Peripapillary retinal nerve fibre thickness in patients with primary open-angle glaucoma

Background The purpose of this study was to determine the difference in peripapillary retinal nerve fiber layer (RNFL) thickness in patients with preperimetric glaucoma and open angle glaucoma (POAG) in comparison to healthy population, as well as to determine the difference in thickness of peripapillary RNFL according to progression of the disease. Methods In this study, 120 patients were included . On the basis of clinical finding four groups of patients were formed: group without glaucoma, early POAG group, moderate POAG group and group with preperimetric glaucoma. Complete ophthalmological examination, visual field and optic coherent tomography of peripapillar region of RNFL were performed. The collected data was entered into a specially created database on a personal account, and the statistical processing was done using the SPSS for Windows. Results are displayed through charts and tables.Results The results showed that the thickness of peripapillary RNFL in patients with mild POAG is lesser than in healthy subjects, and thickness in patients with moderate POAG is lesser than in patients with mild POAG, as well as in healthy subjects (59.69±10.63 μm vs 73.44±12.16 μm vs 105.57±11.34 μm). Thickness of peripapillary RNFL in patients with preperimetric glaucoma is significantly lesser than in healthy subjects (83.65±9.24 μm vs 105.57±11. 34 μm). Parameter S together with mean value of peripapillary RNFL thickness (AvgThic) is the best predictors of appearance and progression of preperimetric glaucoma. There is positive correlation between progression of glaucoma (MD value) and AvgThic. The best predictors of appearance and progression of glaucomatous disease are: AvgThic, RNFL thickness in quadrants- S, I, N; and parameters RNFL- Smax, Savg, Iavg. ROC curve has shown that the following parameters are bad markers for progression of the disease: RNFL thickness in quadrant T and Imax. Conclusions We concluded that the determination of thickness of peripapillary RNFL in

First group (control-healthy) : 30 patients without glaucoma or other eye conditions, with best corrected visual acuity ≥0.9, intraocular pressure (IOP) between 10mmHg and 21mmHg, normal cup-to-disc ratio (C/D) and normal visual field finding, regardless of gender, race and ethnic background.
Second group (early glaucoma): 30 patients with POAG, with characteristic defects of the optic disc and RNFL, with a mean deviation -2dB < (MD) <-6dB in standardized automated perimetry (Hodap classification), with characteristic glaucomatous visual field defects, without other eye conditions, without anamnestic data about previous laser or surgical intervention on the examined eye, with best corrected visual acuity ≥0.5, regardless of gender, race and ethnic background.
Third group (moderate glaucoma): 30 patients with POAG, with characteristic defects of the optic disc and RNFL, with a mean deviation (MD) lower than -6dB and higher than -12dB in standardized automated perimetry (Hodap classification), without other eye conditions, without anamnestic data about previous laser or surgical interventions on the examined eye, with best corrected visual acuity ≥0.5, regardless of gender, race and ethnic background.
The fourth group (pre-perimetric glaucoma): 30 patients with characteristic changes in the optic nerve head that represent glaucoma neuropathy, without functional outbreaks. The standard automated perimetry shows normal values of MD (from -2dB to + 2.0dB), with the best corrected visual acuity ≥ 0.9, regardless of the IOP. In our research descriptive statistics were used: arithmetic mean, standard deviation, median, quartiles, frequencies, and percentages. For the assessment of mean values of variables of two populations, the test for independent samples and the Man-Vitni test were used. An analysis of the variance and Kraskal-Volis test were used to compare the mean values of the variables of more populations. The correlation of categorical variables was examined using the χ2 test for the contour tables or by the Fischer test. The predictive quality of the variables on the outcome was evaluated using ROC curves. In all tests, the obtained level of statistical significance was expressed, statistically significant is considered to be p <0.05. The collected data was entered into a specially created database on a personal account, and the statistical processing was done using the SPSS for Windows. Results are displayed through charts and tables with a text commentary. Results 120 patients over the age of 18 were included in this study. Pathology of only one eye of the patient was analysed and followed. Based on the clinical findings and the inclusion criteria, four groups were formed. Each group encompassed an equal number of patients, thirty.
Gender distribution among patients is represented in table 1. The average age of the patients in the whole sample was 55.93±13.77 years. The youngest group was healthy group, then preperemetric group, then early POAG, and eldes was moderate POAG group. ANOVA showed that there is a statistically significant difference (p<0.001) in average age of the patients between groups.
The average values for each group: age of the patients, visual field parameters (MD, PSD) and RNFL quadrant thickness are presented in table 3. Table 3. Differences in mean values of age, MD, PSD and RNFL thickness by quadrants between groups 1 and 2, 1 and 3, as well as 1 and 4. p1-p-value between groups 1 and 2 (healthy and early POAG group) p2-p-value between groups 1 and 3 (healthy and moderate POAG group) p3-p-value between groups 1 and 4 (healthy and preperimetric glaucoma group) Other RNFL thickness values per quadrants are distributed between these two endpoints.    The ROC curve showed that the parameters: S, I, Smax, Savg, Iavg, AvgThic are good predictors for disease progression in patients with POAG. Other analysed parameters are poor pre-indicators. AvgThic is the best predictor for disease progression (p <0.0005).
Cut-off value for AvgThic is 63.945; Sensitivity is 67% and the specificity is 83.3%.

Discussion
Even though gender is not considered as a risk factor for developing POAG, Framigham, Barbados, Blue Mountains and other studies [10][11][12] have shown that a greater number of males than females suffer from POAG. Gender analysis of our 120 study participants shows that, in the whole sample, the majority of participants were females (60% Analysis of average values of the MD visual field parameter showed that a statistically significant decrease happens going from the healthy group to the moderate POAG group. Testing of differences between MD and PSD values among groups (ANOVA) and Post-Hoc (Tukey) analysis had shown that there is a statistically significant difference between the groups (p<0.001). This had shown that MD and PSD values completely separate groups 1, 2 and 3. The search of Sahli et al. [9] as well as some other's researchers [2,[13][14][15][16] have shown a high correlation between MD values and the degree of POAG progression.
The mean value of RNFL thickness for the healthy group in our study was 105.57±11.34μm, which was the highest value compared to other groups. The lowest value of RNFL thickness was in the moderate POAG group (59.6 9±10.63μm). RNFL thickness value decreases with the progression of POAG, which is confirmed by the statistical analysis of the RNFL thickness parameter (AvgThic) differences between groups (p<0.0005).
The results obtained by measuring RNFL thickness in quadrants (S,N,I,T) showed the same distribution in all study groups. RNFL thickness value was found to be the highest in the inferior quadrant, second highest in the superior quadrant, third highest in the nasal quadrant, while it was the lowest in the temporal quadrant. Also, it was shown that the rule of distribution of average RNFL thickness between groups is also applicable to RNFL thickness by quadrants. In all of the quadrants, the greatest RNFL thickness was found in the healthy group and the lowest in the moderate POAG group. Histologic studies done on enucleated eyes [17], as well as studies done on patients [18,8]  Mwanza et al. [20] reported that in initial POAG focal loss of RNFL thickness is in the lower quadrant, while Nouri-Mahdavi et al. [21] analysing OCT finding among healthy group and 59 patients from the early POAG group came to the conclusion that the best discriminating RNFL thickness parameter is the upper quadrant.
The possibility of detecting glaucoma at an early preperimetric stage was confirmed by the study of Sihota et al. [16], and it has particularly emphasized AvgThic as parameter indicating the onset of the disease. Guedes et al. [22] studied the ability of early detection of glaucoma by the OCT apparatus. They compared the changes that occur in the thickness of the macular zone and the peripapillary RNFL zone and concluded that in the competition of numerous parameters, the average thickness of RNFL is far the best in discriminating against patients with early glaucoma. As a cause of this, they hinted that there are almost 100% ganglia retinal cells in the peripapillary zone, and in the macular zone their number is about 50%, and the parameters of the thickness of RNFL are better for determining glaucoma than the parameters of the macular region. The area of ROC curve for AvgThic RNFL was 0.93 in this study, which was higher than the results obtained in our study (0.803).
Receiver Operating Characteristic (ROC) curves were designed to show which parameter could be a good predictor for disease progression, with which sensitivity and with which specificity. A study by Parikh et al. [23] showed that the best chance of discriminating against healthy glaucoma patients has a parameter RNFL thickness in inferior quadrant with sensitivity of 67% and specificity of 84%. The best parameter of the RNFL thickness group is AvgThic with a largest area of ROC curve (Area = 0.803), cut-off value 63.945, sensitivity 67%, and specificity 83.3% which are almost the same results as ours. Leite et al. [24] reported that the largest area below the ROC were at AvgThic, quadrant I and S.
Analyzing the ROC curves of Savini et al. [19], it is seen that AvgThic is the best marker for the occurrence of glaucoma with the largest area under the curve (Area = 0.725), then the quadrant S (Area = 0.672), the quadrant I (Area = 0.655) and the worst N (Area = 0.564) which is confirmed by the results of our study. Sahli et al. [9] also confirmed that the surface of the ROC is the highest in AvgThic and quadrant I (Area = 0.824, Area = 0.822). A slightly smaller area was in the case of quadrant S, but therefore high values were 77% sensitivity and 87% specificity.

Conclusion
Based on the results obtained in this research it can be concluded that the thickness of RNFL in patients with POAG and preperimetric glaucoma are statistically significantly of data, discussion results and conclusions.  Figure 1 Graphical presentation of the patient according to age groups Graphical presentation of ROC curves for the RNFL thickness parameters