## Purpose

To establish a normative database of endothelial cell density (ECD) using in vivo specular microscopy in children under 5 years old.

## Design

Cross-sectional study.

## Methods

Specular microscopy was performed during a clinic visit in cooperative children in the standard upright position. In uncooperative children, specular microscopy was performed in the lateral decubitus position under general anesthesia, before surgery for other reasons. Corneal diameter (CD) was measured in children undergoing general anesthesia and was stratified according to age.

## Results

One hundred and eighteen eyes of 118 patients were included in the study. The mean patient age was 2.6 ± 1.4 years (range 0.1–5 years) and the mean ECD was 3746 ± 370 cells/mm ^{2 }(range 3145–5013 cells/mm ^{2 }). The mean CD under 2 years of age was 11.85 ± 0.57 mm (n = 40, range 10.50–12.75 mm). Up to 2 years of age, ECD was more inversely correlated with CD than with age (r = −0.61, *P *< .0001; r = −0.38, *P *= .01, respectively). In contrast, after the age of 2 years, the ECD was inversely correlated with age but not with CD (r = −0.27, *P *= .02; r = −0.24, *P *= .2). Between the first and second year of life, the rate of ECD decrease was significantly higher than between 2 and 5 years of age (8.2%, 334 cells/mm ^{2 }vs 2.7%, 100 cells/mm ^{2 }a year, respectively).

## Conclusion

In the first 2 years of life there is a rapid decline in ECD, which is likely related to growth in CD and hence surface area. After the cornea reaches adult size, the ECD decreases at a rate similar to that reported in adults.

The corneal endothelium is essential in maintaining the clarity of the cornea by regulating the degree of stromal hydration. Endothelial cell layer or function may be defective in congenital or genetic diseases such as Peters anomaly, posterior polymorphous corneal dystrophy, or congenital hereditary endothelial dystrophy. Similarly aging or degenerative changes, as well as acquired stress, such as trauma or intraocular surgery, can significantly reduce the number or function of these cells, resulting in corneal edema and loss of corneal clarity.

Studies have been performed in older children and adults to assess normal endothelial cell density (ECD). However, very little work has been performed in children under the age of 5 years, at a critical period of eye and vision development. The inability to evaluate the endothelium at a young age owing to patient cooperation creates a gap in our current knowledge. Recently, a technique to obtain ECD under general anesthesia in the lateral decubitus position has been validated by our group.

It is known that the human corneal endothelium has no proliferative or regenerative capacity, and therefore following a stressful event no mitosis occurs to compensate for the loss of endothelial cells. Rather, lateral cell migration takes place to distribute the remaining cells evenly over the entire corneal surface. It is believed that there is a rapid decline in the corneal endothelial cell density in the first years of life, which slows down afterwards. This knowledge derives primarily from a small number of animal and postmortem studies, with only 2 human in vivo studies, which were limited by the small number of children under 5 years old. In addition, it is not clear whether this decline in ECD represents a change in corneal surface area in infancy or is attributable to actual cell apoptosis. Speedwell and associates hypothesized that the average 37.9 mm ^{2} growth in corneal endothelial surface area during the first year of life accounts for the decrease in ECD without an actual loss of endothelial cells. However, this hypothesis was based on the investigation of only 2 infants’ corneas with longitudinal follow-up over 6 months within the first year, without showing any ECD data beyond the first year. In addition, Speedwell and associates did not validate their technique for specular imaging of the infant corneal endothelium. Therefore, this topic mandates further investigation.

The purpose of our study is to explore the normal ECD and its relationship with corneal diameter in young children using a validated technique. To our knowledge, this is the largest human cohort investigated for ECD under the age of 5 years.

## Methods

This study was approved by the Research Ethics Board of the Hospital for Sick Children, Toronto, Ontario, Canada, and adheres to the tenets of the Declaration of Helsinki. Informed consent to participate in the study was obtained from all patients’ guardians. Consecutive children with normal anterior segments under the age of 5 years underwent specular microscopy imaging using a noncontact specular microscope (Konan Medical, Hyogo, Japan). Images were taken either during examination under anesthesia in the lateral decubitus position or during a clinic visit in the upright position, depending on patient cooperation and only after a normal anterior segment was validated. In bilateral acquisition only 1 eye was randomly included. ECD was calculated using the center method technique by 2 qualified graders, which is comparable to the variable frame technique originally described by the Cornea Donor Study Group. Image acquisition was performed repeatedly until a high-quality-grade image with at least 50 analyzable cells was attained. *An image was considered unanalyzable when the quality of image was poor with indistinct cell borders or when adequate numbers of contiguous cells could not be marked for the final analysis to include at least 50 cells *.

A percentage difference between the cell counts of the 2 graders was calculated using the lower of the 2 densities. If a percent difference of less than 5% was obtained, an average of both analyses was calculated and used as the final ECD. If the intergrader difference was greater than 5% for a given image, that image was reanalyzed by an adjudicator. The average between the adjudicator’s analysis and analysis of the reader that was closer to the adjudicator was calculated. Horizontal corneal diameter was recorded during examination under anesthesia using a set of purpose-made circular rings of known diameter in 0.25-mm intervals. The data were then stratified according to 12-monthly age groups from 0 to 5 years. Pearson correlation was used to assess the correlation between age and ECD and between the changes in corneal diameter and ECD. Fitted linear or quadratic line plots were used to assess the relationship between ECD and corneal diameter and between ECD and age as appropriate. A *P *value of .05 and under was considered statistically significant. Power calculation following our previous validation study had determined that a sample size of 86 eyes produces a 95% confidence interval equal to the sample mean plus or minus 49.741 cells/mm ^{2 }when the estimated standard deviation between age groups is 232 cells/mm ^{2 }.

## Results

One hundred and twenty-three eyes of 123 patients 5 years of age and under were recruited to the study. Five patients were excluded owing to poor-quality images that were not amenable to reliable analysis (ie, indistinct cell borders with less than 50 cells marked or analyzed). The average number of cells analyzed was 108 ± 19 cells ( Table ). The mean patient age was 2.6 ± 1.4 years (range 0.1–5 years) and the mean ECD was 3746 ± 370 cells/mm ^{2 }(range 3145–5013 cells/mm ^{2 }). The Table shows the mean measured ECD for each age group with an overall decrease of 11.7% over 5 years.

Age Group, Years (n) | Mean Age, Years ± SD | Mean ECD, Cells/mm ^{2 }± SD |
Corneal Diameter, mm ± SD (n) | Number of Cells Analyzed, Mean ± SD (Range) |
---|---|---|---|---|

0–1 (20) | 0.5 ± 0.3 | 4056 ± 451 | 11.5 ± 0.6 (18) | 101 ± 21 (64–138) |

1–2 (26) | 1.6 ± 0.3 | 3722 ± 386 | 12.1 ± 0.3 (22) | 109 ± 21 (66–147) |

2–3 (22) | 2.4 ± 0.3 | 3782 ± 302 | 12.1 ± 0.4 (15) | 106 ± 21 (54–143) |

3–4 (26) | 3.5 ± 0.3 | 3686 ± 313 | 11.9 ± 0.4 (10) | 112 ± 15 (64–133) |

4–5 (24) | 4.5 ± 0.3 | 3582 ± 254 | 12.2 ± 0.2 (5) | 108 ± 19 (58–133) |

Between the first and second year of life, the rate of ECD decrease was significantly higher than between 2 and 5 years of age (8.2%, 334 cells/mm ^{2 }vs 2.7%, 100 cells/mm ^{2 }a year, respectively, Table ). The mean corneal diameter under the age of 2 years was 11.85 ± 0.57 mm (n = 40, range 10.50–12.75 mm, Table ). The mean corneal diameter over 2 years was 12.01 ± 0.40 mm (n = 30, range 11.25–12.75 mm, Table ). Linear regression analysis of ECD vs age and of ECD vs corneal diameter showed that, up to 2 years of age, ECD was more inversely correlated with corneal diameter than with age (r = −0.61, *P *< .0001; r = −0.38, *P *= .01, respectively, Figure 1 , Top left and Bottom left). In contrast, after the age of 2 years, the ECD was inversely correlated with age but not with corneal diameter (r = −0.27, *P *= .02; r = −0.24, *P *= .2, respectively Figure 1 , Bottom right and Top right). The rate of ECD decrease before and after the age of 2 was 259 cells/mm ^{2 }a year (6.2%) and 90 cells/mm ^{2 }a year (2.3%), respectively ( Figure 1 , Bottom left and Bottom right).

Linear regression analysis demonstrated 2 different rates of ECD decline in relation to age, before and after 2 years ( Figure 1 , Bottom left and Bottom right, respectively). Therefore we applied a quadratic equation to model predictive changes in ECD over 5 years. Our quadratic fit model showed a total of 8.7% decrease in ECD in the first 2 years, reflecting an annual decrease rate of 4.4% a year. Thereafter, a plateauing in ECD was demonstrated in the following 3 years with a total of 3.8% decrease reflecting an annual rate of decrease of 1.3% ( Figure 2 ).