To examine the distribution of corneal endothelial cell density (ECD) and relating factors in ophthalmologically normal Japanese in a population-based setting.
Cross-sectional, population-based study.
All residents of Kumejima Island, Japan, located in southwestern Japan (eastern longitude 126° 48′ and northern latitude 26° 20′), aged 40 years and older, were asked to undergo a comprehensive questionnaire and ocular examination, including noncontact specular microscopy of corneal endothelial cells.
Of the 4632 residents, 3762 (81.2%) underwent the examination. The mean ECD among all ophthalmologically normal participants (n = 2602), men (n = 1329), and women (n = 1273) was 2943 ± 387 cells/mm 2 , 2927 ± 385 cells/mm 2 , and 2959 ± 389 cells/mm 2 , respectively, with a significant inter-sex difference after adjusting for age ( P = .001). Mean ECD was significantly lower in subgroups with a history of outdoor work compared to corresponding subgroups after adjusting for age and sex ( P = .034). Linear regression analyses with an adjustment for age and/or sex showed that age was significantly negatively correlated with ECD with a slope of −7.43/mm 2 /year ( P < .001), indicating a cell loss rate of 0.25% per year of age. Higher intraocular pressure was significantly correlated with lower ECD, with a slope of −9.87/mm 2 /mm Hg ( P < .001).
Mean ECD in ophthalmologically normal Japanese in Kumejima aged 40 years or older was 2943 cells/mm 2 . Older age, male sex, higher intraocular pressure, and history of outdoor work were also identified as factors correlating with lower ECD.
Corneal endothelial cells, which derive from the neural crest and form the corneal endothelial lining as a single layer of hexagonal cells, function to maintain corneal clarity. Because proliferation of human corneal endothelial cells does not continue throughout a person’s lifetime, wound healing in human corneal endothelium is mainly accomplished by cell spreading. Thus, when corneal endothelial cells are damaged by various causes such as intraocular surgery, trauma, inflammation, and the aging process, corneal endothelial cell density (ECD) decreases. Clinical observations indicate that an ECD of 400 to 600 cells/mm 2 is a crucial point at which endothelial decompensation develops. Therefore, ECD is clinically a very important parameter.
Previous studies examined the morphologic parameters of corneal endothelial cells, including ECD, in patients before cataract surgery, in normal volunteers, and in patients without anterior segment pathologies. Most of those studies demonstrated decreases in ECD and the percentage of hexagonal cells, and an increase in the coefficient of variation in cell size with aging. To the best of our knowledge, there are no reports of normal ECD values or its distribution across age, sex, and related factors in a population-based setting. The present study was conducted to examine the distribution of ECD and its relating factors in ophthalmologically normal Japanese based on the data obtained in a population-based study performed in a rural southwestern island of Japan and to provide reference data of corneal ECD in ophthalmologically normal Japanese.
Subjects and Methods
The distribution of ECD and its relating factors were examined as a part of a population-based epidemiologic survey on ocular diseases in residents of Kumejima Island aged 40 years or older. Kumejima is a 63.2-km 2 island located in the southwestern part of Japan (eastern longitude of 126° 48′ and northern latitude of 26° 20′) west of the main island of Okinawa and has a population size of approximately 9000, with residents almost exclusively originating from the Okinawa prefecture. This study was conducted between May 1, 2005, and August 31, 2006. According to the official household registration database, Kumejima had 5249 residents aged 40 years or older in 2005. After excluding residents who died, moved, or could not be located in Kumejima during the study period (n = 617), 4632 residents were eligible for the study. All of these residents were asked by letter and telephone to undergo examinations held at the public hospital of Kumejima. Home visits and examinations were performed for inpatient, paralyzed, and disabled residents.
All participants provided written informed consent prior to the examinations. After body weight, height, and brachial blood pressure measurements were obtained, a structured questionnaire was administered that included questions about occupation, health history, surgery and trauma history, smoking habit, history of outdoor work, and use of hats and sunglasses. Occupation was categorized into 6 groups: farming, fishing, service, office work, housewife, and other.
A detailed ophthalmic examination was performed by experienced examiners and ophthalmologists, and included uncorrected and best-corrected visual acuity, refraction, slit-lamp examination of the anterior segment, intraocular pressure (IOP), central corneal thickness, specular microscopic examination of corneal endothelial cell morphology, anterior chamber depth, axial length of the eye, ophthalmoscopy, photography of ocular fundus, and visual field. Refraction was measured using an autorefractometer (ARK-730; Topcon, Tokyo, Japan). IOP was measured 3 times using a Goldmann applanation tonometer under topical anesthesia and the median value was adopted. Corneal endothelial cell morphology and central corneal thickness were examined with noncontact specular microscopy (SP-2000; Topcon). Anterior chamber depth and the axial length of eye were measured with the IOL Master (Carl Zeiss Meditec, Dublin, California, USA). Digital color fundus photographs (30 and 45 degrees) were obtained using a nonmydriatic ocular fundus camera system (Image Net TRC-NW7; Topcon). The examinations that did not require direct eye contact, including tests of refraction, visual acuity, specular microscopy, IOL Master, fundus photography, and slit-lamp examination, were performed first. IOP measurement and gonioscopy were performed last.
When participants could not visit the hospital, ophthalmologists visited their homes and performed the examinations, including the slit-lamp examination with a hand-held slit lamp (SL-15; Kowa, Nagoya, Japan), IOP measurements with a Perkins tonometer (Haag-Streit UK, Harlow, United Kingdom) or hand-held tonometer (Tonopen XL; Bio-Rad Laboratories, Hercules, California, USA), and indirect and direct ophthalmoscopy (BS-II and BXα-13; Neitz, Tokyo, Japan).
For the evaluation of corneal endothelial cell morphology, the examiner hand-digitized the center of each cell in a contiguous group of cells and used a computer algorithm to calculate the ECD.
All data were stored at the University of the Ryukyus and University of Tokyo. Data analyses were performed using SPSS 15.0J for Windows (SPSS Japan Inc, Tokyo, Japan). The χ 2 test, paired and unpaired t test, and Pearson correlation analysis were used. Analysis of variance (ANOVA) and linear regression analysis were used to examine the association of each variable with ECD after adjusting for age and/or sex. Data are shown as mean ± standard deviation unless otherwise specified. Low R values in the Pearson correlation analysis imply loose association, which may not be clinically significant, even though statistical significance was observed because of large sample size. Thus, R values less than 0.1 were considered not to be clinically significant.
Of the 4632 eligible residents, 3762 (81.2%) underwent the examination. The 3762 participants were younger than the 870 nonparticipants (59.1 ± 14.9 vs 61.8 ± 14.0 years, P < .001, unpaired t test) and there were more women among the participants (male/female ratio, 1833/1929 vs 555/315, P < .001, χ 2 test).
Of 7524 eyes (3762 participants), 1160 right eyes and 1165 left eyes were excluded for various reasons, including difficulties in obtaining clear images of corneal endothelial cells with specular microscopy, history of intraocular surgery, presence of ocular disease, and best-corrected visual acuity of 20/200 or worse ( Table 1 ). One right eye and 1 left eye that suffered from ocular trauma were included in the eyes that were excluded from this study because of history of intraocular surgery. Only these 2 eyes suffered from ocular trauma that was thought to have influence on ECD. Ophthalmologically normal eyes were defined as eyes that had no apparent ocular diseases, no history of ocular diseases or intraocular surgery, and best-corrected visual acuity over 20/200. As a result, ophthalmologically normal 2602 right eyes and 2597 left eyes were included in the analysis.
|Reason for Exclusion a||Right Eyes||Left Eyes|
|Screened in own homes||190||190|
|Difficulties measuring endothelial cell density b||41||40|
|History of intraocular surgery||604||607|
|Acute or chronic angle closure||100||101|
|Glaucoma, ocular hypertension, or pseudoexfoliation||187||181|
|Best-corrected visual acuity of 20/200 or worse||24||23|
b Reasons for difficulties measuring endothelial cell density were dementia (3 right and 3 left eyes), bedridden (6, 6), artificial eye (4, 2), phthisis (1, 4), severe corneal opacity (17, 12), large pterygium (3, 3), nystagmus or strabismus (4, 4), and difficulties or refusal to open their eye (3, 6).
Among the 2357 subjects for whom both eyes were eligible, the mean ECD in the right eyes was 2955 ± 382 cells/mm 2 (95% CI, 2940-2971) and that in the left eyes was 2954 ± 384 cells/mm 2 (95% CI, 2939-2970), with a significant positive correlation between eyes (R = 0.749, P < .001, Pearson correlation analysis) and no significant difference between right and left eyes ( P = 0.871, paired t test). The mean absolute difference in ECD between the right and left eyes was 218 ± 161 cells/mm 2 . The absolute difference in ECD between the right and left eyes did not significantly correlate with age (R = -0.025, P = .227, Pearson correlation analysis). Because ECD was highly correlated between the right and left eyes, and the results of the left eyes were similar to those of the right eyes, only the results from the ophthalmologically normal 2602 right eyes are presented below.
The distribution of ECD is shown in the Figure . The mean ECD in each sex and age range is shown in Table 2 . The mean ECD in all ophthalmologically normal participants (n = 2602) was 2943 ± 387 cells/mm 2 (95% CI, 2928-2958), 2927 ± 385 cells/mm 2 (95% CI, 2906-2948) in men (n = 1329) and 2959 ± 389 cells/mm 2 (95% CI, 2938-2981) in women (n = 1273), with a significant inter-sex difference after adjusting for age ( P = .001, ANOVA).
|N||Mean ± SD||N||Mean ± SD||N||Mean ± SD|
|Age range (years)|
|40–49||427||3012 ± 348||400||3050 ± 369||827||3031 ± 359|
|50–59||412||2976 ± 363||303||2999 ± 384||715||2986 ± 372|
|60–69||236||2876 ± 383||222||2920 ± 342||458||2897 ± 364|
|70–79||205||2770 ± 386||263||2869 ± 403||468||2825 ± 398|
|80–||49||2671 ± 519||85||2774 ± 442||134||2736 ± 473|
|Total||1329||2929 ± 385||1273||2959 ± 389||2602||2943 ± 387|
Mean ECD in the subgroups are shown in Table 3 , stratified based on sex, and on the presence or absence of each of the following factors: diabetes mellitus, hypertension, smoking habit, contact lens wear, and outdoor work. Mean ECD was significantly lower in the subgroups with a history of outdoor work compared to the corresponding subgroups after adjusting for age and sex ( P = .034, ANOVA).
|N||ECD a||P b|
|Male||1329||2927 ± 385||.001|
|Female||1273||2959 ± 389|
|+||219||2880 ± 384||.24|
|−||2383||2949 ± 387|
|+||984||2902 ± 399||.84|
|−||1618||2967 ± 378|
|+||1112||2934 ± 379||.53|
|−||1490||2950 ± 393|
|Contact lens wear|
|+||68||2978 ± 377||.15|
|−||2534||2942 ± 388|
|+||1612||2909 ± 394||.034|
|−||990||2997 ± 370|