Risk Factors for Normal-Tension Glaucoma Among Subgroups of Patients

Objective  To identify risk factors for normal-tension glaucoma among subgroups of patients.

Methods  In 93 patients with unilateral normal-tension glaucoma, intereye comparison of baseline spherical equivalent, central corneal thickness, untreated intraocular pressure, disc area, and zone β variables was performed among the following 4 subgroups classified according to age and visual field pattern standard deviation of the eye with glaucoma: subgroup 1 (age ≤50 years and visual field pattern standard deviation ≤8 dB), subgroup 2 (≤50 years and >8 dB), subgroup 3 (>50 years and ≤8 dB), and subgroup 4 (>50 years and >8 dB).

Results  Fourteen, 27, 30, and 22 patients were included in subgroups 1, 2, 3, and 4, respectively. The untreated intraocular pressure in subgroup 1 (P = .005), the zone β variables in subgroup 2 (P < .001), and both the untreated intraocular pressure (P = .010 and P = .034, respectively) and the zone β variables (P ≤ .008 and P ≤ .006, respectively) in subgroups 3 and 4 were significantly greater in the eyes with glaucoma than in the normal contralateral eyes (by paired t test or Wilcoxon signed rank test). The other variables showed no significant difference between the eyes in any subgroup.

Conclusion  The zone β variables (and not the untreated intraocular pressure) may represent significant risk factors in young patients having normal-tension glaucoma with moderate to severe visual field loss.

Normal-tension glaucoma (NTG) is defined as chronic open-angle glaucoma with a progressive visual field (VF) defect and optic nerve damage, despite an untreated intraocular pressure (IOP) that always measures within the statistically normal range (mean [SD], 15.5 [2.6] mm Hg).¹ Patients having NTG with asymmetric IOP had a tendency for greater VF loss in the eye with higher IOP in previous studies,^2-4 and the neuroretinal rim area of the optic disc has been reported to decrease with increasing IOP in patients having NTG.^5,6 However, a recent prospective study⁷ reported that IOP asymmetry was unrelated to VF asymmetry in patients having NTG, suggesting an unclear pathogenic relationship between IOP and glaucomatous damage in eyes with NTG. In addition, IOP reduction does not seem to always prevent progression of NTG, although it is beneficial to NTG management.⁸ Furthermore, IOP was not reported to be significantly associated with progression in untreated eyes in a previous randomized clinical trial.⁹ Therefore, there are likely other risk factors associated with NTG in addition to IOP that are clinically important but not yet identified. The determination of additional risk factors would likely aid in the treatment of these patients, which now focuses on IOP reduction. However, there is debate about the influence of possible additional risk factors other than IOP, such as the zone β variables of peripapillary atrophy (PPA),^10-17 central corneal thickness (CCT),^18-30 disc size,^31-36 and myopia,^27,37-45 on the development or progression of NTG.

Because NTG is an arbitrarily defined disease entity and is considered to be due to multifactorial influences such as mechanical, vascular, structural, metabolic, autoimmune, or genetic origins (like chronic open-angle glaucoma), patients with NTG likely comprise separate subgroups with different pathogenic mechanisms. Therefore, the objectives of the present study were (1) to identify by intraindividual intereye comparisons the possible risk factors (baseline spherical equivalent, CCT, IOP, disc size, and the zone β variables) in patients having NTG with unilateral VF loss and (2) to assess the different risk factors among patients with NTG by dividing them into subgroups according to age and severity of VF loss.

A review was performed of consecutive patients who had been newly diagnosed as having unilateral NTG from December 1, 2004, through May 31, 2007, at Samsung Medical Center. All patients underwent a complete baseline ophthalmic examination of both eyes, including gonioscopy, manifest refraction, slitlamp biomicroscopy, Goldmann applanation tonometry, dilated stereoscopic examination of the optic disc and fundus, standard automated perimetry (Humphrey field analyzer; Carl Zeiss Meditec, Dublin, California), ultrasonic pachymetry (Humphrey ultrasonic pachometer model 850; Humphrey Instruments Inc, San Leandro, California), and stereometric analysis of the optic nerve head and PPA using the Heidelberg retina tomograph (HRT II; Heidelberg Engineering, Heidelberg, Germany).

The diagnosis of NTG required open iridocorneal angles by gonioscopy and glaucomatous VF defects on Humphrey 30-2 standard automated perimetry. A VF test was considered abnormal if 2 of the following 3 criteria were met on at least 2 consecutive VF examinations with acceptable reliability standards (fixation loss <20%, false-positive rate <33%, and false-negative rate <33%): (1) an abnormal glaucoma hemifield test result (borderline findings were not considered abnormal), (2) at least 3 contiguous nonedge points (allowing 2 nasal step edge points) on Humphrey 30-2 standard automated perimetry with P < .05 on the pattern SD plot and at least 1 point with P < .01, and (3) a corrected pattern SD with P < .05. The location and pattern of the defect had to be consistent between the 2 consecutive VF examinations, and the glaucomatous optic disc damage had to be consistent with the VF abnormality. Global indexes of the VF were averaged from the 2 baseline fields for statistical analysis. Baseline applanation tonometry was performed in the morning (9 AM to 12 PM) on one examination day and in the afternoon (1-5 PM) on another day with less than a 1-month interval, both after topical medication washout. All IOP measurements were obtained by one glaucoma specialist (C.K.) using a slitlamp-mounted Goldmann applanation tonometer and with the patient sitting; the mean of 3 measurements was recorded each time. The washout period was 4 weeks for β-blocker and prostaglandin analogue therapy, 2 weeks for adrenergic and carbonic anhydrase therapy, and 1 week for pilocarpine therapy. Untreated IOP was adjusted by 2.5 mm Hg for every 50 μm that the CCT deviated from 530 μm,⁴⁶ and this adjusted untreated IOP was analyzed to minimize the confounding effect of CCT on IOP. All patients had untreated IOP readings (unadjusted and adjusted) of 21 mm Hg or lower at all times as a requirement to be included in this study; the mean value of 2 baseline IOP readings, in the morning and in the afternoon, was obtained for statistical analysis.

The contralateral eye for each patient had a normal optic disc appearance and VF with open iridocorneal angles by gonioscopy. Patients were excluded if they had a history of angle closure, ocular trauma, corneal opacity, laser iridotomy, inflammatory eye disease, diabetic retinopathy or other retinal disease, untreated IOP (unadjusted or adjusted) exceeding 21 mm Hg, nonglaucomatous optic neuropathy or other neuroophthalmic disease, argon laser trabeculoplasty performed less than 6 months previously or for an untreated IOP exceeding 21 mm Hg, or ocular surgery other than uncomplicated cataract surgery with posterior chamber lens implantation performed more than 1 year before the first baseline VF test in either eye. Patients with pseudoexfoliation, pigment dispersion, tilted optic disc in either eye, consistently unreliable VFs, or best-corrected visual acuity of less than 20/40 were also excluded. Patients with any of the following criteria underwent neuroimaging, carotid Doppler examination, or orbital B-scan ultrasonography to rule out nonglaucomatous causes of disc and VF changes: neurologic visual abnormalities, new onset or increased severity of headaches, and localizing neurologic symptoms other than migraine. Patients with abnormal neuroimaging, carotid Doppler, or orbital B-scan results were excluded from the study.

The methods applied in the study adhered to the tenets of the Declaration of Helsinki for the use of human subjects in biomedical research, and review of the patients' medical records was performed after institutional review board approval, noting their baseline age, CCT, spherical equivalent, untreated IOP (the mean of morning and afternoon baseline IOP readings), and the VF mean deviation (MD) and pattern standard deviation (PSD) (the mean of 2 consecutive baseline VF results). The disc area was obtained from the Heidelberg retina tomograph results, which were obtained at the time of the first baseline VF test. The zone β area and the angular extent of zone β around the optic disc were measured by drawing a new contour line on the baseline image of the optic nerve head taken by the Heidelberg retina tomograph, similar to the methods described in detail elsewhere.¹² Zone β was defined as the central zone of PPA characterized by chorioretinal atrophy with visible large choroidal vessels and sclera. The peripheral zone (zone α) of PPA with irregular hyperpigmentation and hypopigmentation and a peripheral scleral ring (a thin white band of tissue surrounding the optic nerve head) was not included in zone β. Using these data, intraindividual intereye comparisons of the baseline spherical equivalent, CCT, untreated IOP, disc area, and zone β variables (zone β area, angular extent of zone β, and zone β area to disc area ratio) between the eyes with glaucoma and the normal contralateral eyes were performed in all patients and in each of the 4 subgroups classified according to age and the VF PSD in the eyes with glaucoma as follows: subgroup 1 (age ≤50 years and VF PSD ≤8 dB), subgroup 2 (≤50 years and >8 dB), subgroup 3 (>50 years and ≤8 dB), and subgroup 4 (>50 years and >8 dB). Then, correlation analyses between the intereye differences for each possible risk factor and the intereye differences for the VF global indexes were performed. In addition, correlation analyses between each possible risk factor and the VF global indexes were performed for the eyes with glaucoma (the Δ symbol indicates the intereye difference between the eye with glaucoma and the normal contralateral eye).

For the intraindividual intereye comparisons among all patients and in each of the subgroups, paired t test or Wilcoxon signed rank test was performed using commercially available statistical software (SPSS version 11.0 for Windows; SPSS Inc, Chicago, Illinois). A correlation was expressed in terms of the univariate Pearson product moment correlation or Spearman rank correlation coefficients and P values. Bonferroni correction for multiple testing was applied for the possibly related variables, and the level of significance was P < .05 (2-sided) for all statistical tests. The statistical power analysis revealed for sample sizes of 93 (total cohort), 14 (subgroup 1), 27 (subgroup 2), 30 (subgroup 3), and 22 (subgroup 4) that intereye differences of approximately 0.29, 0.81, 0.56, 0.53, and 0.63 SDs would have been detected, respectively (with a significance level of P < .05 and a statistical power of 80%).

Ninety-three patients with unilateral NTG (48 women and 45 men) having a mean (SD) age of 52.1 (12.6) years (age range, 27-89 years; median age, 53.0 years) were included in the study; all of the study patients were Korean. The mean (SD) VF MD and PSD were −5.73 (4.15) dB (range, −16.25 to 1.42 dB; median, −4.58 dB) and 9.08 (4.60) dB (range, 1.37-16.84 dB; median, 8.12 dB), respectively, for the eyes with glaucoma and −0.45 (1.14) dB (range, −3.32 to 1.75 dB; median, −0.36 dB) and 1.97 (0.46) dB (range, 1.22-3.56 dB; median, 1.90 dB), respectively, for the normal contralateral eyes. The untreated IOP and the zone β variables were significantly greater for the eyes with glaucoma vs the normal contralateral eyes, whereas the baseline spherical equivalent, CCT, and disc area were not significantly different between the 2 eyes (Table 1). In addition, the Δzone β variables were significantly correlated with the ΔMD and the ΔPSD, and, after Bonferroni correction was applied, the Δzone β area and the Δzone β area to disc area ratio were significantly correlated with the ΔPSD (Table 2). In the eyes with glaucoma, the zone β area to disc area ratio had a significant negative correlation with the VF MD, and both the zone β area and the zone β area to disc area ratio had a significant correlation with the VF PSD (Table 3). However, no significant correlation was found between the untreated ΔIOP and the ΔVF global indexes (Table 2) or between the untreated IOP and the VF global indexes in the eyes with glaucoma (Table 3). There was no significant relationship between the intereye differences for any of the other possible risk factors evaluated (baseline Δspherical equivalent, ΔCCT, and Δdisc area) and the intereye differences for the VF global indexes (ΔMD and ΔPSD) or between any of the other risk factors assessed (baseline spherical equivalent, CCT, and disc area) and the VF global indexes (MD and PSD) in the eyes with glaucoma.

In Tables 2 and 3, the relationship between age and the ΔVF global indexes or the VF global indexes in the eyes with glaucoma was also evaluated. Age had a significant negative correlation with the ΔPSD and with the PSD in the eyes with glaucoma, although age was not significantly correlated with any of the Δzone β variables or with the zone β variables in the eyes with glaucoma. Even after considering nondefinitive onset of glaucoma among the patients, it is still unusual that age had a significant negative correlation with the PSD in the eyes with glaucoma and with the ΔPSD in the population of 93 consecutively seen patients with unilateral NTG. Because improvement of glaucomatous VF loss (represented as a decrease in the PSD) is unlikely with increasing age owing to the progressive nature of glaucoma and because NTG is considered to have a multifactorial pathogenesis, patients with NTG likely comprise separate subgroups with different pathogenic mechanisms underlying their NTG. Therefore, a subgroup analysis was performed after classifying all of the patients into 4 subgroups as described in the “Methods” section.

Fourteen, 27, 30, and 22 patients with unilateral NTG were included in subgroup 1 (age ≤50 years and VF PSD ≤8 dB), subgroup 2 (≤50 years and >8 dB), subgroup 3 (>50 years and ≤8 dB), and subgroup 4 (>50 years and >8 dB), respectively. The mean (SD) ages were 40.4 (4.0) years (age range, 35-49 years), 39.9 (6.6) years (age range, 27-50 years), 61.8 (7.8) years (age range, 52-89 years), and 61.2 (6.6) years (age range, 52-73 years) in subgroups 1, 2, 3, and 4, respectively; there were no significant age differences between subgroups 1 and 2 (P = .809 by independent t test) or between subgroups 3 and 4 (P = .933 by Mann-Whitney test).

Table 4 and Table 5 give the intereye comparisons of the possible risk factors for NTG and the VF global indexes between the eyes with glaucoma and the normal contralateral eyes in each subgroup. In subgroup 1, only unadjusted and adjusted untreated IOPs were significantly greater in the eyes with glaucoma than in the normal contralateral eyes, whereas in subgroup 2 only the zone β variables were significantly greater in the eyes with glaucoma than in the normal contralateral eyes. In subgroups 3 and 4, both the untreated IOP and the zone β variables were significantly greater in the eyes with glaucoma than in the normal contralateral eyes. The baseline spherical equivalent, CCT, and disc area did not show significant differences between the 2 eyes in any of the subgroups. The statistical significance of these results did not change after Bonferroni correction was applied except for the significance of the intereye comparison of the unadjusted, untreated IOP in subgroup 4, which approached significance.

In addition, the Δzone β area and the Δzone β area to disc area ratio had a significant positive correlation with the ΔPSD only in subgroup 2 (Table 6 and Figure 1). No significant relationship was found between the untreated ΔIOP and the ΔPSD in any of the subgroups. In the eyes with glaucoma, the zone β area and the zone β area to disc area ratio were significantly correlated with the PSD only in subgroup 2 (the eTable [http://www.archophthalmol.com] and Figure 2). No significant relationship was found between the untreated IOP and the PSD in any of the subgroups. In addition, age no longer had a significant negative correlation with the ΔPSD (Table 6) or with the PSD in the eyes with glaucoma (the eTable) in any of the subgroups. There was no significant association between the intereye differences for any of the other possible risk factors studied (baseline Δspherical equivalent, ΔCCT, and Δdisc area) and the intereye differences for the PSD (ΔPSD) or between any of the other risk factors assessed (baseline spherical equivalent, CCT, and disc area) and the PSD in the eyes with glaucoma.

In the present study, the untreated IOP and the zone β variables were significantly greater in the eyes with glaucoma compared with the normal contralateral eyes in patients with unilateral NTG. These risk factors differed among the subgroups of patients with unilateral NTG: the untreated IOP in subgroup 1, the zone β variables in subgroup 2, and both the untreated IOP and the zone β variables in subgroups 3 and 4 were significantly greater in the eyes with glaucoma than in the normal contralateral eyes. These results suggest that in young (≤50 years) patients having NTG with moderate to severe VF loss (PSD >8 dB) an IOP-independent zone β–related pathogenic mechanism contributes to NTG and that in young patients having NTG with mild VF loss (PSD ≤8 dB) an IOP-dominant pathogenic mechanism underlies NTG. Because intereye differences were evaluated in patients with unilateral NTG, the present study could eliminate many of the confounding factors commonly encountered in control groups such as differences in age, sex, and health status. The contribution of zone β–related pathogenic mechanisms in our young patients having NTG with moderate to severe VF loss was further supported by the significant correlation between the Δzone β variables (Δzone β area and Δzone β area to disc area ratio) and the ΔPSD, as well as by the correlation between the zone β variables (zone β area and zone β area to disc area ratio) and the PSD in the eyes with glaucoma found only among subgroup 2 (Table 6 and the eTable). The significant negative correlation between the ΔPSD (or the PSD in the eyes with glaucoma) and age among the total cohort (Tables 2 and 3) might be due to the heterogeneity owing to different pathogenic mechanisms in our patients having NTG because no such correlation was found in any of our subgroups (Table 6 and the eTable).

Our results on the relationship between NTG and the untreated IOP are consistent with the findings of prior studies^2-4 that showed greater VF loss in the eye with higher IOP among patients with NTG; this finding was reproduced in our study after the untreated IOP was adjusted based on an eye's CCT. However, our IOP results were inconsistent with the findings of a recent prospective clinical trial⁷ that did not demonstrate a relationship between the untreated IOP asymmetry and VF asymmetry in patients with NTG. The reason for this inconsistency is unclear; it is possible that the population with NTG is not a homogeneous group of patients with glaucoma and that the multifactorial nature results in varying proportions of patients with pathogenesis caused by IOP. That is, IOP asymmetry becomes significantly associated with VF asymmetry if a sufficiently large proportion of the study population has an IOP-dominant pathogenic mechanism underlying their NTG and vice versa.

There were no significant correlations between the untreated ΔIOP and the ΔVF global indexes or between the untreated IOP and the VF global indexes among all patients and within each subgroup, despite the fact that the untreated IOP was significantly greater in the eyes with glaucoma compared with the normal contralateral eyes among all patients and within each subgroup except for subgroup 2. These results might be explained by the likelihood that patients with NTG have a wide range of optic nerve head susceptibility to IOP.

Peripapillary atrophy was found to be an IOP-independent risk factor for open-angle glaucoma progression in previous studies.^47,48 In addition, the prevalence and extent of PPA were reported to be significantly greater in NTG than in ocular hypertension in an earlier study,¹⁰ and the area of PPA in NTG was significantly larger than that of normal eyes in another study.¹¹ Furthermore, the zone β of PPA was shown to be significantly associated with functional and structural optic nerve damage in NTG,¹² and PPA was reported to be a significant risk factor for progression of VF damage in NTG.^13,14 Although the association of PPA with glaucoma has not been confirmed in some studies,^15,16 the association of zone β with the presence and amount of glaucomatous optic neuropathy has been confirmed in a population-based study.¹⁷ The present study clearly showed that the zone β of PPA was greater in the eyes with glaucoma than in the normal contralateral eyes and was associated with the amount of glaucomatous damage in the patients with unilateral NTG. The zone β variables failed to show a significant difference between the 2 eyes among the patients with unilateral NTG only in subgroup 1; the reason for this result requires further investigation.

In the present study, the PSD was used as a representative measure of the VF loss for classifying the patients into subgroups not only because age had a significant correlation with the ΔPSD and the PSD (and not with the ΔMD or the MD) but also because the PSD is less affected by cataracts than the MD.⁴⁹ The PSD begins to fall despite worsening glaucomatous damage if the damage is very extensive (MD range, less than −17 to −18 dB)⁵⁰; however, there was no MD value in the present study that was worse than −17 dB. In our study, the PSD was strongly correlated with the MD in the eyes with glaucoma (ρ = −0.84 and P < .001, Spearman rank correlation analysis), and an 8-dB PSD corresponded to approximately a −4.60-dB MD by linear regression analysis. Patients were subgrouped according to an age 50-year cutoff in the present study because NTG is unusual in patients 50 years or younger.¹ The 8-dB PSD and age 50-year cutoff are rough medians among the eyes with glaucoma (median, 8.12 dB and 53 years). The statistical significance of the analyses using the standardized cutoffs based on median split was similar to that of our original results.

This study has the following limitations. Our untreated IOP data, which were obtained in the morning (9 AM to 12 PM) on one day and in the afternoon (1-5 PM) on another day, with less than a 1-month interval, partially but not completely represent the diurnal IOP curve. This may limit the significance of our analyses among the eyes with glaucoma with regard to the correlation between the untreated IOP and the VF global indexes, which are summarized in Table 3 and the eTable. However, this does not limit the significance of the intereye comparisons of the untreated IOP (Tables 1, 4, and 5) or of the correlation analyses between the untreated ΔIOP and the ΔVF global indexes (Tables 2 and 6) because the diurnal curves of the untreated IOP in both eyes of the patients with NTG demonstrate a parallel course⁵¹ and because the mean, peak, and trough untreated IOP values are highly correlated between the 2 eyes among patients with NTG.⁵² We did not routinely perform neuroimaging for patients 50 years or younger unless they had headaches, focal neurologic symptoms, or neurologic visual abnormalities. However, no patient has developed neurologic signs or symptoms warranting further consultation after a mean of 2 years of follow-up. The statistical power analysis revealed for sample sizes of 14 (subgroup 1), 27 (subgroup 2), 30 (subgroup 3), and 22 (subgroup 4) that intereye differences of approximately 0.81, 0.56, 0.53, and 0.63 SDs would have been detected, respectively. Therefore, clinical significance might have been hidden in statistical nonsignificance and vice versa, especially for the variables for which intereye difference was not markedly large. The application of our results to patients with NTG in general might be limited not only because patients with tilted optic disc were excluded from our study but also because it is unclear whether unilateral NTG and bilateral NTG are the same disease entity.

In conclusion, in our study the untreated IOP and the zone β of PPA (zone β area, angular extent of zone β, and zone β area to disc area ratio) were significant risk factors for NTG in the intereye comparative analysis of patients with unilateral NTG. For young patients (≤50 years) having NTG with mild VF loss (PSD ≤8 dB), the untreated IOP (and not the zone β variables) was frequently found to be a risk factor for NTG, whereas IOP-independent zone β–related risk factors were found for young patients having NTG with moderate to severe VF loss (PSD >8 dB). In addition, the zone β area and the zone β area to disc area ratio were significantly associated with both the presence and severity of NTG only in young patients having NTG with moderate to severe VF loss among the subgroups. Therefore, treatment strategies other than IOP-lowering medications should be considered for young patients having NTG with moderate to severe VF loss. Further studies on the significance of zone β of PPA in the pathogenesis of NTG are needed.

Correspondence: Changwon Kee, MD, PhD, Department of Ophthalmology, Samsung Medical Center, Sungkyunkwan University School of Medicine, 50 Irwon-Dong, Gangnam-Gu, Seoul 135-710, Korea (ckee@skku.edu).

Submitted for Publication: August 21, 2008; final revision received April 19, 2009; accepted May 18, 2009.

Financial Disclosure: None reported.