Distribution of Ocular Perfusion Pressure and Its Relationship with Open-Angle Glaucoma: The Singapore Malay Eye Study

Abstract

Purpose.: To describe the distribution of ocular perfusion pressure and its relationship with open-angle glaucoma (OAG) in a Malay population.

Methods.: This was a population-based, cross-sectional study comprising 3280 (78.7% response) ethnic Malays. Intraocular pressure (IOP) was measured with Goldmann applanation tonometry. Systolic and diastolic blood pressure (SBP and DBP) was measured with a digital automatic blood pressure monitor. Mean ocular perfusion pressure (MOPP) = ⅔(mean arterial pressure − IOP), where mean arterial pressure (MAP) = DBP + ⅓(SBP − DBP), systolic perfusion pressure (SPP) = SBP − IOP, and diastolic perfusion pressure (DPP) = DBP − IOP, was calculated. The diagnosis of OAG was based on International Society for Geographical and Epidemiologic Ophthalmology criteria.

Results.: A total of 3261 persons (mean age, 58.7 ± 11 years, including 131 [4.0%] cases of OAG) were available for analyses. Among persons without glaucoma, the mean ± SD IOP, MOPP, SPP, and DPP were 15.3 ± 3.5, 52.8 ± 9.3, 131.5 ± 23.3, and 64.5 ± 11.3 mm Hg, respectively. Among persons with OAG, the corresponding values were 16.8 ± 5.9, 51.6 ± 10.2, 134.5 ± 24.6, and 61.4 ± 11.5 mm Hg, respectively. In multiple logistic regression models adjusting for IOP, age, sex, and IOP- and BP-lowering treatments, OAG risk was significantly higher in participants with DBP, MOPP, or DPP in the lowest quartile (Q1) than in participants in the highest quartile (Q4) (Q1 vs. Q4: odds ratio [OR], 1.71 [95% confidence interval (CI), 1.04–2.96] for DBP; OR, 1.73 [95% CI, 1.05–3.15] for MOPP; OR, 1.75 [95% CI, 1.02–3.01] for DPP).

Conclusions.: Low DBP, low MOPP, and low DPP are independent risk factors for OAG in ethnic Malays, providing further evidence of a vascular mechanism in glaucoma pathogenesis across different populations.

A vascular mechanism has been postulated in the pathogenesis of glaucomatous optic nerve damage ^1–4 ; specifically, an inadequate or unstable ocular blood supply causes ischemic damage and/or reperfusion injury to the optic nerve tissue and axons. This vascular concept is support by the reported associations of glaucomatous optic neuropathy with arterial hypertension, arterial hypotension, cardiovascular disease, migraines, vasospasm, and other circulation disorders. ^1–4  

Perfusion pressure, equal to mean blood pressure (BP) minus intraocular pressure (IOP), is an important determinant of ocular blood flow. ⁴ A decrease in perfusion pressure may significantly decrease the ocular blood flow in the absence of vascular autoregulation. ^3–4 Cross-sectional, population-based studies in Baltimore, ⁵ Bolzano, ⁶ southern Arizona, ⁷ and Rotterdam ⁸ and recently published prospective studies in Barbados ⁹ and Sweden ¹⁰ have shown that lower diastolic perfusion pressure (DPP), defined as diastolic blood pressure (DBP) minus IOP, is an independent risk factor for open-angle glaucoma (OAG), after adjustment for IOP and other risk factors. ^11–13 Clinic-based studies have further shown that nonphysiologic nocturnal BP and wider circadian fluctuation in mean ocular perfusion pressure (MOPP) are significantly associated with the development and progression of OAG. ^14,15 However, the relationship between perfusion pressure and glaucoma has not been well documented in Asians. ¹⁶ As both BP and IOP vary considerably in whites, blacks, and Asians, ^17–19 it is probable that the distribution and determinants of perfusion pressure and its relationship with glaucoma is different in Asians. 

In the Singapore Malay Eye Study (SiMES), ²⁰ we have reported that the prevalence of OAG in Malays 40 to 80 years of age was 2.5%, similar to that in white populations. ¹⁷ However, the mean IOP was 15.3 mm Hg, substantially lower than that in white populations. ¹¹ Furthermore, only 17% of those with OAG had IOP greater than 21 mm Hg. ²⁰ Thus, the identification of vascular risk factors may be of particular relevance to Asians. The purpose of this study was to describe the distribution of perfusion pressure and its relationship with OAG in Malays. 

Methods

Study Population

During 2004 to 2006, the SiMES examined 3280 (78.7% response) persons of Malay ethnicity aged 40 to 80 years. Study methods have been published. ^21–23 Ethics approval was obtained from the Institutional Review Board of the Singapore Eye Research Institute, Singapore, and the study was conducted in accordance with the World Medical Association's Declaration of Helsinki. Informed written consent was obtained from every participant. 

Measurement and Definition of Glaucoma

The measurement and definition of glaucoma have been described elsewhere. ^20,24,25 Before pupil dilation, Goldmann applanation tonometry (Haag-Streit, Köniz, Switzerland) was used to measure IOP once in each eye (right eye first). ²¹ Automated perimetry (SITA FAST 24-2, Humphrey Visual Field Analyzer II; Carl Zeiss, Feldbach, Switzerland) was performed with near refractive correction (1) on one in five consecutive participants without suspected glaucoma (n = 641 persons) before examination by study ophthalmologists and (2) on all participants with suspected glaucoma. Central corneal thickness (CCT) was measured with a handheld ultrasound pachymeter (Advent; Mentor O&O, Norville, OH). ²⁶ After pupil dilation, the optic disc was evaluated with a +78-D lens at ×16 magnification, and the vertical cup-to-disc ratio (VCDR) was determined. ²⁴  

Patients with suspected glaucoma all underwent visual field testing and were defined by any of the following criteria: (1) IOP greater then 21 mm Hg; (2) VCDR > 0.6 or VCDR asymmetry >0.2; (3) abnormal anterior segment deposit consistent with pseudoexfoliation or pigment dispersion syndrome; (4) occludable angle, defined as posterior trabecular meshwork seen for 180° or less of the angle circumference during static gonioscopy; (5) peripheral anterior synechiae or other findings consistent with secondary glaucoma; and (6) known history of glaucoma. 

Glaucoma cases were classified into three categories, according to ISGEO criteria ^20,25 : category 1, optic disc abnormality (VCDR/VCDR asymmetry ≥97.5th percentile or neuroretinal rim width between 11 and 1 o'clock or 5 and 7 o'clock <0.1 VCDR) with a corresponding glaucomatous visual field defect; category 2, a severely damaged optic disc (VCDR or VCDR asymmetry ≥99.5th percentile) in the absence of an adequate visual field test; category 3, no visual field or optic disc data, blindness (corrected visual acuity, <3/60), and prior glaucoma surgery IOP >99.5th percentile. For cases classified as category 1 or 2, there could be no other explanation for the VCDR finding (e.g., dysplastic disc or marked anisometropia) or visual field defect (e.g., branch retinal vein occlusion, macular degeneration, or cerebrovascular disease). 

In this study, OAG was diagnosed, after the exclusion of cases of angle-closure, rubeosis or secondary glaucoma (other than pseudoexfoliation). All the participants (including those without visual field data) who were judged not to have definitive glaucoma were classified as nonglaucomatous and were included in our analyses. 

Measurement and Definition of Ocular Perfusion Pressure

BP was measured with a digital automatic blood pressure monitor (Dinamap model Pro Series DP110X-RW, 100V2; GE Medical Systems, Inc., Milwaukee, WI), with the participant seated after 5 minutes of rest, according to a protocol similar to that used in the Multi-ethnic Study of Atherosclerosis. ²⁷ BP was measured twice, 5 minutes apart. A third measurement was made if systolic blood pressure (SBP) differed by more than 10 mm Hg or DBP differed by more than 5 mm Hg. ²¹ The mean between the two closest readings was then taken as the BP of that individual. Hypertension was defined as SBP ≥ 140 mm Hg or DBP ≥ 90 mm Hg or a diagnosis by physician. ²¹ Mean arterial BP (MAP) was calculated as DBP + ⅓(SBP − DBP). MOPP was defined as ⅔(MAP − IOP), systolic perfusion pressure (SPP) as SBP − IOP, and diastolic perfusion pressure (DPP) as DBP − IOP. For persons without glaucoma, the IOP of the right eye was used to calculate perfusion pressure, as there was no significant difference in IOP between the eyes. For persons with glaucoma, based on the approach described in the Barbados Eye Study, ⁹ the IOP was chosen based on the affected eye in unilateral cases and on the worse eye (based on a higher VCDR, in our study) in bilateral cases. Diabetes mellitus was defined as nonfasting glucose ≥11.1 mM, use of diabetic medication, or self-reported history of diabetes. ²⁸  

Statistical Analysis

R statistical software (R version 2.4.1; R-Project, available at http://cran.r-project.org) was used for data analyses. We first assessed the relationship between perfusion pressure and OAG by constructing logistic regression models adjusted for age, sex, IOP, and IOP- and BP-lowering treatments. CCT and the presence of diabetes were not significantly associated with OAG in univariate regression models (both P > 0.05), and thus they were not retained in multiple regression models. Perfusion pressures were analyzed either as continuous variables or as categorical variables, by using quartile cutoffs based on the distribution of study population. Given the inherent correlation between perfusion pressure and IOP, we examined the standard errors (in the form of confidence intervals [CIs]) and variance inflation factors (VIFs, measures of the degree of collinearity) for the regression model. A high VIF (close to 10) or large CIs indicate strong collinearity. Furthermore, we used a generalized additive model (GAM, based on the GAMLSS package in R), ²⁹ with the Loess smoothing function to determine the nonlinear relationship of perfusion pressure and IOP with the presence of OAG. Finally, we performed a strict power calculation according to the method described by Tosteson et al. ³⁰ The statistical power would be 0.80 if we made the following assumptions: prevalence of OAG = 4%, odds ratio (OR) for 1 SD decrease in DPP (as the exposure) = 1.3, OR for 1 SD increase in IOP (as the covariate) = 1.2, correlation between DPP and IOP = 0.20, sample size = 3261, and two-sided α = 0.05. On the basis of similar assumptions, the statistical power would be 0.19, for detecting the association between DPP and high-tension OAG (prevalence estimate = 0.6%, defined as OAG cases with IOP higher than the 97.5th percentile [21.5 mm Hg]). In addition, the statistical power for the study of OAG would decrease from 0.80 to 0.62, if we added an additional assumption that there was measurement error on DPP (assuming the correlation between measured DPP and true DPP = 0.8). The power would be further reduced to 0.40, if the correlation between measured and true DPP was assumed to be 0.6. 

Results

Among all 3280 participants, 8 cases of primary angle-closure glaucoma, 8 cases of nonspecific glaucoma, 2 cases of developmental glaucoma, and 1 case of rubeotic glaucoma were excluded, leaving 131 cases of definitive OAG and 3130 persons without glaucoma for further analyses. 

Among the 131 persons with OAG, the mean ± SD for IOP, SBP, DBP, MOPP, SPP, and DPP was 16.8 ± 5.9 (median, 16.0), 151.3 ± 24.5 (median, 150.5), 78.3 ± 10.6 (median, 77.0), 51.6 ± 10.2 (median, 52.0), 134.5 ± 24.6 (median, 135.0), and 61.4 ± 11.5 mm Hg (median, 61.0), respectively. The mean deviation (MD) and pattern SD (PSD) were −10.6 and 8.4 dB, respectively. Of the 131 patients, 94 (71.8%) had systemic hypertension (34/94 cases were on antihypertension treatment). Only nine patients had IOP-lowering treatment, probably because more than 90% of them had been undiagnosed. ²⁰  

Among the 3130 persons without glaucoma, the mean ± SD for IOP, SBP, DBP, MOPP, SPP, and DPP was 15.3 ± 3.5 (median, 15.0), 146.9 ± 23.7 (median, 144.0), 79.8 ± 11.2 (median, 78.0), 52.8 ± 9.3 (median, 51.6), 131.5 ± 23.3 (median, 128.5), and 64.5 ± 11.3 mm Hg (median, 63.0 mm Hg), respectively. Of these, 2138 (68.3%) had systemic hypertension (669/2138 were on antihypertension treatment). MOPP and SPP increased with age (both P < 0.001), whereas DPP was not related to age (P = 0.59, Fig. 1). Men tended to have higher MOPP and higher DPP, but lower SPP, than did women (t-test, both P < 0.05). 

Figure 1.

 

Distributions of SPP, DBP, and MOPP by age and sex. Data are derived from persons without glaucoma. Diamonds: SPP; triangles: DPP; circles: MOPP; filled symbols: men; open symbols: women; Error bar, 95% CI.

In multivariate logistic regression models treating the BP-related factors as categorical variables and added to the model, one at a time, lower DBP (OR, 1.71; 95% CI, 1.04–2.96, comparing persons in the lowest quartile with those in highest quartile [Q1 vs. Q4] of DBP), lower MOPP (OR, 1.73; 95% CI, 1.05–3.15; [Q1 vs. Q4]) and lower DPP (OR, 1.75; 95% CI, 1.02–3.01; [Q1 vs. Q4]) were significantly associated with OAG risk (Tables 1, 2). When we treated the BP-related factors as continuous variables and added them to the model one at a time, the multivariate-adjusted ORs were the same (OR, 1.22) for every 10-mm Hg decrease in DBP (P = 0.02), MOPP (P = 0.058), and DPP (P = 0.02). SBP, MAP, and SPP were not associated with OAG (all P > 0.05). Perfusion pressure and IOP were not collinear (the CIs for both variables were narrow, data not shown, and the VIFs for the models were ∼1.0–1.2; Tables 1, 2). 

Table 1.

 

Association of MOPP with OAG

	MOPP (mm Hg), Quartiles				P for Trend	VIF
	Q4	Q3	Q2	Q1
All persons
n cases/n at risk*	29/780	38/772	22/782	42/776
Values, mm Hg	>58	51–58	46–51	≤46
OR (95% CI) of OAG	1.00 (reference)	1.57 (0.93–2.64)	0.95 (0.53–1.72)	1.73 (1.05–3.15)	0.01	1.06
Age groups
Younger (<60 y)
n cases/n at risk*	7/432	11/420	7/440	16/423
OR (95% CI) of OAG	1.00 (reference)	1.58 (0.59–4.20)	1.09 (0.37–3.15)	2.51 (0.99–6.36)	0.04	1.06
Older (≥60 years)
n cases/n at risk*	15/354	26/344	23/346	26/351
OR (95% CI) of OAG	1.00 (reference)	1.95 (0.97–3.92)	1.79 (0.88–3.64)	1.55 (0.75–3.22)	0.09	1.04
Sex
Male
n cases/n at risk*	20/371	22/359	11/386	22/375
OR (95% CI) of OAG	1.00 (reference)	1.41 (0.74–2.71)	0.64 (0.29–1.41)	1.12 (0.56–2.27)	0.50	1.07
Female
n cases/n at risk*	9/408	15/404	13/405	19/402
OR (95% CI) of OAG	1.00 (reference)	2.28 (0.92–5.66)	2.03 (0.79–5.26)	3.26 (1.30–8.14)	0.007	1.07
IOP status
Low, <16 mm Hg
n cases/n at risk*	13/419	20/420	14/421	13/425
OR (95% CI) of OAG	1.00 (reference)	2.10 (0.97–4.51)	1.63 (0.72–3.68)	1.91 (0.82–4.47)	0.22	1.06
High, ≥16 mm Hg
n cases/n at risk*	17/355	17/358	10/364	27/348
OR (95% CI) of OAG	1.00 (reference)	1.11 (0.54–2.29)	0.55 (0.23–1.32)	1.60 (0.80–3.20)	0.05	1.06

 Data shown are ORs (95% CIs) adjusted for age (except for age-group-stratified analyses), sex (except for sex-stratified analyses), IOP (except for IOP-stratified analyses), IOP-lowering, and antihypertensive treatment.

*  Total number in each group may not be the same, because of missing value.

Table 2.

 

Association of DPP with OAG

	DPP (mmHg), Quartiles				P for Trend	VIF
	Q4	Q3	Q2	Q1
All persons
n cases/n at risk*	22/757	29/794	34/750	46/809
Values, mmHg	>71	63–71	56–63	≤56
OR (95% CI) of OAG	1.00 (reference)	1.30 (0.74–2.30)	1.57 (0.90–2.74)	1.75 (1.02–3.01)	0.002	1.08
Age groups
Younger, <60 y
n cases/n at risk*	6/418	9/426	10/437	16/434
OR (95% CI) of OAG	1.00 (reference)	1.38 (0.47–4.04)	1.70 (0.61–4.76)	2.55 (0.96–6.78)	0.02	1.10
Older, ≥60 y
n cases/n at risk*	16/352	22/345	22/323	30/375
OR (95% CI) of OAG	1.00 (reference)	1.44 (0.74–2.81)	1.45 (0.74–2.86)	1.48 (0.76–2.86)	0.07	1.07
Sex
Male
n cases/n at risk*	16/356	16/377	17/385	26/373
OR (95% CI) of OAG	1.00 (reference)	0.89 (0.43–1.83)	0.93 (0.46–1.89)	1.02 (0.52–2.02)	0.28	1.06
Female
n cases/n at risk*	7/388	14/408	15/402	20/421
OR (95% CI) of OAG	1.00 (reference)	2.10 (0.82–5.35)	2.31 (0.92–5.84)	2.53 (1.03–6.24)	0.02	1.06
IOP status
Low, <16 mm Hg
n cases/n at risk*	10/410	13/434	16/419	21/422
OR (95% CI) of OAG	1.00 (reference)	1.32 (0.57–3.06)	1.61 (0.71–3.64)	2.42 (1.10–5.31)	0.03	1.07
High, ≥16 mm Hg
n cases/n at risk*	12/340	17/362	17/373	25/349
OR (95% CI) of OAG	1.00 (reference)	1.46 (0.67–3.14)	1.31 (0.61–2.84)	1.46 (0.68–3.11)	0.07	1.09

 Data shown are ORs (95% CIs) adjusted for age (except for age-group-stratified analyses), sex (except for sex-stratified analyses), IOP (except for IOP-stratified analyses), IOP-lowering, and antihypertensive treatment.

*  Total number in each group may not be the same, because of missing value.

In a logistic regression model adjusting for age, sex, IOP, and IOP- and BP-lowering treatment, the OR of OAG was 0.61 (95% CI, 0.38–0.97; P = 0.02) in persons with systemic hypertension. The significant association remained (OR, 0.61; P = 0.048) when SBP was included in the model, but the significance disappeared (P > 0.05) when DBP or MBP was included. Including SBP, DBP, or MBP in the models did not improve the VIFs (range, 1.0–1.2). 

The association between DPP and OAG was stronger in younger persons (Q1 vs. Q4; OR, 2.55; P = 0.02) than in older persons (Q1 vs. Q4; OR, 1.48; P = 0.07), in lower IOP (Q1 vs. Q4; OR, 2.42; P = 0.03) than in higher IOP (Q1 vs. Q4; OR, 1.46; P = 0.07), and in women (Q1 vs. Q4; OR, 2.53; P = 0.02) than in men (Q1 vs. Q4; OR, 1.02; P = 0.28; Table 2). Similar associations were found between MOPP and OAG (Table 1). 

To explore the possible nonlinear relationship of DPP and IOP with OAG risk, we implemented a GAM model with the Loess smoothing function in R. ²⁹ The Akaike information criterion (AIC) was used to assist in selecting explanatory variable terms. The starting GAM model included IOP, DPP, age, sex, and IOP- and BP-lowering treatment as explanatory variables and prevalence of OAG as the responder. Any explanatory variable could be excluded if its exclusion would result in a better fit. In addition, every continuous explanatory variable (IOP, DPP, or age) could be fitted with either a linear function or a Loess smoothing function. As a result, the final best-fit GAM model (AIC = 1023.27, df = 6.73) for our data included IOP (Loess smoothing function, span = 0.9), DPP (linear function), age (linear function), and sex (binary factor) as explanatory variables. 

Discussion

In this population-based study of Asian Malays, we report three major findings. First, the mean systolic BP (146.9 mm Hg) was high, but the mean IOP (15.3 mm Hg) measured with applanation tonometry was similar to or lower than that that reported in whites or African Americans (15.4 mm Hg in the Beaver Dam Eye Study [BDES] and 18.0 mm Hg in the Barbados Eye Study [BES]). ^9,31 Thus, it is interesting to note that the mean DPPs are similar when compared in subjects from the SiMES (64.5 mm Hg) and the BES (63.3 mm Hg), ⁹ despite the dissimilarities in BP and IOP level between the two populations. Second, we showed that the presence of systemic hypertension was significantly protective against OAG (multivariate-adjusted OR, 0.61; P = 0.02), consistent with some, but not all, previous reports. ^6,13,32 Of interest, the significant association with systemic hypertension disappeared after adjustment for DBP or MAP, suggesting that DBP or MAP was a mediator between systemic hypertension and OAG. Finally, we showed that MOPP and DPP were significantly associated with OAG risk in Asian Malay persons, consistent with the findings from population-based studies in white, black, and Hispanic populations. ^5–9 Our study adds further support to the hypothesized vascular mechanism that lower ocular perfusion pressure results in a reduction in optic blood supply, contributing to defective autoregulation, and ultimately leading to glaucomatous optic damage. ^1–4 By definition, the variations in MOPP and DPP are largely explained by the variation in BP. If IOP alone is the major contributor to the association between MOPP/DPP and OAG, one would expect that the magnitude of associations would be stronger among persons with higher IOP. However, our subgroup analyses showed that this was not the case. 

As the effect of perfusion pressure on OAG risk was small, the value of single time-point measurement of perfusion pressure in clinical settings is uncertain. In the nonlinear model in which DPP was the only explanatory variable against the prevalence of OAG, there was a dramatic increase in OAG prevalence in persons with DPP below 50 mm Hg (Fig. 2, left), consistent with the graphs illustrated in the Baltimore Eye Survey, ⁵ the Egna-Neumarkt Study (in which prevalence of OAG began to rise when DPP became lower than 70 mm Hg), ⁶ and the Proyecto VER study. ⁷ However, when both DPP and IOP were included as explanatory variables in the nonlinear model (GAM), there was only a weak association of DPP with OAG prevalence (Fig. 2, right). The implication is that IOP remains the major risk factor for glaucoma, although DPP plays an independent role. Furthermore, in the Baltimore Eye survey, only 28.5% (45/158) of OAG cases had a DPP < 50 mm Hg. ⁵ In our study, only 32.1% (42/131) and 35.1% (46/131) of OAG cases fell within the lowest quartile of MOPP (≤46 mm Hg) and DPP (≤56 mm Hg), respectively. These results further indicate that multiple mechanisms most likely contribute to glaucomatous optic neuropathy. ³³ For some cases, decreased perfusion is a contributing factor, whereas others may have other mechanisms playing a role such as weaker supporting tissue or unhealthy astroglia, exacerbated by elevated IOP. ³⁴  

Figure 2.

 

Nonlinear relationship of DPP and IOP with the percentage of subjects with OAG. When DPP is the only explanatory variable, the prevalence of OAG increases exponentially at a lower DPP (left). When both DPP and IOP are the explanatory variables, the prevalence increases mildly with decreasing DPP (right). Data on the left are fitted with a Loess line; data on the right are derived from the GAM, fitting the Loess smoothing function for IOP.

In contrast to our findings, the BMES in white Australians found that a higher SPP was associated with OAG risk (OR, 1.09; P = 0.05) cross-sectionally, although a higher DPP was associated with reduced risk of ocular hypertension (OR, 0.78; P = 0.0008). ³² Possible explanations for the difference in findings include errors in the measurement of BP or IOP in this study or the BMES and higher rates of treatment of hypertension in the BMES population (70.0% in BMES vs. 31.5% in SiMES). ³² A true association between BP and IOP may have been removed due to treatment benefits. Our findings were also in contrast to those reported by the Beijing Eye Study in the Chinese, in which no significant association was found between perfusion pressure and OAG. ¹⁶ The discrepancy between our findings and those from the Beijing Eye Study may be attributed to differences in the definition of glaucoma (the Beijing Eye Study relied only on optic nerve head appearance for diagnosing glaucoma ¹⁶ ), genetic background, and/or unmeasured lifestyle factors. 

Our subgroup analyses suggested that the association between DPP and risk of OAG was stronger in women (OR, 2.53; Q1 vs. Q4) than in men (OR, 1.02; Q1 vs. Q4). Such a difference between the sexes was also seen in the relationship between MOPP and OAG risk. These seemed to be consistent with the findings in white populations that vascular dysregulations occur more often in women and that normal-tension glaucoma is more common in women than in men. ³ We also examined the relationship between antihypertension treatment and OAG risk and found no significant association between the two (data not shown), consistent with the finding reported from the BES. ⁹ Given the previous observations that calcium channel antagonists use may increase the risk of OAG ³⁵ and that systemic β-blocking agents may be protective, ³⁶ further studies with larger sample size or pooling analysis are needed to confirm the role of specific antihypertension treatment. 

This study's strengths include a population-based study design, standardized protocol, a comprehensive assessment of BP-related risk factors, and parametric and nonparametric statistical analyses. Limitations should also be highlighted: First, due to logistic constraints in the population-based setting, both BP and IOP were measured during one visit and thus the effect of circadian perfusion pressure fluctuation was not assessed. Previous findings have shown that the circadian MOPP is a consistent risk factor for glaucoma severity in patients with normal-tension glaucoma. ^14,15 Thus, measurements of 24-hour MOPP or DPP may provide more precise information for assessing the relationship of perfusion pressure and risk of OAG. Second, we used only one measurement of IOP and thus the IOP reading may have been subject to measurement error. As demonstrated in our power analyses, the greater the measurement error, the lower the statistical power to detect the link between perfusion pressure and OAG. Third, our data were cross-sectional, and thus we cannot attribute causality from the recorded associations between perfusion pressure and OAG risk. Fourth, we included all the nonglaucomatous participants (including persons who did not undergo visual field testing) for our analyses. Thus, on the basis of the ISGEO scheme, a small proportion of glaucoma patients with glaucomatous minicups may have been misclassified as nonglaucomatous. However, we have no reason to believe that this misclassification would bias the association between OAG and perfusion pressure. Finally, as already noted in our power analyses, we had limited statistical power to separate the OAG cases into normal- and high-tension OAG for supplementary analysis. Given the reduced sample size, we were also unable to exclude systemic hypertensive patients with antihypertension treatment. 

In conclusion, we showed that low DBP, low MOPP, and low DPP are statistically significant and independent risk factors for OAG in Asian Malays. The findings provide further evidence of a vascular mechanism in glaucoma pathogenesis across different populations. 

Footnotes

 Supported by National Medical Research Council Grant 0796/2003 and Biomedical Research Council Grant 501/1/25-5.

Footnotes

 Disclosure: Y. Zheng, None; T.Y. Wong, None; P. Mitchell, None; D.S. Friedman, None; M. He, None; T. Aung, None