Spectral domain optical coherence tomography in children: normative data and biometric correlations

Optical Coherence Tomography (OCT) is a noninvasive, noncontact, transpupillary imaging method that performs objective high-resolution cross-sectional images of retinal tissue. The recently introduced spectral domain OCT (SD-OCT) provides measurements of the retinal nerve fiber layer (RNFL) and macula with greatly improved image acquisition speed and image resolution up to 5 μm [1]. This is particularly helpful when applying this technology in uncooperative children.

Several studies have proved the feasibility of OCT in the pediatric population [2‐9]. Nevertheless, all OCT devices have an integrated normative database only for adult subjects 18 years of age and older. Earlier studies have reported normative values in children using the time domain OCT (TD-OCT) devices [3‐11]; similar reports using SD-OCT are much less available. Although RNFL measurements taken from TD-OCT and SD-OCT are comparable, significant differences exist and values cannot be used interchangeably [12, 13]. Comparing macular thickness between the two instruments is even more complicated, being dependant on pathology and location [14]. Measurement protocols vary even among different SD-OCT instruments [15]. Only a few studies in the literature aimed at reporting normative reference ranges using SD-OCT [16‐21]. Normative measurements in children using Cirrus are least reported, especially with regards to macular parameters.

The clinical applications of SD-OCT are increasingly expanding [22]; normal reference values for RNFL and macular thickness are needed in the pediatric population where the software has no nomogram for comparison. The purpose of this study was to collect normative values for SD-OCT measurements of macular thickness and volume and peripapillary RNFL thickness in healthy eyes of normal children using the most recent commercially available Cirrus SD-OCT, and to study the effects of age, gender, axial length, and refractive error on these values.

A total of 113 children and adolescents were enrolled in the study, five subjects were later excluded due to decentered OCT scans. We included 108 patients (mean age 10.7+/−3.1 years), 43 males (mean age 10.7+/−3.4 years) and 65 females (mean age 10.7+/−3.0 years). Data from right eyes of all patients were recorded for analysis. Average spherical equivalent (SE) refraction was −0.02+/−1.77(−4.25 to +5.00) diopters and average axial length was 23.5+/−1.0 (21.5 to 25.8) mm.

Mean RNFL and macular thicknesses followed a normal distribution (Figure 1). Patients were divided into four subgroups based on age: 6 up to 9 years of age (n = 35), 9 up to12 years (n = 28), 12 up to15 years (n = 28) and 15 up to18 years (n = 17). Mean RNFL thickness for all patients was 95.6+/−8.7 μm; quadrant thicknesses followed a pattern similar to that in adults with the thickest being the inferior followed by superior then nasal then temporal (Table 1).

Mean central macular thickness for all patients measured 249.1+/−20.2 μm, while average macular thickness value was 279.6+/−12.5 μm, and mean macular volume was 10.1+/−0.5 mm³ (Table 2). Inner circle values were significantly increased as compared to the outer macular thicknesses (p < 0.0001). There was a significant positive correlation between average RNFL and average macular thickness (r = 0.30, p = 0.002).

As expected, in the correlation analysis, a significant negative correlation was seen between age and SE(r = −0.33, p = 0.001) and between SE and AL(r = −0.70, p < 0.001). A significant positive correlation was observed between age and AL (r = 0.30, p = 0.02).

On multivariate analysis, RNFL measurements were not affected by gender whereas central macular thickness values were significantly increased in males (p < 0.001). RNFL measurements did not correlate with age but did correlate positively with SE. When controlling for axial length, the correlation with SE remained significant for average RNFL thickness but not for quadrant thicknesses. All macular parameters were consistently positively correlated with age and most were positively correlated with SE (Table 3).When controlling for axial length, correlation with SE was lost and only the positive correlation of the inner circle macular thicknesses with age persisted.

This study reported normative values for RNFL thickness, macular thickness and macular volume in children 6–17 years of age using spectral domain OCT (Cirrus). Correlations with biometric data showed that central macular but not RNFL thickness values were higher in males; additionally, the macular inner circle thickness correlated positively with age and average RNFL thickness correlated positively with SE. Optical coherence tomography, being non-invasive and especially fast, is gaining more popularity in evaluating vitreoretinal diseases of childhood including pediatric glaucoma [20, 22]. Spectral domain -OCT, the latest generation of the technology, provides higher resolution and decreased acquisition time, hence is more useful in the pediatric population. Direct comparison of RNFL and macular parameters between TD-OCT and SD-OCT is not possible due to different scanning algorithms [12‐14]. Studies using the earlier TD-OCT in children have shown good reproducibility [4, 23]. Altemir et al. showed good reliability and repeatability in children using Cirrus SD-OCT [2]. With Stratus OCT, good quality scans could be obtained in 93% and 96% of children [1, 3]. Using Spectralis SD-OCT, Turk et al. reported higher feasibility of 99% [16]. We similarly report high quality Cirrus OCT scans in 96% of our enrolled children.

Most available literature reporting normative OCT values in children used TD-OCT [5‐11]. Recently, normal values have been reported using Spectralis in Turkish and North American children [16, 17] and RTVue-100 OCT in Chinese children [18]. Using Cirrus, Elia et al. reported RNFL but not macular parameters in Caucasian children aged 6–13 years; however, ophthalmologic examination was performed without cycloplegia and OCT parameters were not correlated with biometric data [19]. Our work compliments the latter study by: confirming RNFL normative values on Cirrus, expanding the age group to include 14–17 year-old adolescents, reporting additional normative values for macular thicknesses and volume, and correlating RNFL and macular parameters with age, gender, refractive error and axial length. At the time of our data analysis, we noted a recent report of similar work using Cirrus in another study from Spain; Barrio-Barrio et al. reported RNFL and macular measurements in Caucasian children from three Spanish centers [21]. While Elia and Barrio-Barrio reported normative values in the Spanish population, the current study reported normative Cirrus OCT values in children from the Middle East. The study by Barrio-Barrio tried to address similar objectives as the current study; differences in design include the multicenter nature of their study and the different age group (4–17 years in their study, 6–17 years in our study). Reported normative data in their study were very close to our measurements (Tables 4 & 5), correlations were also in agreement: positive correlation of RNFL with SE and central macular thickness with age and male gender. They similarly excluded subjects with high refractive errors (>5.5 diopters of SE).

Tables 4 and 5 compare our OCT values with earlier literature. Average RNFL thickness in this study measured 96 μm comparable to others using Cirrus (97-98 μm) but lower than values recorded using Stratus (104-113 μm), Spectralis (106-107 μm) and RTVue (109 μm). The quadrant thickness distribution followed the classic “double hump” pattern seen in adults with the thickest being the inferior followed by the superior then nasal and temporal, consistent with the “ISNT rule”. However, a number of reports have shown exceptions to this rule [3, 5, 7, 9, 18].

Macular parameters show the greatest variability among different OCT devices and algorithms [14]; discrepancies have also been noted across different versions of OCT [15, 24]. Few studies have reported macular parameters in normal children (Table 5). Mean macular volume in our study was 10.1 mm³ similar to other studies using Cirrus but much higher than those obtained by Stratus (7 mm³) or Spectralis (0.25 mm³); this also applied to central macular thickness values. Average macular thickness was highest on Spectralis (326 μm), followed by Cirrus (280-284 μm) and Stratus (240-270 μm). Yanni et al. reported mean macular thickness (271 μm) on Spectralis but proceeded to segment the retinal layers and perform normative segmentation values, hence their data were not included in Table 5[17]. In the current study, inner macular thicknesses were statistically significantly higher than outer macular values in all quadrants similar to results from Stratus [4, 5].

Discrepancies noted in recorded normative OCT values could also be related to confounding variables like ethnicity, race, gender, age, SE and AL measurements. All our subjects were white and Middle Eastern while the other two reports using Cirrus came from a Spanish population. Gender differences applied only in central macular thickness measurements which were significantly increased in males. This finding is in agreement with Barrio-Barrio [21] and Huynh et al. [8]. Gender differences may need to be accounted for during OCT interpretations.

In this study, RNFL values were not affected by age in agreement with other reports [5, 9, 16, 21]. Some authors suggested that nerve fiber layer losses happened later in life after the age of 50 years, hence the absence of RNFL correlation with age in children [25]. By contrast, we found a strong positive correlation between age and macular parameters, similar to the findings by Barrio-Barrio et al. [21] and Huynh et al. [8].

High refractive errors may affect OCT measurements; therefore, we excluded patients with SE more than 7 diopters. Still we noted a strong persistent positive correlation between SE and average RNFL thickness measurements after controlling for axial length. Many authors have observed positive correlations with SE [3, 8, 11, 18]. Tsai et al. reported an increase in average RNFL thickness of 1.7 μm for every diopter of hyperopia [18]. On the other hand, Turk et al. did not detect any correlations with SE or axial length [16].

Strengths of this study include the large age range (6–17 years) of enrolled children, the use of the new generation Cirrus SD-OCT, the completion of a comprehensive eye examination (with cycloplegic retinoscopy) by a pediatric ophthalmologist, the recording of both normative RNFL and macular parameters, and the detailed biometric correlations. Limitations of this work include the mostly uniform ethnic group (white and Middle Eastern) so the effect of race and ethnicity could not be tested. We also excluded patients with high refractive errors and increased cup to disc ratios; normative data for these groups were not established. Additionally, our study was hospital-based and not population-based. However, patients in this setting received a comprehensive examination, and biometric data were recorded.

This study established normal reference ranges for RNFL and macular parameters measured by Cirrus SD-OCT in healthy Middle Eastern children 6–17 years of age. This adds another database to the available literature on normative values using other OCT devices and facilitates evaluation of OCT measurements in children with optic neuropathies, glaucoma and macular diseases. The data presented are for white Middle Eastern children; hence, other races and ethnicities should be studied in future research. Variability with age and gender warrants special consideration during OCT interpretations.