Using spectral-domain optical coherence tomography to detect optic neuropathy in patients with craniosynostosis Linda R. Dagi, MD,a Laura M. Tiedemann, BS,a Gena Heidary, MD, PhD,a Caroline D. Robson, MB, ChB,b Amber M. Hall, MPH,c and David Zurakowski, PhDc,d BACKGROUND
Detecting and monitoring optic neuropathy in patients with craniosynostosis is a clinical challenge due to limited cooperation, and subjective measures of visual function. The purpose of this study was to appraise the correlation of peripapillary retinal nerve fiber layer (RNFL) thickness measured by spectral-domain ocular coherence tomography (SD-OCT) with indication of optic neuropathy based on fundus examination.
METHODS
The medical records of all patients with craniosynostosis presenting for ophthalmic evaluation during 2013 were retrospectively reviewed. The following data were abstracted from the record: diagnosis, historical evidence of elevated intracranial pressure, current ophthalmic evaluation and visual field results, and current peripapillary RNFL thickness.
RESULTS
A total of 54 patients were included (mean age, 10.6 years [range, 2.4-33.8 years]). Thirteen (24%) had evidence of optic neuropathy based on current fundus examination. Of these, 10 (77%) demonstrated either peripapillary RNFL elevation and papilledema or depression with optic atrophy. Sensitivity for detecting optic atrophy was 88%; for papilledema, 60%; and for either form of optic neuropathy, 77%. Specificity was 94%, 90%, and 83%, respectively. Kappa agreement was substantial for optic atrophy (k 5 0.73) and moderate for papilledema (k 5 0.39) and for either form of optic neuropathy (k 5 0.54). Logistic regression indicated that peripapillary RNFL thickness was predictive of optic neuropathy (P \ 0.001). Multivariable analysis demonstrated that RNFL thickness measurements were more sensitive at detecting optic neuropathy than visual field testing (likelihood ratio 5 10.02; P 5 0.002). Sensitivity and specificity of logMAR visual acuity in detecting optic neuropathy were 15% and 95%, respectively.
CONCLUSIONS
Peripapillary RNFL thickness measured by SD-OCT provides adjunctive evidence for identifying optic neuropathy in patients with craniosynostosis and appears more sensitive at detecting optic atrophy than papilledema. ( J AAPOS 2014;18:543-549)
C
hildren with craniosynostosis (CS) suffer premature fusion of one or more cranial sutures in utero and/or during the first years of life. Early fusion can result in increased intracranial pressure (ICP) prior to or after cranial vault expansion.1,2 Persistent elevation
Author affiliations: Departments of aOphthalmology, bRadiology, cAnesthesia, and dSurgery, Boston Children’s Hospital; Harvard Medical School, Boston, Massachusetts Financial support: Endowment funds from the Children’s Hospital Ophthalmology Foundation Chair and the Discovery Award, both granted to Linda R. Dagi, MD. Presented at the 40th Annual Meeting of the American Association for Pediatric Ophthalmology and Strabismus, Palm Springs, California, April 2-6, 2014. Submitted May 22, 2014. Revision accepted July 29, 2014. Correspondence: Linda R. Dagi, MD, Department of Ophthalmology, Boston Children’s Hospital, 300 Longwood Avenues, Fegan 4, Boston, MA 02115 (email: Linda.Dagi@ childrens.harvard.edu). Copyright Ó 2014 by the American Association for Pediatric Ophthalmology and Strabismus. 1091-8531/$36.00 http://dx.doi.org/10.1016/j.jaapos.2014.07.177
Journal of AAPOS
in ICP may be related to continued arrest of cranial growth, obstructive hydrocephalus, or complications of sleep apnea.1-5 Monitoring this vulnerable population for impending optic neuropathy is challenging due to competing etiologies of vision loss (astigmatism, amblyopia, strabismus, and exposure keratopathy) and limited cooperation due to age, cognitive skills, or photophobia.6-8 Pattern-reversal visual evoked potentials can aid in evaluation, but pediatric normative data is not universally available, testing is time consuming, and interpretation requires highly skilled support staff. Frequent direct measurement of ICP by lumbar puncture or pressure bolt augments risk and is invasive.9 The purpose of this study was to assess the ability of peripapillary retinal nerve fiber layer (RNFL) thickness measured by spectral-domain optical coherence tomography (SD-OCT) to detect optic neuropathy compared to fundus examination in patients with CS. Additionally, we compared indication of optic neuropathy by RNFL
543
544
Dagi et al
thickness and fundus examination with evidence of altered visual function measured by logMAR visual acuity and visual field testing, and we investigated the correlation of abnormal fundus examination with historical evidence of elevated ICP.
Subjects and Methods This study was approved by the Boston Children’s Hospital Institutional Review Board and conformed to the requirements of the US Health Insurance Portability and Accountability Act of 1996. The medical records of patients $2 years of age with CS of any type presenting at Boston Children’s Hospital for ophthalmic evaluation during 2013 were retrospectively reviewed. Patients with a history of premature birth, anterior segment anomaly, glaucoma, retinal degeneration, or dystrophy were excluded.10,11 Age, sex, CS diagnosis, and history of elevation in ICP based on radiographic confirmation or copper beaten skull deformity documented at the time of cranial vault expansion were recorded. Current best-corrected logMAR visual acuity in the bettersighted (nonamblyopic) eye, cycloplegic refraction, color vision by Hardy Rand Rittler (HRR) or Ishihara color plates, visual fields obtained by Humphrey Field Analyzer (Zeiss 750i-series Humphrey Field Analyzer II, Dublin, CA) or Goldmann kinetic perimetry, optic nerve appearance, and peripapillary RNFL thickness measured by Heidelberg Spectralis SD-OCT (Heidelberg Engineering, GmbH, Heidelberg, Germany) were also documented. Patients were classified as moderately hyperopic if spherical equivalent (SE) was between 13.25 D and 16.00 D and highly hyperopic if between 16.25 D and 19.50 D.12 Classification of excess myopia required SE . 4.00 D.13
Visual Fields Visual fields undertaken with Goldmann kinetic perimetry were performed on patients too young or cognitively limited to perform Humphrey field testing and analyzed using a modified Wall and George grading system.14 The Humphrey SITA fast central 24-2 test was chosen for older or more cooperative patients. These fields were graded using guidelines described by Thomas and George.15
Spectral-domain OCT Peripapillary RNFL thickness measurements were obtained by two experienced technicians using the N-site Analytics scan protocol (Heidelberg Engineering, GmbH, Heidelberg, Germany). The Spectralis captures details at a speed of 40,000 A-scans per second, with axial resolution of 3.9 mm and a transverse resolution of 14 mm.16 Automated real-time tracking eliminates artifacts from eye movements in order to generate high-resolution scans. The N-site axonal protocol involves a circle B-scan of approximately 3.45 mm centered around the optic nerve.16 Measurements initiate and end nasally, allowing for detailed assessment of the papillomacular bundle.17 Scans with artifacts, misalignment errors, or quality scores below 23 were rejected.
Volume 18 Number 6 / December 2014 Historical Evidence of Elevated ICP Radiological findings were assessed by an experienced pediatric neuroradiologist who was masked to ophthalmic findings. Historical CT and/or MR scans of the brain were reviewed. Presence of copper-beaten pattern, progressive ventriculomegaly, shunted hydrocephalus, or jugular foraminal stenosis (JFS) with enlarged occipitomastoid emissary veins (LEV)18,19 was deemed sufficient to confirm previous elevation in ICP. Additionally, patients with copper-beaten skull deformity documented at cranial vault expansion were included in the group of patients with historical evidence of elevated ICP.
Designation of Optic Neuropathy A pediatric ophthalmologist with 25 years’ experience determined the presence of optic neuropathy (papilledema or optic atrophy) based on fundus examination and photographic confirmation utilizing the Non-Mydriatric Retinal Camera (Topcon Corp, Model TRC-NW8F). Optic atrophy was defined as clinically and photographically apparent optic nerve pallor or hypoplasia, sectorally or globally, with diminution of vascularity and confirmatory nerve fiber layer drop-out on red-free fundus view. Papilledema was designated in the presence of clinically and photographically apparent optic nerve swelling, with at least some obscuration of vasculature at the disk border. Current optic nerve designation (normal, atrophic, or edematous) was compared with current peripapillary RNFL thickness. Correlation of current optic nerve appearance with logMAR acuity, visual field results, and historical evidence of elevation in ICP were also investigated.
Statistical Analysis One eye per patient was included for analysis. In patients with amblyopia, the nonamblyopic eye was chosen. If neither eye was amblyopic, the eye with the higher quality SD-OCT RNFL scan was included. Peripapillary RNFL thickness was compared to published pediatric norms for patients under age 1813,20 and to validated adult norms for older patients. Pediatric normative data was derived from published values13,20 (N 5 83; N 5 107) and these combined to yield an N of 190 patients. Standard error mean13 was used to calculate standard deviation for each RNFL region—nasal, nasal inferior, temporal inferior, temporal, temporal superior, nasal superior, and global (SD 5 SEM * SQRT[sample size]). Mean thicknesses from the studies were combined for each RNFL region (Combined Mean 5 [N1 * Mean1 1 N2 * Mean2]/[N11N2]), as were the standard deviations (Combined SD 5 SQRT [(SD1)2 1 (SD2)2]). From the combined data, we determined the 1st, 5th, 95th, and 99th percentiles for each RNFL region (P1 5 mean [2.33 * SD]; P5 5 mean [1.64 * SD]; P95 5 mean 1 [1.64 * SD]; P99 5 mean 1 [2.33 * SD]). See Table 1 for combined percentile data. For each pediatric patient, a thickness (mm) versus position ( ) graph beginning ( 180 ) and ending (180 ) nasally was created using curved lines to compare patient data to combined pediatric 1st, 5th, 95th, and 99th percentiles. This visual representation appears similar to adult Heidelberg SD-OCT peripapillary RNFL thickness output graph. An example is provided in Figure 1.
Journal of AAPOS
Volume 18 Number 6 / December 2014
Dagi et al
545
Table 1. Combined pediatric normative data, derived from Turk et al20 and Yanni et al13 Region Nasal Nasal inferior Temporal inferior Temporal Temporal superior Nasal superior
Position, degrees 180; 180 135 45 0 45 135
1st Percentile
5th Percentile
95th Percentile
99th Percentile
30.6 37.0 85.8 29.3 79.5 38.6
44.4 60.0 103.5 42.9 97.9 59.3
110.0 169.4 187.8 107.6 185.4 158.0
123.8 192.4 205.6 121.2 203.8 178.8
Sensitivity and specificity were calculated with 95% confidence intervals to evaluate the diagnostic characteristics of SD-OCT RNFL thickness, logMAR acuity, visual field assessment, and history of elevated ICP compared to current fundus appearance.21 Cohen’s kappa coefficient (k) was used as the chance-corrected measure of agreement with fundus examination. Standard criteria on the strength of agreement is from Landis and Koch.22 Positive and negative predictive values of SD-OCT were computed using Bayes’ theorem and a 25% estimated prevalence of optic neuropathy from fundus examination. Statistical power was calculated using nQuery Advisor, version 7.0 (Statistical Solutions, Saugus, MA). To ascertain whether SD-OCT provides additional predictive information, multivariable logistic regression was applied in the subgroup that had both SD-OCT and visual field results (n 5 36), using the likelihood ratio test as the criteria for statistical significance.23,24 Point-biserial correlation coefficient (rpb) comparing logMAR acuity to optic neuropathy as identified by fundus examination was performed. Statistical analysis was completed using SPSS version 21.0 (IBM, Armonk, NY). Two-tailed P values of \0.05 were considered statistically significant.
Results A total of 65 patients were studied; 11 were excluded because of limited cooperation, severe nystagmus, poor scan quality, or retinal degeneration. Mean age at examination was 10.6 5.4 years (median, 9.3 4.6 years; range, 2.4-33.8 years). The cohort consisted of 33 females (61%) and 21 males (39%) and included a wide range of diagnoses (Apert, Pfeiffer, Crouzon, Saethre-Chotzen, isolated metopic synostosis, Muenke, unicoronal, bicoronal, and atypical multisuture variants). Abstracted clinical characteristics are summarized in Table 2. No patient displayed a relative afferent pupillary defect. All patients had a best-corrected visual acuity in their nonamblyopic or included eye of 0.48 logMAR or better (mean, 0.03). Of the 54 included patients, 47 patients had normal visual acuity ( 0.12 to 0.10 logMAR), 5 had a visual acuity of 0.18 to 0.30, and 2 patients had a visual acuity of 0.40 to 0.48. Point-biserial correction coefficient comparing logMAR acuity to optic neuropathy, as identified by fundus exam, demonstrated a moderate positive correlation (rpb 5 0.329, P 5 0.015), confirming that visual acuities were worse among some patients with evidence of optic neuropathy based on fundus examination.
Journal of AAPOS
FIG 1. Pediatric peripapillary RNFL thickness, fundus examination, and Goldmann perimetry. Graphs were created using combined pediatric normative data from Turk and colleauges20 and Yanni and colleagues.13 Patient 1 has papilledema noted by (A) peripapillary RNFL, (B) fundus examination, and (C) visual fields. D-F demonstrate these features for patient 2 with optic atrophy. BA, borderline above; BB, borderline below.
546
Dagi et al
Volume 18 Number 6 / December 2014
Table 2. Patient characteristics Sex, age, and diagnosis Male Female Median age at exam, years Diagnosis Apert Bicoronal Crouzon Metopic Pfeiffer Othera Saethre-Chozen Sagittal Unilateral coronal
Radiographic and clinical features 21 (39%) 33 (61%) 9.3 (4.6) 4 4 4 4 3 10 7 10 8
JFS and LEVb JFS without LEVb Hydrocephalus/Ventriculomegalyb Copper-beaten patternb Chiari malformationb History indicating elevated ICPb Amblyopia Mean best-corrected visual acuity (logMAR) Mean spherical equivalent Color abnormalities Field deficitsc Papilledema on fundus exam Optic atrophy on fundus exam
21 (44%) 4 (0.8%) 15 (31%) 14 (29%) 8 (17%) 30 (63%) 13 (24%) 0.03 0.82 3 (6%) 11 (31%) 5 (9%) 8 (15%)
JFS, jugular foraminal stenosis; LEV, emissary veins. a Includes Apert/Pfeiffer-like, Crouzoid, Crouzon-like, Muenke, and multiple suture. b Based on imaging and intraoperative findings available for 48 patients. c Based on data base of 36 reliable visual fields.
Sensitivity was 15% and specificity was 95% for the association of logMAR acuity $0.3 with optic neuropathy. Two patients had mild color vision defects on HRR testing, missing two test plates, and 1 patient had classic protanomaly/deuteranomaly dyschromatopsia. Neither had abnormal fundus appearance, RNFL thickness, or history of elevated ICP. Spectral-domain OCT Analyzed scans had a mean quality score of 31.3 5.1 (range, 23-43). All but 3 were obtained on the day of fundus examination. Those not obtained that day were completed within 3 weeks, without intervening intervention. There was evidence of optic neuropathy on fundus examination in 13 patients (8 with optic atrophy; 5 with papilledema). Of these, 10 with optic neuropathy had thinner-than-normative range RNFL and optic atrophy and thicker-than normative range with papilledema. SD-OCT sensitivity was 88% for classifying patients with optic atrophy; 60% for those with papilledema; and 77% for those with either form of optic neuropathy. One patient with disk edema and another with optic atrophy had RNFL thickness measurements within the normal range. Peripapillary RNFL thickness outside the pediatric normative range (\18 years of age) and outside the standard adult normative range for all others was seen in 7 of 41 patients (17%) with a normal fundus appearance. SD-OCT specificity was 94% in classifying patients without evidence of optic atrophy; 90% for those without papilledema; and 83% for those without either. Kappa agreement between SD-OCT results and fundus appearance was moderate to substantial: k 5 0.73 for optic atrophy; k 5 0.39 for papilledema; and k 5 0.54 for either form of optic neuropathy (Table 3). Logistic regression analysis confirmed that SDOCT was predictive of optic neuropathy findings on fundus
examination (OR: 16.2; 95% CI, 3.5-74.4; likelihood ratio test 5 15.75; P \ 0.001). Among the patients classified as abnormal on SD-OCT, 59% would be expected to have optic neuropathy (positive predictive value). Among patients with a normal SD-OCT test, 92% would be expected to have a normal fundus examination (negative predictive value). nQuery Advisor software demonstrated that our sample size of 54 patients provides 80% power to assess the predictive value of RNFL thickness abnormality as a predictor of optic neuropathy. Patients with apparent optic atrophy on SD-OCT (n 5 11) exhibited it more temporally (90.0%) than nasally (50.0%), but there was a more even distribution temporally (62.5%) and nasally (87.5%) in patients with papilledema on SD-OCT (n 5 8). Of the 4 patients with moderate and 1 with high hyperopia, 3 had normal RNFL thickness measurements, 1 had clinical and SD-OCT evidence of optic atrophy (14.25 SE), and 1 demonstrated a measured increase in RNFL thickness (15.25 SE), without other ophthalmic features or history of elevation in ICP. Considering SD-OCT as a possible predictor of papilledema on fundus examination and controlling for moderate to high hyperopia as a possible confounding variable, we found that SD-OCT was a significant predictor of papilledema irrespective of the presence of moderate to high hyperopia (OR: 22.0; 95% CI, 3.0-159.7; likelihood ratio test 5 10.13; P \ 0.001). Among the 10 patients who presented with myopia, only 1 had higher myopia ( 5.25 SE), and she had frank papilledema and an associated increase in RNFL thickness. There were 49 pediatric patients, 14 (29%) of whom had abnormal examinations based on pediatric and adult normative data. Twenty-one of 35 pediatric patients (60%) had unremarkable SD-OCT results based on pediatric normative data but would have been considered abnormal based on adult norms standardly provided at this time.
Journal of AAPOS
Volume 18 Number 6 / December 2014
Dagi et al
547
Table 3. OCT, history, visual fields, and LogMAR visual acuity as predictors of optic neuropathya Variable
Sensitivity
95% CI
Specificity
95% CI
Kappa
OCT: optic atrophy OCT: papilledema OCT: optic neuropathy Historical evidence [ICPb Visual fieldc LogMAR VA
7/8 (88%) 3/5 (60%) 10/13 (77%) 11/13 (85%)
53%-98% 23%-88% 50%-92% 58%-96%
43/46 (94%) 44/49 (90%) 34/41 (83%) 16/35 (46%)
83%-98% 78%-96% 69%-91% 30%-62%
0.73 0.39 0.54 0.22
6/10 (60%) 2/13 (15%)
32%-83% 4%-42%
21/26 (81%) 39/41 (95%)
62%-91% 84%-99%
0.40 0.14
CI, confidence interval; OCT, optical coherence tomography; VA, visual acuity. a Diagnostic characteristics based on fundus appearance as the gold standard. b Historical evidence based on radiographic features and intraoperative findings consistent with elevated ICP based on 48 patients with radiology available for review. c Based on 36 patients with visual field data and fundus examination data.
Visual Fields A subset of 36 patients had complete and reliable visual fields. In comparing agreement of visual field findings with fundus appearance, sensitivity was 60% and specificity was 81%, with moderate kappa agreement (k 5 0.40). Eleven patients had a visual field abnormality, 8 with abnormal peripapillary RNFL thickness. Visual field testing was normal for 25 patients, but 6 had abnormal RNFL thickness. Therefore, agreement between SD-OCT and visual field modalities was moderate (75% agreement; k 5 0.45). Multivariable logistic regression indicated that, independent of visual field results, information obtained by SD-OCT provides significant prediction of what is seen by fundus examination regarding the presence of optic neuropathy (OR: 13.3; 95% CI, 2.2-80.5; likelihood ratio test 5 10.02; P 5 0.002). The presence of an abnormal visual field along with optic neuropathy on fundus examination guaranteed predicted SD-OCT abnormalities. History of Elevated ICP No neuroimaging was available for 6 of 54 patients. For the rest, imaging studies consisted of both CT and MR of the brain (n 5 19), CT alone (n 5 25), or MR alone (n 5 4). Imaging characteristics are summarized in Table 2 and detailed in e-Supplement 1 (available at jaapos.org). In 30 patients, radiographic or intraoperative features confirmed prior elevation in ICP. Table 3 compares RNFL thickness by SD-OCT with historical evidence of elevation in ICP, visual field results, and logMAR acuity testing as predictors of current optic neuropathy based on fundus examination. Figure 2 illustrates this graphically with likelihood ratios.
Discussion Peripapillary RNFL thickness measured by SD-OCT confirmed fundus examination findings of optic neuropathy with a sensitivity of 77% and a specificity of 83%. However, sensitivity for detecting optic atrophy was much higher (88%) than for detecting papilledema (60%). Positive predictive value (59%) was found to be
Journal of AAPOS
FIG 2. Likelihood ratio of SD-OCT, history, visual fields, and logMAR acuity predicting optic neuropathy. The likelihood ratio reflects the likelihood of optic neuropathy given positive test results. If the SD-OCT result is positive for optic neuropathy, the likelihood that the patient has optic neuropathy is nearly 5 times greater than if it is negative.
more modest than negative predictive value (92%). Lower positive predictive value is anticipated when disease incidence is low—only 13 of our 54 patients had optic neuropathy. Having abnormal visual fields, along with evidence of optic neuropathy on fundus examination, guaranteed predicted RNFL findings. Whereas a “gold standard” combining altered visual function (visual acuity and visual field abnormality) with abnormal fundus examination would have resulted in an absolute correlation between the presence of optic neuropathy on clinical examination and abnormal peripapillary RNFL measurements, we did not design our study with this comprehensive approach. Measured visual acuity may not be accurate in this population and would be unlikely to show decline with mild papilledema. Furthermore, accurate visual field testing is often unavailable in those with limited cognitive function. We found that only two-thirds of patients had
548
Dagi et al
Volume 18 Number 6 / December 2014
FIG 3. Comparison of information provided by fundus examination, SD-OCT, and visual fields for a patient presenting with and treated for papilledema. A, Initial presentation. B, Two months after treatment with acetazolamide there was minimal change on fundus examination but clear reduction in peripapillary RNFL thickness and improved visual field. C, Five months after initiating treatment noted decline in papilledema on fundus examination and SD-OCT but visual field was unchanged.
reliable visual field data. To assess the correlation of function with optic nerve appearance, we chose logMAR visual acuity worse than 0.3 as criteria for evidence of visual acuity loss. The 2 patients with poorest logMAR acuity (0.48) had evidence of optic atrophy. Nevertheless, only a modest correlation was confirmed between diminished visual acuity and optic neuropathy: only 4 patients had “diminished” acuity by our standard. We suspect that early referral and attentive management protected most from significant visual acuity loss. Although there is no consensus about the impact of amblyopia on peripapillary RNFL thickness, amblyopia has been reported to reduce it.25,26 Since we sought evidence of vision loss from optic neuropathy, and not from amblyopia, we studied acuity and RNFL thickness characteristics from the nonamblyopic eye. Abnormal peripapillary RNFL thickness was documented in 7 of 41 patients with normal fundus appearance. The 5 with elevated thickness had the following characteristics: history of elevation in ICP (3), chronic sleep apnea (1), moderate hyperopia (1), and/or optic nerve head drusen documented with echography (1). Excessive hyperopia and optic nerve head drusen are known confounders, increasing RNFL thickness.14,27,28 The 2 patients with reduced thickness had either a history of elevated ICP or microcephaly. Perhaps SD-OCT identified some patients with optic neuropathy that were missed on clinical examination. Myopia is also a confounder, associated with the
decreased RNFL thickness.29 Our patient with 5.25 D of myopia had frank papilledema, thus overwhelming this potential confounder. Longitudinal monitoring of peripapillary RNFL measurements in this population may provide added value. The patient described in Figure 3 presented with papilledema and was treated with acetazolamide in anticipation of shunt placement. RNFL thickness measurements taken over several months demonstrated a more definitive change after 2 months of treatment than did fundus appearance. We suspect that SD-OCT monitoring will prove even more important for patients with otherwise occult increase in optic atrophy over time. This study has several limitations. A combination of fundus appearance and functional loss was contemplated as a gold standard for optic neuropathy but proved impractical, given considerations noted above. Assessment with patternreversal visual evoked potentials might have enriched our analysis. Axial length, known to affect peripapillary RNFL thickness,30 was not collected, but significant hyperopia and myopia were used as reasonable surrogates. Peripapillary total RNFL thickness31,32 or volume33 may be more sensitive at detecting papilledema compared to standard RNFL thickness but were not obtained. Fundus appraisal was made by one observer. Lastly, the classification of patients into those with evidence of elevated ICP was subject to historical and incomplete data including imaging studies which might have predated the onset of elevated ICP.
Journal of AAPOS
Volume 18 Number 6 / December 2014
References 1. Liasis A, Thompson DA, Hayward R, Nischal KK. Sustained raised intracranial pressure implicated only by pattern reversal visual evoked potentials after cranial vault expansion surgery. Pediatr Neurosurg 2003;39:75-80. 2. Campbell JW, Albright AL, Losken HW, Biglan AW. Intracranial hypertension after cranial vault decompression for craniosynostosis. Pediatr Neurosurg 1995;22:270-73. 3. Fok H, Jones BM, Gault DG, Andar U, Hayward R. Relationship between intracranial pressure and intracranial volume in craniosynostosis. Br J Plast Surg 1992;45:394-7. 4. Liasis A, Nischal KK, Walters B, et al. Monitoring visual function in children with syndromic craniosynostosis: a comparison of 3 methods. Arch Ophthalmol 2006;124:1119-26. 5. Tuite GF, Chong WK, Evanson J, et al. The effectiveness of papilledema as an indicator of raised intracranial pressure in children with craniosynostosis. Neurosurgery 1996;38:272-8. 6. Khan SH, Nischal KK, Dean F, Hayward RD, Walker J. Visual outcomes and amblyogenic risk factors in craniosynostotic syndromes: a review of 141 cases. Br J Ophthalmol 2003;87:999-1003. 7. Nischal KK. Ocular aspects of craniofacial disorders. Am Orthopt J 2002;52:58-68. 8. McGraw PV, Winn B, Gray LS, Elliott DB. Improving the reliability of visual acuity measures in young children. Ophthalmic Physiol Opt 2000;20:173-84. 9. Rangwala LM, Liu GT. Pediatric idiopathic intracranial hypertension. Surv Ophthalmol 2007;52:597-617. 10. Goldberg MF, Duke JR. Ocular histopathology in Hunter’s syndrome: systemic mucopolysaccharidosis type II. Arch Ophthalmol 1967;77:503-12. 11. Liang F, Audo I, Sahel JA, Paques M. Retinal degeneration in mucopolysaccharidose type II. Graefes Arch Clin Exp Ophthalmol 2013; 251:1871-2. Epub December 4, 2012. 12. Tas M, Oner V, Turkcu FM, et al. Peripapillary retinal nerve fiber layer thickness in hyperopic children. Optom Vis Sci 2012;89:1009-13. 13. Yanni SE, Wang J, Cheng CS, et al. Normative reference ranges for the retinal nerve fiber layer, macula, and retinal layer thicknesses in children. Am J Ophthalmol 2013;155:354-360.e1. 14. Wall M, George GE. Idiopathic intracranial hypertension a prospective study of 50 patients. Brain 1991;114(Pt 1A):115-80. 15. Thomas R, George R. Interpreting automated perimetry. Indian J Ophthalmol 2001;49:125-40. 16. Spectralis HRA1OCT User Guide Software Version 5.3. Heidelberg, Germany: Heidelberg Engineering GmbH; 2010. 17. Rebolleda G, Gonzalez-Lopez JJ, Munoz-Negrete FJ, Oblanca N, Costa-Frossard L, Alvarez Cermeno JC. Color-code agreement among stratus, cirrus, and spectralis optical coherence tomography in relapsing-remitting multiple sclerosis with and without prior optic neuritis. Am J Ophthalmol 2013;155:890-97. Epub February 8, 2013.
Journal of AAPOS
Dagi et al
549
18. Robson CD, Mulliken JB, Robertson RL, et al. Prominent basal emissary foramina in syndromic craniosynostosis: correlation with phenotypic and molecular diagnoses. Am J Neuroradiol 2000;21:1707-17. 19. Taylor WJ, Hayward RD, Lasjaunias P, et al. Enigma of raised intracranial pressure in patients with complex craniosynostosis: the role of abnormal intracranial venous drainage. J Neurosurg 2001;94:377-85. 20. Turk A, Ceylan OM, Arici C, et al. Evaluation of the nerve fiber layer and macula in the eyes of healthy children using spectral-domain optical coherence tomography. Am J Ophthalmol 2012;153:552-559.e1. 21. Fletcher RW, Fletcher SW. Diagnosis. In: Clinical Epidemiology: The Essentials. 4th ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2005:35-58. 22. Landis RJ, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159-74. 23. Katz MH. Interpreting the Results. In: Multivariable Analysis: A Practical Guide for Clinicians and Public Health Researchers. 3rd ed. New York, NY: Cambridge University Press; 2011:140-61. 24. Hidalgo B, Goodman M. Multivariate or Multivariable Regression? Am J Public Health 2013;103:39-40. 25. Bandyopadhyay SK, Chatterjee A, Banerjee R. Peripapillary nerve fibre layer thickness and macular thickness in children with anisometropic amblyopia attending a referral centre in Eastern India. J Indian Med Assoc 2012;110:542-5. 26. Colen TP, de Faber JT, Lemij HG. Retinal nerve fiber layer thickness in human strabismic amblyopia. Bincol Vis Strabismus Q 2000;15: 141-6. 27. Sarac O, Tasci YY, Gurdal C, Can I. Differentiation of optic disc edema from optic nerve head drusen with spectral-domain optical coherence tomography. J Neuroophthalmol 2012;32:207-11. 28. Lee KM, Woo SJ, Hwang J-M. Differentiation of optic nerve head drusen and optic disc edema with spectral-domain optical coherence tomography. Ophthalmology 2011;118:971-7. 29. Choi SW, Lee SJ. Thickness changes in the fovea and peripapillary retinal nerve fiber layer depend on the degree of myopia. Korean J Ophthalmol 2006;20:215-19. 30. Oner V, Aykut V, Tas M, Alakus MF, Iscan Y. Effect of refractive status on peripapillary retinal nerve fibre layer thickness: a study by RTVue spectral domain optical coherence tomography. Br J Ophthalmol 2013;97:75-9. 31. Vartin VC, Nguyen AM, Balmitgere T, Bernard M, Tilikete C, Vighetto A. Detection of mild papilloedema using special domain optical coherence tomography. Br J Ophthalmol 2012;96:375-9. 32. Scott CJ, Kardon RH, Lee AG, Frisen L, Wall M. Diagnosis and grading of papilledema in patients with raised intracranial pressure using optical coherence tomography vs clinical expert assessment using a clinical staging scale. Arch Ophthalmol 2010;128:705-11. 33. Wang J-K, Kardon RH, Kupersmith MJ, Garvin MK. Automated quantification of volumetric optic disc swelling in papilledema using spectral-domain optical coherence tomography. Invest Ophthalmol Vis Sci 2012;53:4069-75.
Detecting and monitoring optic neuropathy in patients with craniosynostosis is a clinical challenge due to limited cooperation, and subjective measures of visual function. The purpose of this study was to appraise the correlation of peripapillary retinal nerve fiber layer (RNFL) thickness measured by spectral-domain ocular coherence tomography (SD-OCT) with indication of optic neuropathy based on fundus examination.
METHODS
The medical records of all patients with craniosynostosis presenting for ophthalmic evaluation during 2013 were retrospectively reviewed. The following data were abstracted from the record: diagnosis, historical evidence of elevated intracranial pressure, current ophthalmic evaluation and visual field results, and current peripapillary RNFL thickness.
RESULTS
A total of 54 patients were included (mean age, 10.6 years [range, 2.4-33.8 years]). Thirteen (24%) had evidence of optic neuropathy based on current fundus examination. Of these, 10 (77%) demonstrated either peripapillary RNFL elevation and papilledema or depression with optic atrophy. Sensitivity for detecting optic atrophy was 88%; for papilledema, 60%; and for either form of optic neuropathy, 77%. Specificity was 94%, 90%, and 83%, respectively. Kappa agreement was substantial for optic atrophy (k 5 0.73) and moderate for papilledema (k 5 0.39) and for either form of optic neuropathy (k 5 0.54). Logistic regression indicated that peripapillary RNFL thickness was predictive of optic neuropathy (P \ 0.001). Multivariable analysis demonstrated that RNFL thickness measurements were more sensitive at detecting optic neuropathy than visual field testing (likelihood ratio 5 10.02; P 5 0.002). Sensitivity and specificity of logMAR visual acuity in detecting optic neuropathy were 15% and 95%, respectively.
CONCLUSIONS
Peripapillary RNFL thickness measured by SD-OCT provides adjunctive evidence for identifying optic neuropathy in patients with craniosynostosis and appears more sensitive at detecting optic atrophy than papilledema. ( J AAPOS 2014;18:543-549)
C
hildren with craniosynostosis (CS) suffer premature fusion of one or more cranial sutures in utero and/or during the first years of life. Early fusion can result in increased intracranial pressure (ICP) prior to or after cranial vault expansion.1,2 Persistent elevation
Author affiliations: Departments of aOphthalmology, bRadiology, cAnesthesia, and dSurgery, Boston Children’s Hospital; Harvard Medical School, Boston, Massachusetts Financial support: Endowment funds from the Children’s Hospital Ophthalmology Foundation Chair and the Discovery Award, both granted to Linda R. Dagi, MD. Presented at the 40th Annual Meeting of the American Association for Pediatric Ophthalmology and Strabismus, Palm Springs, California, April 2-6, 2014. Submitted May 22, 2014. Revision accepted July 29, 2014. Correspondence: Linda R. Dagi, MD, Department of Ophthalmology, Boston Children’s Hospital, 300 Longwood Avenues, Fegan 4, Boston, MA 02115 (email: Linda.Dagi@ childrens.harvard.edu). Copyright Ó 2014 by the American Association for Pediatric Ophthalmology and Strabismus. 1091-8531/$36.00 http://dx.doi.org/10.1016/j.jaapos.2014.07.177
Journal of AAPOS
in ICP may be related to continued arrest of cranial growth, obstructive hydrocephalus, or complications of sleep apnea.1-5 Monitoring this vulnerable population for impending optic neuropathy is challenging due to competing etiologies of vision loss (astigmatism, amblyopia, strabismus, and exposure keratopathy) and limited cooperation due to age, cognitive skills, or photophobia.6-8 Pattern-reversal visual evoked potentials can aid in evaluation, but pediatric normative data is not universally available, testing is time consuming, and interpretation requires highly skilled support staff. Frequent direct measurement of ICP by lumbar puncture or pressure bolt augments risk and is invasive.9 The purpose of this study was to assess the ability of peripapillary retinal nerve fiber layer (RNFL) thickness measured by spectral-domain optical coherence tomography (SD-OCT) to detect optic neuropathy compared to fundus examination in patients with CS. Additionally, we compared indication of optic neuropathy by RNFL
543
544
Dagi et al
thickness and fundus examination with evidence of altered visual function measured by logMAR visual acuity and visual field testing, and we investigated the correlation of abnormal fundus examination with historical evidence of elevated ICP.
Subjects and Methods This study was approved by the Boston Children’s Hospital Institutional Review Board and conformed to the requirements of the US Health Insurance Portability and Accountability Act of 1996. The medical records of patients $2 years of age with CS of any type presenting at Boston Children’s Hospital for ophthalmic evaluation during 2013 were retrospectively reviewed. Patients with a history of premature birth, anterior segment anomaly, glaucoma, retinal degeneration, or dystrophy were excluded.10,11 Age, sex, CS diagnosis, and history of elevation in ICP based on radiographic confirmation or copper beaten skull deformity documented at the time of cranial vault expansion were recorded. Current best-corrected logMAR visual acuity in the bettersighted (nonamblyopic) eye, cycloplegic refraction, color vision by Hardy Rand Rittler (HRR) or Ishihara color plates, visual fields obtained by Humphrey Field Analyzer (Zeiss 750i-series Humphrey Field Analyzer II, Dublin, CA) or Goldmann kinetic perimetry, optic nerve appearance, and peripapillary RNFL thickness measured by Heidelberg Spectralis SD-OCT (Heidelberg Engineering, GmbH, Heidelberg, Germany) were also documented. Patients were classified as moderately hyperopic if spherical equivalent (SE) was between 13.25 D and 16.00 D and highly hyperopic if between 16.25 D and 19.50 D.12 Classification of excess myopia required SE . 4.00 D.13
Visual Fields Visual fields undertaken with Goldmann kinetic perimetry were performed on patients too young or cognitively limited to perform Humphrey field testing and analyzed using a modified Wall and George grading system.14 The Humphrey SITA fast central 24-2 test was chosen for older or more cooperative patients. These fields were graded using guidelines described by Thomas and George.15
Spectral-domain OCT Peripapillary RNFL thickness measurements were obtained by two experienced technicians using the N-site Analytics scan protocol (Heidelberg Engineering, GmbH, Heidelberg, Germany). The Spectralis captures details at a speed of 40,000 A-scans per second, with axial resolution of 3.9 mm and a transverse resolution of 14 mm.16 Automated real-time tracking eliminates artifacts from eye movements in order to generate high-resolution scans. The N-site axonal protocol involves a circle B-scan of approximately 3.45 mm centered around the optic nerve.16 Measurements initiate and end nasally, allowing for detailed assessment of the papillomacular bundle.17 Scans with artifacts, misalignment errors, or quality scores below 23 were rejected.
Volume 18 Number 6 / December 2014 Historical Evidence of Elevated ICP Radiological findings were assessed by an experienced pediatric neuroradiologist who was masked to ophthalmic findings. Historical CT and/or MR scans of the brain were reviewed. Presence of copper-beaten pattern, progressive ventriculomegaly, shunted hydrocephalus, or jugular foraminal stenosis (JFS) with enlarged occipitomastoid emissary veins (LEV)18,19 was deemed sufficient to confirm previous elevation in ICP. Additionally, patients with copper-beaten skull deformity documented at cranial vault expansion were included in the group of patients with historical evidence of elevated ICP.
Designation of Optic Neuropathy A pediatric ophthalmologist with 25 years’ experience determined the presence of optic neuropathy (papilledema or optic atrophy) based on fundus examination and photographic confirmation utilizing the Non-Mydriatric Retinal Camera (Topcon Corp, Model TRC-NW8F). Optic atrophy was defined as clinically and photographically apparent optic nerve pallor or hypoplasia, sectorally or globally, with diminution of vascularity and confirmatory nerve fiber layer drop-out on red-free fundus view. Papilledema was designated in the presence of clinically and photographically apparent optic nerve swelling, with at least some obscuration of vasculature at the disk border. Current optic nerve designation (normal, atrophic, or edematous) was compared with current peripapillary RNFL thickness. Correlation of current optic nerve appearance with logMAR acuity, visual field results, and historical evidence of elevation in ICP were also investigated.
Statistical Analysis One eye per patient was included for analysis. In patients with amblyopia, the nonamblyopic eye was chosen. If neither eye was amblyopic, the eye with the higher quality SD-OCT RNFL scan was included. Peripapillary RNFL thickness was compared to published pediatric norms for patients under age 1813,20 and to validated adult norms for older patients. Pediatric normative data was derived from published values13,20 (N 5 83; N 5 107) and these combined to yield an N of 190 patients. Standard error mean13 was used to calculate standard deviation for each RNFL region—nasal, nasal inferior, temporal inferior, temporal, temporal superior, nasal superior, and global (SD 5 SEM * SQRT[sample size]). Mean thicknesses from the studies were combined for each RNFL region (Combined Mean 5 [N1 * Mean1 1 N2 * Mean2]/[N11N2]), as were the standard deviations (Combined SD 5 SQRT [(SD1)2 1 (SD2)2]). From the combined data, we determined the 1st, 5th, 95th, and 99th percentiles for each RNFL region (P1 5 mean [2.33 * SD]; P5 5 mean [1.64 * SD]; P95 5 mean 1 [1.64 * SD]; P99 5 mean 1 [2.33 * SD]). See Table 1 for combined percentile data. For each pediatric patient, a thickness (mm) versus position ( ) graph beginning ( 180 ) and ending (180 ) nasally was created using curved lines to compare patient data to combined pediatric 1st, 5th, 95th, and 99th percentiles. This visual representation appears similar to adult Heidelberg SD-OCT peripapillary RNFL thickness output graph. An example is provided in Figure 1.
Journal of AAPOS
Volume 18 Number 6 / December 2014
Dagi et al
545
Table 1. Combined pediatric normative data, derived from Turk et al20 and Yanni et al13 Region Nasal Nasal inferior Temporal inferior Temporal Temporal superior Nasal superior
Position, degrees 180; 180 135 45 0 45 135
1st Percentile
5th Percentile
95th Percentile
99th Percentile
30.6 37.0 85.8 29.3 79.5 38.6
44.4 60.0 103.5 42.9 97.9 59.3
110.0 169.4 187.8 107.6 185.4 158.0
123.8 192.4 205.6 121.2 203.8 178.8
Sensitivity and specificity were calculated with 95% confidence intervals to evaluate the diagnostic characteristics of SD-OCT RNFL thickness, logMAR acuity, visual field assessment, and history of elevated ICP compared to current fundus appearance.21 Cohen’s kappa coefficient (k) was used as the chance-corrected measure of agreement with fundus examination. Standard criteria on the strength of agreement is from Landis and Koch.22 Positive and negative predictive values of SD-OCT were computed using Bayes’ theorem and a 25% estimated prevalence of optic neuropathy from fundus examination. Statistical power was calculated using nQuery Advisor, version 7.0 (Statistical Solutions, Saugus, MA). To ascertain whether SD-OCT provides additional predictive information, multivariable logistic regression was applied in the subgroup that had both SD-OCT and visual field results (n 5 36), using the likelihood ratio test as the criteria for statistical significance.23,24 Point-biserial correlation coefficient (rpb) comparing logMAR acuity to optic neuropathy as identified by fundus examination was performed. Statistical analysis was completed using SPSS version 21.0 (IBM, Armonk, NY). Two-tailed P values of \0.05 were considered statistically significant.
Results A total of 65 patients were studied; 11 were excluded because of limited cooperation, severe nystagmus, poor scan quality, or retinal degeneration. Mean age at examination was 10.6 5.4 years (median, 9.3 4.6 years; range, 2.4-33.8 years). The cohort consisted of 33 females (61%) and 21 males (39%) and included a wide range of diagnoses (Apert, Pfeiffer, Crouzon, Saethre-Chotzen, isolated metopic synostosis, Muenke, unicoronal, bicoronal, and atypical multisuture variants). Abstracted clinical characteristics are summarized in Table 2. No patient displayed a relative afferent pupillary defect. All patients had a best-corrected visual acuity in their nonamblyopic or included eye of 0.48 logMAR or better (mean, 0.03). Of the 54 included patients, 47 patients had normal visual acuity ( 0.12 to 0.10 logMAR), 5 had a visual acuity of 0.18 to 0.30, and 2 patients had a visual acuity of 0.40 to 0.48. Point-biserial correction coefficient comparing logMAR acuity to optic neuropathy, as identified by fundus exam, demonstrated a moderate positive correlation (rpb 5 0.329, P 5 0.015), confirming that visual acuities were worse among some patients with evidence of optic neuropathy based on fundus examination.
Journal of AAPOS
FIG 1. Pediatric peripapillary RNFL thickness, fundus examination, and Goldmann perimetry. Graphs were created using combined pediatric normative data from Turk and colleauges20 and Yanni and colleagues.13 Patient 1 has papilledema noted by (A) peripapillary RNFL, (B) fundus examination, and (C) visual fields. D-F demonstrate these features for patient 2 with optic atrophy. BA, borderline above; BB, borderline below.
546
Dagi et al
Volume 18 Number 6 / December 2014
Table 2. Patient characteristics Sex, age, and diagnosis Male Female Median age at exam, years Diagnosis Apert Bicoronal Crouzon Metopic Pfeiffer Othera Saethre-Chozen Sagittal Unilateral coronal
Radiographic and clinical features 21 (39%) 33 (61%) 9.3 (4.6) 4 4 4 4 3 10 7 10 8
JFS and LEVb JFS without LEVb Hydrocephalus/Ventriculomegalyb Copper-beaten patternb Chiari malformationb History indicating elevated ICPb Amblyopia Mean best-corrected visual acuity (logMAR) Mean spherical equivalent Color abnormalities Field deficitsc Papilledema on fundus exam Optic atrophy on fundus exam
21 (44%) 4 (0.8%) 15 (31%) 14 (29%) 8 (17%) 30 (63%) 13 (24%) 0.03 0.82 3 (6%) 11 (31%) 5 (9%) 8 (15%)
JFS, jugular foraminal stenosis; LEV, emissary veins. a Includes Apert/Pfeiffer-like, Crouzoid, Crouzon-like, Muenke, and multiple suture. b Based on imaging and intraoperative findings available for 48 patients. c Based on data base of 36 reliable visual fields.
Sensitivity was 15% and specificity was 95% for the association of logMAR acuity $0.3 with optic neuropathy. Two patients had mild color vision defects on HRR testing, missing two test plates, and 1 patient had classic protanomaly/deuteranomaly dyschromatopsia. Neither had abnormal fundus appearance, RNFL thickness, or history of elevated ICP. Spectral-domain OCT Analyzed scans had a mean quality score of 31.3 5.1 (range, 23-43). All but 3 were obtained on the day of fundus examination. Those not obtained that day were completed within 3 weeks, without intervening intervention. There was evidence of optic neuropathy on fundus examination in 13 patients (8 with optic atrophy; 5 with papilledema). Of these, 10 with optic neuropathy had thinner-than-normative range RNFL and optic atrophy and thicker-than normative range with papilledema. SD-OCT sensitivity was 88% for classifying patients with optic atrophy; 60% for those with papilledema; and 77% for those with either form of optic neuropathy. One patient with disk edema and another with optic atrophy had RNFL thickness measurements within the normal range. Peripapillary RNFL thickness outside the pediatric normative range (\18 years of age) and outside the standard adult normative range for all others was seen in 7 of 41 patients (17%) with a normal fundus appearance. SD-OCT specificity was 94% in classifying patients without evidence of optic atrophy; 90% for those without papilledema; and 83% for those without either. Kappa agreement between SD-OCT results and fundus appearance was moderate to substantial: k 5 0.73 for optic atrophy; k 5 0.39 for papilledema; and k 5 0.54 for either form of optic neuropathy (Table 3). Logistic regression analysis confirmed that SDOCT was predictive of optic neuropathy findings on fundus
examination (OR: 16.2; 95% CI, 3.5-74.4; likelihood ratio test 5 15.75; P \ 0.001). Among the patients classified as abnormal on SD-OCT, 59% would be expected to have optic neuropathy (positive predictive value). Among patients with a normal SD-OCT test, 92% would be expected to have a normal fundus examination (negative predictive value). nQuery Advisor software demonstrated that our sample size of 54 patients provides 80% power to assess the predictive value of RNFL thickness abnormality as a predictor of optic neuropathy. Patients with apparent optic atrophy on SD-OCT (n 5 11) exhibited it more temporally (90.0%) than nasally (50.0%), but there was a more even distribution temporally (62.5%) and nasally (87.5%) in patients with papilledema on SD-OCT (n 5 8). Of the 4 patients with moderate and 1 with high hyperopia, 3 had normal RNFL thickness measurements, 1 had clinical and SD-OCT evidence of optic atrophy (14.25 SE), and 1 demonstrated a measured increase in RNFL thickness (15.25 SE), without other ophthalmic features or history of elevation in ICP. Considering SD-OCT as a possible predictor of papilledema on fundus examination and controlling for moderate to high hyperopia as a possible confounding variable, we found that SD-OCT was a significant predictor of papilledema irrespective of the presence of moderate to high hyperopia (OR: 22.0; 95% CI, 3.0-159.7; likelihood ratio test 5 10.13; P \ 0.001). Among the 10 patients who presented with myopia, only 1 had higher myopia ( 5.25 SE), and she had frank papilledema and an associated increase in RNFL thickness. There were 49 pediatric patients, 14 (29%) of whom had abnormal examinations based on pediatric and adult normative data. Twenty-one of 35 pediatric patients (60%) had unremarkable SD-OCT results based on pediatric normative data but would have been considered abnormal based on adult norms standardly provided at this time.
Journal of AAPOS
Volume 18 Number 6 / December 2014
Dagi et al
547
Table 3. OCT, history, visual fields, and LogMAR visual acuity as predictors of optic neuropathya Variable
Sensitivity
95% CI
Specificity
95% CI
Kappa
OCT: optic atrophy OCT: papilledema OCT: optic neuropathy Historical evidence [ICPb Visual fieldc LogMAR VA
7/8 (88%) 3/5 (60%) 10/13 (77%) 11/13 (85%)
53%-98% 23%-88% 50%-92% 58%-96%
43/46 (94%) 44/49 (90%) 34/41 (83%) 16/35 (46%)
83%-98% 78%-96% 69%-91% 30%-62%
0.73 0.39 0.54 0.22
6/10 (60%) 2/13 (15%)
32%-83% 4%-42%
21/26 (81%) 39/41 (95%)
62%-91% 84%-99%
0.40 0.14
CI, confidence interval; OCT, optical coherence tomography; VA, visual acuity. a Diagnostic characteristics based on fundus appearance as the gold standard. b Historical evidence based on radiographic features and intraoperative findings consistent with elevated ICP based on 48 patients with radiology available for review. c Based on 36 patients with visual field data and fundus examination data.
Visual Fields A subset of 36 patients had complete and reliable visual fields. In comparing agreement of visual field findings with fundus appearance, sensitivity was 60% and specificity was 81%, with moderate kappa agreement (k 5 0.40). Eleven patients had a visual field abnormality, 8 with abnormal peripapillary RNFL thickness. Visual field testing was normal for 25 patients, but 6 had abnormal RNFL thickness. Therefore, agreement between SD-OCT and visual field modalities was moderate (75% agreement; k 5 0.45). Multivariable logistic regression indicated that, independent of visual field results, information obtained by SD-OCT provides significant prediction of what is seen by fundus examination regarding the presence of optic neuropathy (OR: 13.3; 95% CI, 2.2-80.5; likelihood ratio test 5 10.02; P 5 0.002). The presence of an abnormal visual field along with optic neuropathy on fundus examination guaranteed predicted SD-OCT abnormalities. History of Elevated ICP No neuroimaging was available for 6 of 54 patients. For the rest, imaging studies consisted of both CT and MR of the brain (n 5 19), CT alone (n 5 25), or MR alone (n 5 4). Imaging characteristics are summarized in Table 2 and detailed in e-Supplement 1 (available at jaapos.org). In 30 patients, radiographic or intraoperative features confirmed prior elevation in ICP. Table 3 compares RNFL thickness by SD-OCT with historical evidence of elevation in ICP, visual field results, and logMAR acuity testing as predictors of current optic neuropathy based on fundus examination. Figure 2 illustrates this graphically with likelihood ratios.
Discussion Peripapillary RNFL thickness measured by SD-OCT confirmed fundus examination findings of optic neuropathy with a sensitivity of 77% and a specificity of 83%. However, sensitivity for detecting optic atrophy was much higher (88%) than for detecting papilledema (60%). Positive predictive value (59%) was found to be
Journal of AAPOS
FIG 2. Likelihood ratio of SD-OCT, history, visual fields, and logMAR acuity predicting optic neuropathy. The likelihood ratio reflects the likelihood of optic neuropathy given positive test results. If the SD-OCT result is positive for optic neuropathy, the likelihood that the patient has optic neuropathy is nearly 5 times greater than if it is negative.
more modest than negative predictive value (92%). Lower positive predictive value is anticipated when disease incidence is low—only 13 of our 54 patients had optic neuropathy. Having abnormal visual fields, along with evidence of optic neuropathy on fundus examination, guaranteed predicted RNFL findings. Whereas a “gold standard” combining altered visual function (visual acuity and visual field abnormality) with abnormal fundus examination would have resulted in an absolute correlation between the presence of optic neuropathy on clinical examination and abnormal peripapillary RNFL measurements, we did not design our study with this comprehensive approach. Measured visual acuity may not be accurate in this population and would be unlikely to show decline with mild papilledema. Furthermore, accurate visual field testing is often unavailable in those with limited cognitive function. We found that only two-thirds of patients had
548
Dagi et al
Volume 18 Number 6 / December 2014
FIG 3. Comparison of information provided by fundus examination, SD-OCT, and visual fields for a patient presenting with and treated for papilledema. A, Initial presentation. B, Two months after treatment with acetazolamide there was minimal change on fundus examination but clear reduction in peripapillary RNFL thickness and improved visual field. C, Five months after initiating treatment noted decline in papilledema on fundus examination and SD-OCT but visual field was unchanged.
reliable visual field data. To assess the correlation of function with optic nerve appearance, we chose logMAR visual acuity worse than 0.3 as criteria for evidence of visual acuity loss. The 2 patients with poorest logMAR acuity (0.48) had evidence of optic atrophy. Nevertheless, only a modest correlation was confirmed between diminished visual acuity and optic neuropathy: only 4 patients had “diminished” acuity by our standard. We suspect that early referral and attentive management protected most from significant visual acuity loss. Although there is no consensus about the impact of amblyopia on peripapillary RNFL thickness, amblyopia has been reported to reduce it.25,26 Since we sought evidence of vision loss from optic neuropathy, and not from amblyopia, we studied acuity and RNFL thickness characteristics from the nonamblyopic eye. Abnormal peripapillary RNFL thickness was documented in 7 of 41 patients with normal fundus appearance. The 5 with elevated thickness had the following characteristics: history of elevation in ICP (3), chronic sleep apnea (1), moderate hyperopia (1), and/or optic nerve head drusen documented with echography (1). Excessive hyperopia and optic nerve head drusen are known confounders, increasing RNFL thickness.14,27,28 The 2 patients with reduced thickness had either a history of elevated ICP or microcephaly. Perhaps SD-OCT identified some patients with optic neuropathy that were missed on clinical examination. Myopia is also a confounder, associated with the
decreased RNFL thickness.29 Our patient with 5.25 D of myopia had frank papilledema, thus overwhelming this potential confounder. Longitudinal monitoring of peripapillary RNFL measurements in this population may provide added value. The patient described in Figure 3 presented with papilledema and was treated with acetazolamide in anticipation of shunt placement. RNFL thickness measurements taken over several months demonstrated a more definitive change after 2 months of treatment than did fundus appearance. We suspect that SD-OCT monitoring will prove even more important for patients with otherwise occult increase in optic atrophy over time. This study has several limitations. A combination of fundus appearance and functional loss was contemplated as a gold standard for optic neuropathy but proved impractical, given considerations noted above. Assessment with patternreversal visual evoked potentials might have enriched our analysis. Axial length, known to affect peripapillary RNFL thickness,30 was not collected, but significant hyperopia and myopia were used as reasonable surrogates. Peripapillary total RNFL thickness31,32 or volume33 may be more sensitive at detecting papilledema compared to standard RNFL thickness but were not obtained. Fundus appraisal was made by one observer. Lastly, the classification of patients into those with evidence of elevated ICP was subject to historical and incomplete data including imaging studies which might have predated the onset of elevated ICP.
Journal of AAPOS
Volume 18 Number 6 / December 2014
References 1. Liasis A, Thompson DA, Hayward R, Nischal KK. Sustained raised intracranial pressure implicated only by pattern reversal visual evoked potentials after cranial vault expansion surgery. Pediatr Neurosurg 2003;39:75-80. 2. Campbell JW, Albright AL, Losken HW, Biglan AW. Intracranial hypertension after cranial vault decompression for craniosynostosis. Pediatr Neurosurg 1995;22:270-73. 3. Fok H, Jones BM, Gault DG, Andar U, Hayward R. Relationship between intracranial pressure and intracranial volume in craniosynostosis. Br J Plast Surg 1992;45:394-7. 4. Liasis A, Nischal KK, Walters B, et al. Monitoring visual function in children with syndromic craniosynostosis: a comparison of 3 methods. Arch Ophthalmol 2006;124:1119-26. 5. Tuite GF, Chong WK, Evanson J, et al. The effectiveness of papilledema as an indicator of raised intracranial pressure in children with craniosynostosis. Neurosurgery 1996;38:272-8. 6. Khan SH, Nischal KK, Dean F, Hayward RD, Walker J. Visual outcomes and amblyogenic risk factors in craniosynostotic syndromes: a review of 141 cases. Br J Ophthalmol 2003;87:999-1003. 7. Nischal KK. Ocular aspects of craniofacial disorders. Am Orthopt J 2002;52:58-68. 8. McGraw PV, Winn B, Gray LS, Elliott DB. Improving the reliability of visual acuity measures in young children. Ophthalmic Physiol Opt 2000;20:173-84. 9. Rangwala LM, Liu GT. Pediatric idiopathic intracranial hypertension. Surv Ophthalmol 2007;52:597-617. 10. Goldberg MF, Duke JR. Ocular histopathology in Hunter’s syndrome: systemic mucopolysaccharidosis type II. Arch Ophthalmol 1967;77:503-12. 11. Liang F, Audo I, Sahel JA, Paques M. Retinal degeneration in mucopolysaccharidose type II. Graefes Arch Clin Exp Ophthalmol 2013; 251:1871-2. Epub December 4, 2012. 12. Tas M, Oner V, Turkcu FM, et al. Peripapillary retinal nerve fiber layer thickness in hyperopic children. Optom Vis Sci 2012;89:1009-13. 13. Yanni SE, Wang J, Cheng CS, et al. Normative reference ranges for the retinal nerve fiber layer, macula, and retinal layer thicknesses in children. Am J Ophthalmol 2013;155:354-360.e1. 14. Wall M, George GE. Idiopathic intracranial hypertension a prospective study of 50 patients. Brain 1991;114(Pt 1A):115-80. 15. Thomas R, George R. Interpreting automated perimetry. Indian J Ophthalmol 2001;49:125-40. 16. Spectralis HRA1OCT User Guide Software Version 5.3. Heidelberg, Germany: Heidelberg Engineering GmbH; 2010. 17. Rebolleda G, Gonzalez-Lopez JJ, Munoz-Negrete FJ, Oblanca N, Costa-Frossard L, Alvarez Cermeno JC. Color-code agreement among stratus, cirrus, and spectralis optical coherence tomography in relapsing-remitting multiple sclerosis with and without prior optic neuritis. Am J Ophthalmol 2013;155:890-97. Epub February 8, 2013.
Journal of AAPOS
Dagi et al
549
18. Robson CD, Mulliken JB, Robertson RL, et al. Prominent basal emissary foramina in syndromic craniosynostosis: correlation with phenotypic and molecular diagnoses. Am J Neuroradiol 2000;21:1707-17. 19. Taylor WJ, Hayward RD, Lasjaunias P, et al. Enigma of raised intracranial pressure in patients with complex craniosynostosis: the role of abnormal intracranial venous drainage. J Neurosurg 2001;94:377-85. 20. Turk A, Ceylan OM, Arici C, et al. Evaluation of the nerve fiber layer and macula in the eyes of healthy children using spectral-domain optical coherence tomography. Am J Ophthalmol 2012;153:552-559.e1. 21. Fletcher RW, Fletcher SW. Diagnosis. In: Clinical Epidemiology: The Essentials. 4th ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2005:35-58. 22. Landis RJ, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159-74. 23. Katz MH. Interpreting the Results. In: Multivariable Analysis: A Practical Guide for Clinicians and Public Health Researchers. 3rd ed. New York, NY: Cambridge University Press; 2011:140-61. 24. Hidalgo B, Goodman M. Multivariate or Multivariable Regression? Am J Public Health 2013;103:39-40. 25. Bandyopadhyay SK, Chatterjee A, Banerjee R. Peripapillary nerve fibre layer thickness and macular thickness in children with anisometropic amblyopia attending a referral centre in Eastern India. J Indian Med Assoc 2012;110:542-5. 26. Colen TP, de Faber JT, Lemij HG. Retinal nerve fiber layer thickness in human strabismic amblyopia. Bincol Vis Strabismus Q 2000;15: 141-6. 27. Sarac O, Tasci YY, Gurdal C, Can I. Differentiation of optic disc edema from optic nerve head drusen with spectral-domain optical coherence tomography. J Neuroophthalmol 2012;32:207-11. 28. Lee KM, Woo SJ, Hwang J-M. Differentiation of optic nerve head drusen and optic disc edema with spectral-domain optical coherence tomography. Ophthalmology 2011;118:971-7. 29. Choi SW, Lee SJ. Thickness changes in the fovea and peripapillary retinal nerve fiber layer depend on the degree of myopia. Korean J Ophthalmol 2006;20:215-19. 30. Oner V, Aykut V, Tas M, Alakus MF, Iscan Y. Effect of refractive status on peripapillary retinal nerve fibre layer thickness: a study by RTVue spectral domain optical coherence tomography. Br J Ophthalmol 2013;97:75-9. 31. Vartin VC, Nguyen AM, Balmitgere T, Bernard M, Tilikete C, Vighetto A. Detection of mild papilloedema using special domain optical coherence tomography. Br J Ophthalmol 2012;96:375-9. 32. Scott CJ, Kardon RH, Lee AG, Frisen L, Wall M. Diagnosis and grading of papilledema in patients with raised intracranial pressure using optical coherence tomography vs clinical expert assessment using a clinical staging scale. Arch Ophthalmol 2010;128:705-11. 33. Wang J-K, Kardon RH, Kupersmith MJ, Garvin MK. Automated quantification of volumetric optic disc swelling in papilledema using spectral-domain optical coherence tomography. Invest Ophthalmol Vis Sci 2012;53:4069-75.
