Ocular Perfusion Pressure and the Risk of Open-Angle Glaucoma: Systematic Review and Meta-analysis

Low ocular perfusion pressure (OPP) has been proposed as an important risk factor for glaucoma development and progression, but controversy still exists between studies. Therefore, we conducted a systematic review and meta-analysis to analyze the association between OPP and open-angle glaucoma (OAG). Studies were identified by searching PubMed and EMBASE databases. The pooled absolute and standardised mean difference in OPP between OAG patients and controls were evaluated using the random-effects model. Meta-regression analysis was conducted to investigate the factors associated with OPP difference between OAG patients and controls. A total of 43 studies were identified including 3,009 OAG patients, 369 patients with ocular hypertension, and 29,502 controls. The pooled absolute mean difference in OPP between OAG patients and controls was −2.52 mmHg (95% CI −4.06 to −0.98), meaning significantly lower OPP in OAG patients (P = 0.001). Subgroup analyses showed that OAG patients with baseline IOP > 21 mmHg (P = 0.019) and ocular hypertension patients also had significantly lower OPP than controls (P < 0.001), but such difference in OPP was not significant between OAG patients with baseline IOP of ≤21 mmHg and controls (P = 0.996). In conclusion, although no causal relationship was proven in the present study, our findings suggest that in patients with high baseline IOP, who already have a higher risk of glaucoma, low OPP might be another risk factor.

Data extraction and quality assessment. Two investigators (KEK and SUB) independently extracted data in a masked manner using a data extraction form. Discrepancies between the investigators were resolved by the third investigator (JWJ). Following data were extracted from the studies: (1) study characteristics, including the year of publication, name of the first author, country, study design, number of included eyes, patient demographics; (2)  The quality of included studies was assessed using the Risk of Bias Assessment Tool for Nonrandomized Studies (RoBANS 2.0) method by two independent investigators (KEK and SUB) 16 . The RoBANS tool consists of eight domains: comparability of participants, selection of participants, confounding variables, intervention (exposure) measurement, blinding of outcome assessment, outcome evaluation, incomplete outcome data, and selective outcome reporting. The risk of bias for each domain was categorized as low risk, high risk, and unclear risk.
Statistical methods. The pooled OPP difference, which is to say, the difference in OPP between the glaucoma and control groups from the meta-analysis of the included studies, was presented as the mean difference and the standardised mean difference with 95% confidence interval (CI). Heterogeneity in meta-analysis refers to the variation in study outcomes between or among studies. The I 2 statistic, representing the percentage of variation across studies that is due to heterogeneity rather than to sampling error, was evaluated for the degree of unexplained variation in the OPP/glaucoma association 17,18 . The values can range from 0 to 100%, 0% indicating statistical homogeneity and 100% statistical heterogeneity. It has been suggested that the adjectives low, moderate, and high be ascribed to I 2 values of 25, 50, and 75%, respectively 19 . Due to high levels of heterogeneity, differences in mean IOP, OPP, BP levels between glaucoma patients and control were analyzed using random-effects model, www.nature.com/scientificreports www.nature.com/scientificreports/ which assumes that the true underlying effect between studies varies 20 . Meta-regression analysis was used to find possible potential factors that could result in OPP difference between glaucoma patients and controls. Variables including age, gender (proportion of men), MAP, SBP, DBP and IOP were included in the analyses. For studies not having MAPs but only SBP and DBPs, MAPs were calculated as DBP + [1/3 (SBP -DBP)]. All statistical analyses used 95% CI and P-values with a cut-off point of 0.05. All statistical analyses were performed using the software package R version 3.6.2 21 .

Results
characteristics of included studies. We identified 1,152 studies through database searches. After reviewing abstracts, we excluded 1,080 studies that were not relevant, leaving 72 studies for full-text evaluation (Fig. 1). Of these, 43 studies were finally included in the current systematic review and meta-analyses. Among them, 4 studies were population-based, cross-sectional studies 8,[22][23][24] and the others were hospital-based, clinical case-control studies 14, . These 43 studies included 3,009 patients with OAG (1,294 OAG, 926 POAG, and 789 NTG), 369 OHT patients, and 29,502 controls from 19 countries and their characteristics are summarized in the Table 1. The quality of the evidence was generally good, but the risk of bias caused by confounding variables was high in 12 (27.9%) studies, and unclear in 3 (7.0%) studies. Ocular perfusion pressure difference between glaucoma patients and control groups. The main outcome of the present study was the difference in OPP (measured in mmHg) between patients with and without OAG and its significance. The pooled average difference in OPP between patients with and without OAG was −2.52 mmHg (95% CI, −4.06 to −0.98, P = 0.001), with a high degree of heterogeneity (I 2 = 92.3%), presented in Fig. 2A. The pooled standardised average difference in OPP between OAG patients and control was −0.38 (95% CI, −0.56 to −0.20, P < 0.001), also with high degree of heterogeneity (I 2 = 90.1%, Fig. 2B). These showed that OAG patients had significantly lower OPP than controls.
Additional analyses. The study that contributed the most to the heterogeneity for the OPP difference between controls and OAG patients was Mroczkowska et al. 36 Meta-regression analyses using random-effects model were performed to investigate the potential risk factors associated with the pooled standardised average difference in OPP between OAG patients and controls. Random effects meta-regression analyses showed that age, systolic BP, diastolic BP, mean arterial pressure, and study design were not significantly associated with the standardised mean difference in OPP. However, studies with larger proportion of men showed increasing trend of standardised mean difference in OPP (P = 0.040, Fig. 5A) and the OPP difference was larger in studies with lower mean OPP level (P = 0.029, Fig. 5B).

Discussion
Our systematic review and meta-analysis showed that patients with OAG had lower mean OPP compared to controls. The pooled mean absolute difference in OPP level between OAG patients and controls was −2.52 mmHg. Moreover, patients with POAG and OHT also showed significantly lower OPP than controls, with the pooled mean absolute difference of −4.20 mmHg and −6.01 mmHg, respectively. However, this trend of the relationship was not significant in NTG patients. We concluded that low OPP may be a significant risk factor for OAG patients with high baseline IOP, and thus, control of IOP leading to appropriate OPP may be important in terms of regulating vascular factors for glaucoma treatment.    www.nature.com/scientificreports www.nature.com/scientificreports/ Vascular factors have long been suspected of playing an important role in the glaucomatous process in addition to IOP. In this aspect, previous studies have reported a significant association between low OPP and glaucoma. Low OPP represents hypo-perfusion to ONH, ultimately leading to ONH deprived of nutrition and oxygenation. However, several studies have shown no association between them 14 . This may be due to the fact that IOP and BP, the major constitutes of OPP, were measured in different clinical settings (e.g. study population, types of glaucoma, use of topical drugs). Moreover, various BP-and OPP-related parameters (e.g., MOPP, systolic OPP, diastolic OPP, MAP, SBP, DBP) were used, which rendered interpretation of the effect of OPP on glaucoma more complex, in line with the study by Barbosa-Breda et al. 63 Therefore, to address the gap in consideration of heterogeneity between studies, we conducted systematic review and meta-analysis and confirmed that OAG patients had significantly lower MOPP than controls.

First author Year
Our results showed that POAG patients, whose baseline IOP of more than 21 mmHg, showed significantly lower OPP than controls. Previous studies have suggested that NTG patients may be more affected by ischemic injury associated with vascular factors than mechanical injury by elevated IOP 44,[64][65][66][67][68][69] . In this aspect, one may expect significantly lower OPP in NTG patients than in controls. However, this was not proven in the present study. Following reasons may explain our findings. First, OPP itself is not the only vascular factor causing ischemic injury to the optic nerve head. Moreover, several studies reported that OPP fluctuation or degree of its variability, rather than one single OPP value may be more important in development and progression of NTG 65,70,71 . Second, several studies have suggested that NTG patients may have vascular dysregulation or weak www.nature.com/scientificreports www.nature.com/scientificreports/ vascular regulating system to defense against provocative stimulation. Thus, their vulnerability to vascular insults may not be revealed under normal (resting) condition 14,72 . All of the included studies measured OPP at resting and sitting position, and thus, our results could not reveal the OPP results in NTG patients under provocative stimulation. Third, the present study only included studies with definite MOPP values presented as mean ± SD. Thus, several large, population-based studies including Baltimore Eye Survey 73 , Egna-Neumarkt Study 12 , Los Angeles Latino Eye Study 74 which reported the association between low OPP and increased risk of glaucoma in terms of odds ratio were excluded. Since these studies also had a large proportion of patients with baseline IOP less than 21 mmHg, these could have affected the present meta-analysis.
The actual OPP should be determined by the difference between arterial pressure at the entrance to the eye and the venous pressure at the exit of the eye. However, currently available methods cannot directly measure such pressures. Therefore, OPP has been estimated by the difference between arterial pressure measured in the arm and IOP, which may not reflect actual measures. Based on current equation, either decrease in BP or increase in IOP may influence the decrease in OPP. However, these parameters cannot be evaluated separately for the association with glaucoma, since they are all included in the same equation. We conducted direct comparison of OPP, IOP, and BP between POAG and NTG subgroups in an attempt to explain the effects of these parameters on OPP difference. However, it was not possible, since only 18.6% (8/43) of studies had both POAG and NTG groups. Also, some studies presented "untreated" IOP values, while others presented "treated" IOP values for the subjects' www.nature.com/scientificreports www.nature.com/scientificreports/ baseline characteristics. Thus, we indirectly calculated the pooled average OPP, IOP, and BP differences between POAG and NTG groups using network meta-analysis (Supplemental Table S1) 75 . Despite the high heterogeneity and the limited number of studies, we confirmed that the POAG group showed lower OPP and higher IOP compared with the NTG group. By contrast, there was no significant difference in pooled average BP difference between POAG and NTG patients by network meta-analysis. These additional findings seem to imply that high IOP might be the basis for low OPP in patients with POAG.
Several studies reported that diastolic BP is more important in determining OPP than other BP parameters. As the degree of BP is larger than that of IOP, OPP may be more sensitive to changes in BP than those in IOP. We used meta-regression analyses to examine the potential evidence as to which BP parameter would be associated with OPP difference. However, none of the BP parameters showed any association with OPP difference. This could be attributable to the fact that all of the studies included patients with no hypertension or hypertension under controlled BP with or without medication. Another possible cause is the fact that not all of the studies had available BP values. To overcome the limitation that only 32.6% (14/43) studies had mean arterial pressures, we even calculated them additionally based on systolic BP and diastolic BP provided in 65.1% (28/43) of studies, but the association was insignificant. As only mean OPP was used in the analyses, the future investigation is needed on whether BP parameters could have effect on other OPP parameters (e.g. diastolic, systolic).
Several limitations should be considered for the interpretation of our results. First, our meta-analyses are based on cross-sectional studies. Thus, despite a significant association between low OPP and glaucoma, our results cannot provide evidence for a causal relationship between them. The Barbados Eye Study, a prospective, longitudinal study reported that low MOPP, systolic OPP, and diastolic OPP were all associated with a higher risk of developing glaucoma at 4 and 9 years of follow-up 10 . Despite this, further longitudinal studies providing sufficient clinical evidence are needed to address this causal relationship. Second, the heterogeneity of pooled studies was substantial. This may result from differences in study population, study design, and participant characteristics. We used random-effects analyses to overcome such limitation and no publication bias was detected by Egger's test. Finally, the present study showed the absolute mean difference of OPP as approximately 2 mmHg between OAG patients and controls, but these were based on single measurements. Thus, further investigation is required to prove the clinical relevance of OPP fluctuation or dynamic range of OPP on glaucoma.
In conclusion, our systematic review and meta-analysis point to evidence for low OPP in OAG patients, which is in line with the current notion that decreased vascular supply to the ONH may increase its vulnerability to glaucomatous structural damage. Additionally, among OAG patients, those with high baseline IOP particularly, rather than those with low baseline IOP, showed significantly lower OPP than the controls. Although further investigation might be needed, our results imply that in patients with high baseline IOP, who already have a higher risk of glaucoma, low OPP might be another risk factor. www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/