Two clusters of patients with open angle glaucoma were computed based on ocular blood flow parameters. Although some uncertainty regarding the nature of the parameters measured with the devices used in the present study persists,25
all are related to ocular blood flow. In the present study, the two computed groups differed mainly in choroidal and optic nerve blood flow parameters. Between these two clusters, no differences in sex distribution, propensity to have normal tension glaucoma, age, endothelin-1 plasma levels, visual field damage, intraocular pressure, systemic blood pressure parameters, or ocular perfusion values were observed. The only statistically significant difference between the two groups, besides the parameters used to compute the two clusters, was the propensity to show a vasospastic response in nailfold capillaroscopy, with 46% of the patients of the group with high values of ocular blood flow parameters showing a capillary blood flow stop, while this proportion reached 80% among the patients with low values of ocular blood flow parameters. The present study did not address the effect of low values of blood flow parameters on the prognosis of glaucomatous disease, but suggested a statistical association between altered ocular blood flow parameters and a vasospastic propensity. This observation is of importance, indicating that systemic blood flow alterations are related to ocular blood flow disturbances in glaucoma.
We used a somewhat unusual approach in the present analysis. In contrast with traditional hypothesis testing designed to verify differences between groups of patients, an exploratory data analysis was employed. Parameters of ocular blood flow were used to compute two groups, using cluster analysis, a multivariate exploratory technique designed specifically to identify patterns in multivariate data sets. Theoretically, any set of variables such as intraocular pressure or visual field damage can be used to compute clusters, each approach addressing a different question. Computationally, cluster analysis may be thought of as analysis of variance “in reverse.” The computation will start with random clusters, and then move patients between those clusters with the goal to (1) minimise variability within clusters and (2) maximise variability between clusters. In contrast with traditional hypothesis testing where groups of patients are defined based on obvious criteria, a cluster analysis allows us to define groups of patients based on more complex characteristics. Once the groups are defined, they can be compared with regard to parameters other than those used to compute the clusters.
The present cluster analysis yielded intriguing observations. The most marked differences were noted for choroidal and optic nerve blood flow parameters (both eyes showing the same trend). End diastolic velocity in the ophthalmic artery (probably the least “ocular” of all tested vessels) was higher, but only borderline significantly and only in the left eyes of patients with generally low ocular blood flow. Therefore, this finding does, in our opinion, not invalidate our consideration of the two clusters as one group with low ocular blood flow and one group with high ocular blood flow. Patients with low values of ocular blood flow parameters did not have more advanced visual field damage, suggesting that blood flow alterations do not, at least not only, occur secondary to visual field damage. The same patients were not necessarily patients with normal tension glaucoma and did not have markedly lower untreated IOP readings in a diurnal tension curve, suggesting that ocular blood flow alterations are not, at least not only, secondary to increased IOP. Furthermore, low values of ocular blood flow parameters were not related to low systemic blood pressure, suggesting that ocular blood flow alterations occur not only in response to systemic perturbations. The only positive association with low values of ocular blood flow parameters was observed with a vasospastic responsiveness of the nailfold capillaries. The fact that up to half of the patients with high values of ocular blood flow parameters show a vasospastic responsiveness of the nailfold capillaries suggests that the presence of systemic vasospastic responsiveness does not necessarily mean that ocular blood flow is altered. This is also in accordance with the relative high frequency of a vasospastic propensity in a healthy population.26
Nevertheless, the fact that four out of five patients with low values of ocular blood flow parameters showing a vasospastic responsiveness of the nailfold capillaries, suggests an association between altered ocular blood flow and vasospastic propensity. It is unlikely that the latter association is because the observed lower values of ocular blood flow parameters are brought about by a systemic vasoconstrictive response to an external perturbation. However, although this interpretation is purely speculative, it is possible that similar phenomena occur within the ocular vascular bed of some, but not all, glaucoma patients with a vasospastic responsiveness of the nailfold capillaries, leading ultimately to lower blood flow parameters. Most importantly, results of a cluster analysis are to be understood as giving a hunch of what should be investigated. An inferring interpretation of the results would not be appropriate and the present findings need further confirmation.
Regarding risk factors in glaucoma, the differentiation between mechanical damage, as a consequence of IOP,27
and vascular theories, considering glaucomatous optic neuropathy as a consequence of insufficient blood supply,28
had little therapeutic impact so far, as IOP reduction was the only intervention available for treatment of glaucoma. Clinical intervention studies have proved the role of IOP and the benefits of IOP lowering treatment.29,30
Given the success of IOP reduction, consideration of ancillary risk factors may seem distracting. However, once pharmacological means and laser surgery have failed, what remains is a filtering procedure with all its risks and possible complications. Elucidating how other factors influence the course of glaucomatous disease may open new therapeutic avenues as welcome alternatives to surgical treatment.