Background/aims To characterise the association of iris surface features (crypts, furrows and colour) with iris volume and curvature assessed by swept-source optical coherence tomography (SSOCT) in Asian eyes.
Methods Iris crypts (by number and size) and furrows (by number and circumferential extent) were graded from iris photographs. Iris colour was measured by a customised algorithm written on MATLAB (MathWorks, Natick, Massachusetts, USA). The iris was imaged by SSOCT (SS-1000, CASIA, Tomey, Nagoya, Japan). The associations of surface features with iris parameters were analysed using a generalised estimating equation.
Results A total of 1704 subjects (3297 eyes) were included in the analysis. The majority was Chinese (86.4%), and 63.2% were females, and their mean age (±SD) was 61.4±6.6 years. After adjusting for age, sex, ethnicity, pupil size and corneal arcus, higher iris crypt grade was independently associated with smaller iris volume (β=−0.54, p<0.001), whereas darker irides and higher iris furrow grade were associated with larger iris volume (β=−0.041, p<0.001) and (β=0.233, p<0.001), respectively. Lighter coloured irides with more crypts and/or more furrows were also associated with less convexity (crypts: β=−0.003, p=0.03; furrows: β=−0.004, p=0.007; and colour: β=−0.001, p=0.005).
Conclusions Iris surface features were highly correlated with iris volume and curvature. Irides with more crypts have a smaller volume; and darker irides with more furrows have a larger volume. Lighter irides with more crypts and/or furrows have less convexity.
Statistics from Altmetric.com
Structural and dynamic features of the iris play an important role in mechanistic blockage of aqueous outflow in primary angle closure glaucoma, a leading cause of blindness worldwide.1 ,2 Objective and quantitative assessment of the irides by modern imaging devices such as ultrasound biomicroscopy (UBM) and anterior segment optical coherence tomography (AS-OCT) showed that the irides with larger cross-sectional area, thicker periphery, more convexity and lesser loss of iris volume during physiological mydriasis have a higher risk of angle closure.3–7 However, the assessment of iris by these devices is costly and may need technical expertise.
By contrast, iris surface features (such as iris crypts, furrows and colour) could be easily assessed during standard slit-lamp examination or captured with iris photography. In our previous studies, we developed an iris grading system for Asian eyes and showed that iris surface features were associated with cross-sectional iris thickness (IT), which was assessed by using AS-OCT.8 ,9
The anterior segment swept-source optical coherence tomography (SSOCT, Casia SS-1000, Tomey, Nagoya, Japan) is a new form of AS-OCT that uses 1310 nm wavelength to deliver 30 000 A-scans per second with an axial resolution of <10 µm. With a faster scan speed than AS-OCT, SSOCT takes an entire 360° assessment of the anterior segment of the eye in 2.4 s by capturing 128 cross-sectional slices (one anterior chamber angle every 1.4°).10 Mak et al11 reported that SSOCT iris volume calculation was more robust than that of the AS-OCT iris volume estimation.
In this study, we aimed to assess the associations of iris surface features with iris volume measured with SSOCT in an Asian population.
We recruited 1950 subjects over a 4-month period (June–September 2013) from a government-run polyclinic in the central region of Singapore.12 Inclusion criteria were subjects aged ≥50 years old with phakic eyes, with no known history of glaucoma and intraocular surgery, and no history of penetrating eye trauma. These subjects came to the polyclinic for primary healthcare or who accompanied the patients at the clinic and were asked to participate in the study. The socio-economic and demographic characteristics of the subjects in the central region are similar to those of Singapore as a whole. All subjects underwent the following examinations: (1) measurement of visual acuity, (2) iris photography, (3) non-contact intraocular pressure measurement, (4) biometric measurement by Lenstar (LS900, Haag-Streit, Switzerland), (5) SSOCT imaging, (6) slit-lamp biomicroscopy (Haag-Streit, Switzerland), (7) Goldmann applanation tonometry (Haag-Streit) and (8) gonioscopy.
Iris photography and grading
Colour photographs of irides were taken using an iris imaging system (MEC-5-ASL-D7100-N85, Miles Research, California, USA) that consisted of a 24 megapixel Nikon Camera (Nikon D7100, Nikon, Japan), Nikon 85 mm macro lens (Nikon D3200, Nikon), adjustable side lighting illuminator (MEC-5-ASL, Miles Research) and chinrest/camera support (CRCS-FH4, Miles Research) under normal lighting conditions. The side lighting illuminators were angled at 60°. The camera setting was kept constant at aperture priority dial, aperture stop (f18) shutter speed (1/60), ISO (200), flash power (1/2) and focal length (1 ft/0.286 m). For grading purpose, photographs were viewed on a 1366×768/60 Hz resolution screen, using the viewing software ACDSee Photo Manager V.11.0 (ACD Systems, Washington, USA).
A single observer (ES) used the Asian Iris Grading scheme8 ,9 to grade iris crypts and furrows from iris photographs (see online supplementary figure S18). In brief, based on the number and size of crypts present, irides were classified into five grades as follows: grade 1 (no crypts); grade 2 (1–3 crypts); grade 3 (at least four crypts ≤1 mm in diameter); grade 4 (at least four crypts >1 mm in diameter) and grade 5 (numerous crypts >1 mm in diameter, covering nearly the entire iris). Iris furrows were classified into three grades as follows: grade 1 (no furrows); grade 2 (five furrows or fewer present, extending 180° or less); and grade 3 (five furrows or more present, extending 180° or more).
Iris colour was measured from iris photos and was derived from the middle region of the iris, where it is less affected by ocular conditions such as corneal arcus or eyelid occlusion. The ciliary and pupillary margins of the iris were detected automatically and a colour intensity threshold was applied on the iris photograph to remove extremely dark and bright regions, which were caused by distortions, such as shadows from crypts and reflections from camera flash. The iris region was identified using a custom program written on MATLAB (MathWorks, Natick, Massachusetts, USA). The mean colour obtained from the identified region was then calculated and represented using a single variable (L*) (a value of 100 corresponds to perfect white and that of 0 to black) in the LAB colour space (see online supplementary figure S2). The LAB colour space provides a quantitative and intuitive representation that relates directly to visual observations of colour. The L* component in particular has been used in the quantification of skin colour and has been shown to correlate with melanin content.13–15
SSOCT imaging and analysis
All participants underwent anterior segment imaging using the SSOCT before any contact procedure or eye drops were instilled under dark room conditions. Subjects were examined with the three-dimensional (3-D) angle analysis scan in the primary gaze position while directed towards an internal fixation light. To avoid eyelid artefacts, the operator opened both eyelids, avoiding inadvertent pressure on the globe during scanning. The 3-D angle analysis scan is a volume scan comprising 128 radial scans, each 16 mm in length and 6 mm in depth, to image the iris and anterior chamber. Each image was averaged automatically from three consecutive scans by the algorithm native to the Casia system.
A single observer (TAT) used the 360° SSOCT viewer (V.6.0, Tomey) to analyse 8 out of 128 frames (16 anterior chamber angles, in 22.5° increments) from a single 3-D angle analysis SSOCT scan of each eye. The only user input was to mark the scleral spur location (defined as the inward protrusion of the sclera where a change in curvature of the corneoscleral interface was noted). After the grader manually marked the scleral spurs of eight frames per scan, the program automatically computed SSOCT parameters such as pupil size, IT at 750 and 2000 µm from the scleral spur, iris curvature and iris volume. The algorithm may be subjected to segmentation error (around 10% subjectively assessment by the author) when detecting of the edges and these segmentation errors were manually corrected by the grader.
Pupil size was defined as the distance between the two tips of irides. IT750 was IT measured at 750 µm from scleral spur, whereas IT2000 was measured at 2000 µm. The iris curvature was measured by drawing a perpendicular line from the iris pigment epithelium at the point of greatest iris convexity, to the line extending the most peripheral to the most central points of the iris pigment epithelium (see online supplementary figure S3).
The iris volume was calculated as a summation of pixel volume derived from individual B-scans. A pixel volume is represented by where r is the distance between the pixel and the radial rotation axis and θ is the angle between adjacent B-scans. By using integral calculus, the iris volume was derived from the equation where s represents the number of analysed frames (8 in this study), n represents the pixel number of the iris in a B-scan and r represents the distance between pixel and the radial rotation axis.11 ,16 The scaling factor was 9.7 µm/pixel.
Statistical analysis was performed using R software V.3.22 (R Development Core Team 2008, http://www.R-project.org). The association between iris surface features (independent variable) and SSOCT iris parameters (dependent variables) was assessed by linear regression models with generalised estimating equations to account for inter-eye correlation and were adjusted for potential confounders such as age, sex, ethnicity, pupil size and corneal arcus in different models. Statistical significance was set at p <0.05.
Of the 1950 consecutive recruited subjects, 3297 eyes (1646 right eyes and 1651 left eyes) were included in the final analysis of iris crypt and furrow. We excluded 603 (15.4%) eyes because of ungradable iris photographs (418 eyes), poor SSOCT image quality (74 eyes) and missing clinical data (111 eyes). We further excluded 301 eyes due to failing of iris colour measurement by our customised algorithm so that 2996 eyes were only included in final analysis of iris colour. There was no statistically difference in age, gender and ethnicity distribution between included and excluded subjects (all p>0.05). Table 1 shows the demographic and clinical characteristics of the subjects. The mean age of the included subjects was 61.4±6.6 years old, and of them, 63.2% were females and 86.4% were Chinese. The distribution of all three iris surface features of the study population is shown in figure 1.
Table 2 shows the associations of the iris crypts and furrows with SSOCT iris-related parameters. After adjusting for age, sex and ethnicity (model 1 in table 2), increasing grade of crypt was significantly associated with smaller iris volume (β (change in volume in mm3 per one grade higher)=−0.553, p<0.001), whereas increasing furrow grade was significantly associated with larger iris volume (β=0.226, p<0.001). The associations between these two iris surface features and iris volume remained robust after further adjusting for pupil size and corneal arcus (model 2 in table 2).
In addition, higher grade of crypt and furrow were also associated with lesser convexity of iris (crypt: β=−0.003, p=0.03; furrow: β=−0.004, p=0.007) after adjusting for age, sex, ethnicity, pupil size and corneal arcus (model 2 in table 2). We also confirmed the previously identified association8 that increasing grade of crypt was associated with thinner IT2000 (β=−0.01, p<0.001), while higher grade of furrow was associated with thicker IT750 (β=0.004, p<0.001; table 2).
For iris colour (0 represents dark iris while 100 represents light iris), darker irides were significantly associated with larger iris volume (β=−0.005, p<0.001) and more curved irides (β=−0.001, p<0.001) after adjusting for age, sex and ethnicity (model 1) and the similar result was observed even after further adjusting with pupil size and corneal arcus (model 2; table 3) We also confirmed the previously published association8 that darker irides were associated with thicker IT750 (β=−0.001, p<0.001) and thicker IT2000 (β=−0.001, p<0.001).
The iris surface features such as crypts of Fuchs, contraction furrows and colour could be graded during slit-lamp examination and/or documented by slit-lamp photographs. These features could be a surrogate for assessing iris parameters such as thickness and volume that are measured by AS-OCT or UBM, but not able to be observed during slit-lamp examination. In this study, we assessed the associations of iris surface features with iris volume and other SSOCT iris parameters in Asian eyes and found that irides with more crypts were associated with smaller iris volume, whereas darker irides with more furrow were associated with larger iris volume.
In our previous studies, we found that irides with more crypts were thinner and darker irides with more furrows were associated with thicker IT750.8 In this study, we found similar results for IT and further validated the associations with iris volume with a large sample size recruited from a community clinic. The thickness is measured at a point or a particular section of the iris, whereas the volume provides a more complete representation of the iris as a whole. Moreover, we used a more robust iris volume calculation method by using SSOCT to assess the association between these parameters.
Iris crypts are hypoplasia or atrophy of iris stroma and appear as the pits on the iris surface.17 It could be explained that the presence of crypts would be associated with smaller iris volume due to a lack of iris tissue in the crypts. Iris furrows are contraction folds and become more visible when the pupil dilates,17 which can translate into a larger iris volume. Regarding iris colour, a darker iris will have more melanocytes and melanin content;18 therefore, the presence of a darker iris may be associated with a larger iris volume.
The hints from surface features may provide better insight for assessing iris volume to predict the risk of angle closure diseases. For example, an iris with grade 1 crypt may have a larger volume than that of grade 5 crypt by 5.99 mm3, while an iris with grade 1 furrow may have a smaller volume by 1.36 mm3 compared with that of grade 3 furrows. It was reported that iris volume of subject with closed angle was 2.52 mm3 larger than those with open angles after adjusting for confounders such as age, sex and other ocular biometric parameters under conditions of physiological dilatation.16 Thus, an eye with no crypt and more contraction furrows in periphery may have higher risk for angle closure than that with more crypts and less furrow. However, since only 8% of the study eyes in the present study had occludable angles, further studies involving more angle closure patients would be warranted to directly assess whether these iris surface features could predict the risk of angle closure diseases.
Limitations of this study included the fact that grading of iris surface features and marking of scleral spur were subjective and may have the systemic bias. However, the good inter-grader and intra-grader reliability has been reported elsewhere.10 ,19 Moreover, Seager et al20 have shown that iris area and volume were less affected by scleral spur markings. We analysed only 8 frames out of 128 frames of the SSOCT volume scan of an eye so that our results may therefore be subject to 1.6% mean absolute per cent error in calculation of iris volume.21 Our participants were recruited from a community clinic, and thus some selection bias may exist, yet we have minimised the bias by ensuring that these participants visited the clinic for minor non-ocular health issues.
In summary, we demonstrated the correlation of iris surface features with iris volume and curvature in Asian eyes. Our study showed that irides with more crypts have smaller volume, whereas darker irides with more furrows have larger volume. Lighter irides with more crypts and/or furrows have less convexity.
Contributors Design of study: MCLT, JHMQ, TA, C-YC; conduct of the study: TAT, JC, WS, MT; collection and management of data: TAT, JC, ES, SGT; analysis and Interpretation of data: TAT, JC, YS; preparation of manuscript: TAT, JC, TA, C-YC; review or approval of manuscript: JC, MCLT, JHMQ, TA, C-YC.
Funding National Medical Research Council (NIG/1069/2012 and CIRG15may030).
Competing interests C-YC is supported by National Medical Research Council (CSA/033/2012).
Patient consent Obtained.
Ethics approval The study was approved by the SingHealth Centralised Institutional Review Board, and adhered to the tenets of the Declaration of Helsinki.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.