hoover and colleagues15 but closer to that found by Erkkila (11%).16 This seems to indicate that visual field defects due to ONHD may be less common in children than adults. The fact that pediatric ONHD are more frequently buried within the optic nerve head is well established.3 Because visual field defects less frequently manifest in eyes with buried ONHD,4 it stands to reason that the lack of superficial ONHD in children may explain the decreased frequency of visual field defects in our cohort. We were unable to demonstrate a relationship between visible ONHD and visual field deficits, but this could be limited by the small number of patients with visible ONHD. Many children who are referred to subspecialty clinics to “rule out papilledema” are frequently lost to follow-up once the diagnosis of “pseudopapilledema” is made. Although inherently less dangerous than causes of true optic disk edema (eg, brain tumor, inflammatory or infectious optic nerve infiltration, and increased intracranial pressure), our data support periodic follow-up in these children on the basis of visual field defects alone. Visual field loss from ONHD has been shown to evolve and worsen over time in both children17 and adults.18 Although the majority of CNVM in the present study were small and nasally located, 3 children were found to have evidence of temporal CNVM, but without central visual loss. This complication was likely under-recognized prior to the advent of SD-OCT. In their study using SDOCT, Lee and colleagues10 found that 7% of eyes in a cohort of 63 patients with ONHD had hemorrhagic complications. The higher prevalence (25%) found in our study may be a result of our using EDI, which allows for visualization of deeper buried drusen in association with subtle CNVM. Vision loss as a consequence of CNVM is a possibility. Baillif and colleagues19 reported a case of choroidal neovascularization from ONHD in a 5-year-old who presented with vision loss and subretinal fluid. He was treated with intravitreal anti-VEGF therapy, and vision was restored to baseline. Other studies have supported the efficacy of intravitreal anti-VEGF therapy,20 focal laser photocoagulation,21 and photodynamic therapy with verteporfin.22 Peripapillary CNVM may cause vision loss by either direct expansion of the membrane or by serous/hemorrhagic retinal detachment. Given this possibility of vision loss, we recommend more frequent follow-up of those children with ONHD, especially if they demonstrate evidence of CNVM on SD-OCT scans. We also found RNFL plotting errors in 41% of the SD-OCT scans, 15% of which were not correctable with
Journal of AAPOS
Duncan, Freedman, and El-Dairi
47
Spectralis review software. The increased rate of plotting errors is likely due to the presence of ONHD. Adequate characterization of RNFL thickness is important for quantifying damage to the RNFL in cases of ONHD. RNFL thinning, for example, has been attributed to ONHD and was correlated to the presence of visual field defects.23 Casado and colleagues24 suggest that serial RNFL and macular ganglion cell-inner plexiform layer measurements can be used to indicate early neuronal loss in ONHD. In addition, clinicians frequently use RNFL thickening on SD-OCT as a means of differentiating pseudopapilledema (as a result of ONHD) from true papilledema.25 We found that in cases of RNFL plotting errors on SD-OCT scans, the software tended to overestimate RNFL thickness. Given the frequency with which we detected plotting errors by analytical software, we suggest caution in RNFL interpretation unless the reader is adequately trained in identifying and correcting such errors. There are several limitations to this study, including the small number of patients, its retrospective nature, and the referral nature of the practice. The OCT data is lacking long-term longitudinal follow-up. This information would be important for attempting to describe the natural history of CNVM and for understanding their visual significance. In addition, we did not confirm the activity of CNVM with fluorescein angiography. Visual field data is also limited by the inherent unreliability of visual field testing in children. In addition, we ensured clinical significance by including only those fields that were reproducible in multiple visits. Finally, visual fields, fundus photographs, and SD-OCT images were reviewed by a single examiner, who, although masked to case identity, could have been a source of interpretive bias. In conclusion, we show that pediatric ONHD can cause significant visual field defects and CNVM. Use of SD-OCT is valuable for diagnosing ONHD, but as a result of frequent plotting errors, caution should be exercised when using RNFL data to make clinical decisions regarding the presence of optic disk edema or RNFL thinning. Further studies that look at the longitudinal progression of the CNVM in this cohort are warranted.
Literature Search The authors searched PubMed without date restriction in July 2015 using the following terms in combination: optic nerve head drusen, choroidal neovascular membrane, children, incidence. References 1. Auw-Haedrich C, Staubach F, Witschel H. Optic disc drusen. Surv Ophthalmol 2002;47:515-31. 2. Boldt HC, Byrne SF, DiBernardo C. Echographic evaluation of optic disc drusen. J Clin Neuroophthalmol 1991;11:85-91. 3. Gili P, Flores-Rodriguez P, Yanguela J, Herreros Fernandez ML. Using autoflourescence to detect optic nerve head drusen in children. J AAPOS 2013;17:568-71.
48
Duncan, Freedman, and El-Dairi
4. Kulkarni KM, Pasol J, Rosa PR, Lam BL. Differentiating mild papilledema and buried optic nerve head drusen using spectral domain optical coherence tomography. Ophthalmology 2014;121:959-63. 5. Katz BJ, Pomeranz HD. Visual field defects and retinal nerve fiber layer defects in eyes with buried optic nerve drusen. Am J Ophthalmol 2006;141:248-53. 6. Lee AG, Zimmerman MB. The rate of visual field loss in optic nerve head drusen. Am J Ophthalmol 2005;139:1062-6. 7. Noval S, Visa J, Contreras I. Visual field defects due to optic disk drusen in children. Graefes Arch Clin Exp Ophthalmol 2013;251:2445-50. 8. Gregory-Evans K, Rai P, Patterson J. Successful treatment of subretinal neovascularization with intravitreal ranibizumab in a child with optic nerve head drusen. J AAPOS 2009;21:1-4. 9. Merchant KY, Su D, Park SC, et al. Enhanced depth imaging optical coherence tomography of optic nerve head drusen. Ophthalmology 2013;120:1409-14. 10. Lee KM, Hwang JM, Woo SJ. Hemorrhagic complications of optic nerve head drusen on spectral domain optical coherence tomography. Retina 2014;34:1142-8. 11. Keltner JL, Johnson CA, Cello KE, et al., Optic Neuritis Study Group. Visual field profile of optic neuritis: a final follow-up report from the optic neuritis treatment trial from baseline through 15 years. Arch Ophthalmol 2010;128:330-37. 12. Kovarik JJ, Doshi PN, Collinge JE, Plager DA. Outcome of pediatric patients referred for papilledema. J AAPOS 2015;19:344-8. 13. Gospe SM III, Bhatti MT, El-Dairi MA. Anatomic and visual outcomes of pediatric idiopathic intracranial hypertension. Epub ahead of print, August 12, 2015. Br J Ophthalmol 2015. 14. Akar Y, Yilmaz A, Yucel I. Assessment of an effective visual field testing strategy for a normal pediatric population. Ophthalmologica 2008;222:329-33. 15. Hoover DL, Robb RM, Petersen RA. Optic disc drusen in children. J Pediatr Ophthalmol Strabismus 1998;25:191-5.
Volume 20 Number 1 / February 2016 16. Erkkila H. Clinical appearance of optic disc drusen in childhood. Graefes Arch Clin Exp Ophthalmol 1975;193:1-18. 17. Frisen L. Evolution of drusen of the optic nerve head over 23 years. Acta Ophthalmol 2008;86:111-12. 18. Mustonen E. Pseudopapilledema with and without verified optic disc drusen. A clinical analysis II: visual fields. Acta Ophthalmol 1983;61: 1057-66. 19. Baillif S, Nguyen E, Colleville-El Hayek A, Betis F. Long term followup after a single intravitreal ranibizumab injection for choroidal neovascularisation secondary to optic nerve head drusen in a 5-year-old child. Graefes Arch Clin Exp Ophthalmol 2013;251:1657-9. 20. Knape RM, Zavaleta EM, Clark CL 3rd, Khuddus N, Peden MC. Intravitreal bevacizumab treatment of bilateral peripapillary choroidal neovascularization from optic nerve head drusen. J AAPOS 2011; 15:87-90. 21. Delyfer MN, Rougier MB, Fourmaux E, Cousin P, Korobelnik JF. Laser photocoagulation for choroidal neovascular membrane associated with optic disc drusen. Acta Ophthalmol Scand 2004;82:236-8. 22. Silva R, Torrent T, Loureiro R, Travassos A, de Abreu JR. Bilateral CNV associated with optic nerve drusen treated with photodynamic therapy with verteperforin. Eur J Ophthalmol 2004;14:434-7. 23. Roh S, Noecker RJ, Schuman JS, Hedges TR 3rd, Weiter JJ, Mattox C. Effect of optic nerve head drusen on nerve fiber layer thickness. Ophthalmology 1998;105:878-85. 24. Casado A, Rebolleda G, Guerrero L, et al. Measurement of retinal nerve fiber layer and macular ganglion cell-inner plexiform layer with spectral-domain optical coherence tomography in patients with optic nerve head drusen. Graefes Arch Clin Exp Ophthalmol 2014;252:1653-60. 25. Fard M, Fakhree S, Abdi P, Hassanpoor N, Subramanian PS. Quantification of peripapillary total retinal volume in pseudopapilledema and mild papilledema using spectral-domain optical coherence tomography. Am J Ophthalmol 2014;158:136-43.
Seeing Is Believing So ophthalmologists hate blindness. It represents failure and displays our inability to help, to fix, to heal. And the most heart rending cases are those of blind children, who were perhaps unlucky enough to have had severe retinopathy of prematurity or congenital glaucoma. You can’t help but pity the parents who may have to support the child into their old age. These are the times when life seems the most unfair, and when we physicians feel the most helpless. —Andrew Lam, Saving Sight (Bokeelia, Fla.: Irie Books, 2013), 163.
Journal of AAPOS