The relationship between the optic disc area and the vertical cup/disc ratio is presented by means of a graphic nomogram after studying a total of 365,356 normal optic nerves. For this purpose, the Laguna ONhE application was used, which analyses the distribution of hemoglobin based on the colorimetric analysis of fundus color images. The 1%, 5%, 50% and 95% percentiles are given for disc areas between 1 mm2 and 4 mm2. The highest cup-to-disc ratios (1%–5% percentiles) were close to 0.7 for large discs, while for more common sizes (1.9–2.0 mm2) they were close to 0.6 and for small ones to 0.5. Data allowed the Laguna ONhE Globin Discriminant Function (GDF) index to be corrected, in such a way that it could, then, be verified in 544 normal eyes that it was less dependent on the vertical cup/disc ratio, whereas in 436 confirmed and suspected glaucoma an intimate relationship was reported.
Se expone, mediante un nomograma gráfico, la relación entre el área del discoóptico y el cociente vertical copa/disco, estudiando 365.356 nervios ópticos normales. Para ello se usó la aplicación Laguna ONhE, que analiza la distribución de hemoglobina en base al análisis colorimétrico de retinografías a color. Se señalan los percentiles 1%, 5%, 50% y 95% para áreas de disco comprendidas entre 1 y 4 mm2. Los cocientes copa/disco mayores (percentiles 1%–5%) estuvieron próximos a 0.7 en las papilas grandes, mientras que en las de tamaños más habituales (1.9–2.0 mm2) estuvieron próximos a 0.6 y en las pequeñas a 0.5. Los datos permitieron corregir el índice de Función Discriminante de Globina (GDF) del ONhE de Laguna, para poder comprobar a continuación en 544 ojos normales que era menos dependiente de la relación vertical copa/disco, mientras que en 436 glaucomas confirmados y sospechosos de glaucoma existía una relación íntima.
The Vertical Cup/Disc Ratio (CDR) has been widely used as a diagnostic criterion for glaucoma.1,2 However, it is well known that the size of the optic disc (Disc Area, DA) has an important influence on the variability of the CDR in normal eyes, which makes it difficult to use as a diagnostic criterion.3–5
We have recently published a method for estimating DA in fundus photographs.6 On the other hand, the Laguna ONhE application7 estimates, among other morphological parameters, the CDR value from the distribution of haemoglobin in the optic disc.8 The intention of this paper has been to establish the frequency of CDR, expressed in percentiles with respect to the DA value to facilitate its clinical interpretation and reduce its influence in the evaluation of glaucoma with the GDF index.
Material and methodsThe frequency distribution of CDR for different DA values was studied in 365,356 optic disc photographs obtained, during a screening activity in Scandinavia, with the Topcon NW-400 fundus camera (Topcon. 75-1 Hasunuma-cho, Itabashi-ku, Tokyo, Japan), which showed normal values of the Globin Discriminant Functión (GDF) index in the Laguna ONhE application.
Until now, variables used for GDF calculation included relative haemoglobin concentrations in optic nerve sectors particularly sensitive to glaucoma, such as the superior and inferior optic disc, a deep learning classifier, optic disc area and CDR value.
The Laguna ONhE application was modified so that the GDF function excluded the CDR value and the relationship between the two indices was observed in 544 normal eyes and 436 glaucomas, of subjects from the Canary Islands.
The images were obtained in essentially Caucasian subjects, who had previously signed their consent to be used for clinical research, complying with the protocols required by the Declaration of Hensinki. Given the sample size, both eyes were included.
Comparison between variabilities was performed using the t-test and between correlation coefficients using the z-test on Fisher z-transformed correlation coefficients of the MedCalc program (MedCalc® Statistical Software version 23.0.2. MedCalc Software Ltd, Ostend, Belgium; https://www.medcalc.org; 2024).
ResultThe 1%, 5%, 50% and 95% percentiles of the CDR values for the different papillary sizes between 1 and 4 mm2 are shown graphically (Fig. 1).
Although optic disc area was taken into account in the GDF index, large nerves were still influenced by CRD, which was not always related to glaucoma.
Removing CRD from the GDF calculation showed a much lower relationship between the two indices in normal eyes (R2 = 0.0729) than in eyes with confirmed or suspected glaucoma (R2 = 0.6157), (P < .0001), indicating better in the latter the severity of damage (Fig. 2).
As a consequence, GDF results in normal cases show relatively little variability (17.72 ± 10.94). In glaucomatous eyes, on the contrary, the results have a higher variability (−38.40 ± 34.79) (P < .0001), better representing the degree of defect.
DiscussionThe results are consistent with the mean values of the 97.5% (0.63) and 99.5% (0.72) percentiles of CDR that have been described for the North American population, where disc size was not taken into account.9
Although specific studies like ours would be desirable in other races and characteristics, the North American study suggests approximately 0.2−0.4 higher values of CDR in races other than the white, as well as values 0.1 higher than the mean in males and 0.1 lower in females. A limitation of this study is that it essentially addresses the case of the Caucasian race. However, we are applying the results in Brazil for verification in other breeds.
On the other hand, this nomogram may be useful to elucidate whether the asymmetries between the two eyes greater than 0.2 in CDR, which are considered characteristic of asymmetric forms of glaucoma,9,10 are real or caused by significant differences between the two optic disc sizes.
ConclusionThe graph attached as Fig. 1 can serve as a nomogram for the clinical assessment of the CDR index, allowing for case-specific interpretation according to the size of the optic nerve.
Ethical considerationsThe images were obtained in subjects who had previously signed their consent to be used for clinical research, complying with the protocols required by the Declaration of Hensinki.
FundingThe present work has been funded by the company Instrumentación y Oftalmología, INSOFT S.L., Spain.
The authors are partners and owners of the company INSOFT S.L.
