The descriptive characteristics of subjects with the three types of OAG are shown in . Out of 607 participants, 550 (90.6%) were newly diagnosed with POAG, 28 (4.6%) with PIGM, and 29 (4.8%) with PEXG. The mean ages of the subjects in these three groups differed significantly (p<0.0001). POAG subjects (58.0 years) were nine years older on average than those with PIGM (48.9 years) and seven years younger on average than those with PEXG (65.1 years). The distribution of males and females among the three types of glaucoma did not differ significantly (p=0.335). 41.5% (228) of POAG subjects indicated their race was black, whereas only 7.1% of PIGM subjects and 3.5% of PEXG subjects reported that their race was black (p<0.0001). Educational achievement varied somewhat between the groups (p=0.055), with the POAG subjects having the highest percentage (22.6%) with less than a high school education relative to PIGM subjects (3.6%) and PEXG subjects (10.3%). Follow-up time did not significantly differ for the POAG (7.2 years), PIGM (7.2 years), and PEXG (7.7 years) subjects (p=0.447).
Descriptive characteristics of subjects with the three types of OAG.
Those with newly diagnosed POAG tended to have more non-ocular co-morbidities than the other two OAG subtypes. Diabetes was found significantly more frequently (p=0.014) among subjects with POAG (18.2%) vs. subjects with PIGM (3.6%) or PEXG (3.5%). 38.6 % (212) of POAG subjects had systemic hypertension, whereas 21.4% of PIGM subjects and 24.1% of PEXG subjects had hypertension (p=0.063). The percentages of subjects with other vascular or cardiac diseases did not significantly differ among the three groups (p=0.376). In terms of family history of glaucoma, 7.7 % of PIGM subjects had a history of glaucoma within the immediate family whereas 38.8% of POAG and 27.3% of PEXG subjects reported this (p=0.002). The distribution of history of glaucoma in the distant family among subjects with the three types of glaucoma did not differ significantly (p=0.268), nor did smoking status (p=0.222).
Ophthalmic examination findings showed some significant differences among the three OAG subtypes. The mean IOPs at baseline of POAG, PIGM, and PEXG participants’ study eyes were 27.3, 28.1, and 31.9 mmHg, respectively (p<0.0001). Post hoc pairwise comparisons showed a significantly higher mean IOP in those with PEXG relative to either of the other OAG diagnoses, but no difference between POAG and PIGM. Those diagnosed with PIGM were significantly (p<0.0001) more myopic on average (spherical equivalent mean value of −3.81 diopters, D) than the other two groups, whose mean values were −0.87 D (POAG) and −0.15 D (PEXG). The mean visual acuities at baseline, results from Humphrey 24–2 visual field testing, vertical cup to disc ratio, and the presence of disc hemorrhage among the three types of glaucoma were not significantly different. In terms of bilaterality, those with PEXG were more likely to present at diagnosis with only one eye involved (p<0.0001).
The multinomial logistic regression model results () identified four factors that were significantly associated with a diagnosis of PIGM versus a diagnosis of POAG: age, race, history of glaucoma in the immediate family, and spherical equivalent. Relative to POAG, those with PIGM were younger (odds ratio (OR)=0.44 for a 10-year increment in age), more likely to be white (OR=13.70), less likely to have an immediate family history of glaucoma (OR=0.12), and more likely to have a more negative (myopic) spherical equivalent value (OR= 0.77 for a 1 D increment).
Six factors were significantly associated with having a diagnosis of PEXG relative to POAG: age, sex, race, visual acuity, IOP, and bilaterality. Relative to POAG, CIGTS enrollees with newly diagnosed PEXG were significantly older (OR=3.61 for a 10 year increment), more likely to be female (OR=7.57), and more likely to be white (OR=8.01). PEXG subjects had higher IOP than POAG subjects (OR = 2.69 for a 5 mmHg increment), and their visual acuity at baseline was somewhat better (OR=2.16 for a 5-letter increment). PEXG subjects were less likely to have bilateral disease relative to those with POAG (OR=0.20).
In two binary logistic regression models (PIGM vs. POAG and PEXG vs. POAG) where non-significant effects were stepped-out, multi-variable results showed similar effects to that of the multinomial logistic regression results. Univariable results were also similar for all but the gender effect. This effect was weakened due to its collinearity with baseline IOP (r=−0.19) – males had significantly higher baseline IOP than females. See supplementary online Table 1
for a comparison of effects between the multinomial logistic regression, multi-variable binary logistic regression, and univariable binary logistic regression results.
The relationship of glaucoma diagnosis with two treatment outcomes (MD and IOP) was evaluated over seven years of follow-up. Boxplots of these two outcomes over time by diagnosis are shown in and . The association of glaucoma diagnosis with these outcomes was investigated using repeated measures linear regression. Glaucoma diagnosis was added to previously published models of baseline risk factors(13, 14)
to determine if diagnosis added predictive strength. Glaucoma diagnosis was not associated with MD over time (p=0.627). There were no significant interactions between diagnosis and either treatment or time. Glaucoma diagnosis was also not associated with IOP measures during follow-up (p=0.310), with no significant interactions between diagnosis and treatment or time. Since baseline IOP was significantly higher in those with PEXG than in POAG or PIGM, we evaluated and found a significant interaction (p=0.030) between baseline IOP and glaucoma diagnosis on follow-up IOP. The interaction resulted from differing trends in IOP reduction over time relative to baseline IOP in patients with PEXG, in whom the higher the baseline IOP, the lower the follow-up IOP, whereas in patients with POAG and PIGM, baseline IOP and follow-up IOP were directly related.