Retinal Vessel Pulsatile Characteristics Associated With Vascular Stiffness Can Predict the Rate of Functional Progression in Glaucoma Suspects

Abstract

Purpose: Tissue stiffening and alterations in retinal blood flow have both been suggested as causative mechanisms of glaucomatous damage. We tested the hypothesis that retinal blood vessels also stiffen, using laser speckle flowgraphy (LSFG) to characterize vascular resistance.

Methods: In the longitudinal Portland Progression Project, 231 eyes of 124 subjects received LSFG scans of the optic nerve head (ONH) and automated perimetry every 6 months for six visits. Eyes were classified as either “glaucoma suspect” or “glaucoma” eyes based on the presence of functional loss on the first visit. Vascular resistance was quantified using the mean values of several instrument-defined parameterizations of the pulsatile waveform measured by LSFG, either in major vessels within the ONH (serving the retina) or in capillaries within ONH tissue, and age-adjusted using a separate group of 127 healthy eyes of 63 individuals. Parameters were compared against the severity and rate of change of functional loss using mean deviation (MD) over the six visits, within the two groups.

Results: Among 118 “glaucoma suspect” eyes (average MD, −0.4 dB; rate, −0.45 dB/y), higher vascular resistance was related to faster functional loss, but not current severity of loss. Parameters measured in major vessels were stronger predictors of rate than parameters measured in tissue. Among 113 “glaucoma” eyes (average MD, −4.3 dB; rate, −0.53 dB/y), higher vascular resistance was related to more severe current loss but not rate of loss.

Conclusions: Higher retinal vascular resistance and, by likely implication, stiffer retinal vessels were associated with more rapid functional loss in eyes without significant existing loss at baseline.

Ocular blood flow is altered in eyes with primary open-angle glaucoma,¹ yet causality remains unclear. Eyes with fewer remaining retinal ganglion cells have lower metabolic demand, which leads to lower blood flow, but reduced blood flow could also further damage remaining retinal ganglion cells. It is also possible that reduced blood flow is contributory to glaucoma pathophysiology more directly and at earlier stages. Autoregulation ensures nearly constant blood supply as ocular perfusion pressure varies,² and impaired autoregulation may be a risk factor for developing glaucoma,³ as seen, for example, in individuals with Flammer syndrome,⁴ but autoregulation has also been shown to be altered during the course of experimental glaucoma.⁵ It is still unknown whether vascular and mechanical changes represent two discrete causes of glaucomatous damage or are mutually causative and hence intrinsically linked.⁶ 

Tissue stiffness is also believed to have a role in the pathophysiology of glaucoma. In the anterior segment, glaucoma has been linked with increased stiffness of the trabecular meshwork, reducing aqueous humor outflow.⁷ In the posterior segment, the manner in which the optic nerve head (ONH) responds to fluctuations in intraocular pressure (IOP) seems to be driven in large part by scleral stiffness.^8–11 In the mechanosensitive lamina cribrosa cells, sustained strain provokes mechanotransduction, extracellular matrix production, and fibrosis,¹² driving further stiffening. Notably, the peripapillary sclera,¹³ ONH,¹⁴ and conventional outflow pathways¹⁵ all exhibit viscoelastic properties such that they stiffen with elevated IOP and also stiffen in older eyes,^16–18 these being the two biggest known risk factors for developing glaucoma. 

Within the peripapillary retina, it has been suggested that shear strains are higher in the vasculature than in neuroglial tissue.¹⁹ This suggests that it may be the biomechanical properties of the vessels in particular that play a role in the development and progression of glaucoma. It is therefore important to investigate whether the stiffness of blood vessel walls and vascular resistance are related to glaucoma. Vascular stiffness increases with age,²⁰ the primary risk factor for glaucoma. Stiffer vessels and higher vascular resistance would be consistent with reduced blood flow,^21–23 reduced autoregulatory capacity,^3,5 impaired neurovascular coupling,^24,25 and increased risk of disc hemorrhages,^26–28 each of which has been reported to be related to glaucoma. However, retinal vessel stiffness cannot yet be directly measured in vivo in human eyes. Instead, differences in vascular resistance between eyes can be inferred by examination of the pulse wave, which is measurable using laser speckle flowgraphy (LSFG).^29–32 The mean blur rate (MBR) measured by LSFG has been shown to be linearly correlated with blood flow^33,34 and is assessed at 30 frames per second, enabling parameterization of the pulsatile waveform. It is not clear which parameter is most strongly related to resistance, but by investigating multiple parameters clear conclusions may emerge. In this study, we used this technique to test the hypothesis that vascular resistance may be related to the severity and rate of glaucomatous functional progression at two different stages: eyes with and without existing functional loss. 

Methods

Subjects

Longitudinal blood flow measurements in a cohort with glaucoma/glaucoma suspects were compared against a separate normative database, all tested at Devers Eye Institute (Portland, OR, USA). The healthy control cohort consisted of 127 eyes of 64 subjects without systemic hypertension, significant visual field loss, or any conditions likely to cause visual field loss. Of those eyes, 99 were tested twice on the same day, and the remainder were tested once. 

The longitudinal cohort consisted of 231 eyes of 124 subjects tested at Devers Eye Institute as part of the ongoing Portland Progression Project (P3)^35–37 who had open-angle glaucoma or suspected glaucoma as determined by the subject's clinician. Participants in the P3 study undergo a set of functional and structural diagnostic tests once every 6 months, including LSFG, standard automated perimetry, and IOP measurement by tonometry. To be included in this analysis, a series of six visits at which LSFG and perimetry were performed was required; if more than six such visits were available, the most recent six were used. Subjects were excluded if they had significant visual field loss due to causes other than glaucoma, again as determined by their clinician; if they had systemic hypertension (systolic pressure ≥160 mmHg and/or diastolic pressure ≥100 mmHg when measured during the study visit, which excluded five subjects enrolled in the P3 study who would otherwise have been eligible); or if they produced unreliable visual fields, as indicated by a glaucoma hemifield test (GHT)³⁸ result from perimetry of “abnormally high sensitivity” on their first test date (this excluded 17 otherwise eligible eyes from the P3 cohort). Both eyes were tested, even if only one had extant glaucomatous loss, on the basis that this implies that the fellow eye could be considered as a “glaucoma suspect” that is at increased risk of developing glaucomatous loss in the future.³⁹ The cohort was then subdivided into 118 “glaucoma suspect” eyes without functional loss, defined as a GHT³⁸ result of “within normal limits” on the first of the six test dates in the series, and 113 “glaucoma” eyes with existing functional loss, defined as a GHT result of “outside normal limits” (90 eyes) or “borderline” (23 eyes) on the first test date. 

Laser Speckle Flowgraphy

Measurements were performed using the LSFG-NAVI instrument (Softcare Co., Ltd., Fukuoka, Japan) to obtain parameters of ONH blood flow and pulsatile hemodynamics. LSFG has been described in detail in previous studies.^29,33,40,41 Briefly, a fundus camera equipped within the LSFG device was focused on a 750 × 360 pixel area (approximately 6 × 3.8 mm)³³ centered on the ONH. An 830-nm laser generates a speckle pattern, due to random interference of scattered light from the illuminated tissue area. The speckle pattern is imaged by a charge-coupled device at a frequency of 30 frames per second for a period of 4 seconds. The MBR from speckle contrast within the images is computed by the manufacturer's LSFG Analyzer software (Softcare Co.). MBR at a given pixel and frame is defined as MBR = (M/D)2, where M/26 is the mean intensity across 26 pairs of pixels (the pixel of interest at frame T together with each of the surrounding 8 pixels in frame T and each of those 9 pixels in frames T – 1 and T + 1), and D/26 is the mean difference within those pairs of pixels.⁴² As a ratio, MBR is reported in arbitrary units. MBR varies temporally and spatially according to the amount and velocity of blood cell movement and correlates well with blood flow within the ONH.^33,34 The user then defines an elliptical region of interest encompassing the ONH. The analysis software of the instrument splits this region into major vessels (which serve the retina) versus ONH tissue, classifying pixels by whether their average MBR is above or below a set threshold so that separate parameterizations can be performed for each. Up to three such scans were conducted per eye per day, in short succession, so that scans of poor quality (typically, scans in which the subject blinked too much) could readily be excluded; if more than one scan per day had adequate image quality, results from those scans were averaged to give a single value for each parameter. 

The native software of the LSFG instrument applies a Fourier transform to the pulsatile waveform of MBR over the 4-second measurement. This is used to calculate: 

• • Average mean blur rate (MBR_Ave)—MBR_Ave is a surrogate measure of mean blood flow.
• • Beat strength (BS)—BS is the maximum power in the Fourier-transformed data, which is closely correlated with the pulsatile range from the systolic maximum of MBR to the diastolic minimum of MBR.
• • Normalized beat strength (BOM)—BOM is defined as BS/MBR_Ave, which is therefore similar to the commonly used hemodynamic pulsatility index defined as (Velocity_systole – Velocity_diasole)/Velocity_mean, which is known to be positively correlated with peripheral vascular resistance.³¹ BOM has been shown to be elevated in oxygen-induced ischemic retinopathy.³²
• • Total capillary resistance (TCR)—TCR is defined as (BS[vessel] – BS[tissue])/(MBR_Ave[vessel] – MBR_Ave[tissue]). This reflects the vascular resistance of the retinal vasculature, because the flow within the major vessels traversing the ONH primarily serves the retina. The parameter TCR is elevated in central retinal vein occlusion⁴³ and branch retinal vein occlusion.⁴⁴

The LSFG-NAVI software also generates a number of parameterizations of the average raw pulsatile waveform, without using a Fourier transform.^30,45 These parameters are simpler to calculate and visualize but at the expense of a higher proportion of missing data, as they cannot be calculated in a waveform that includes too many blinks. Thus, for these parameters, results for a full series of six test dates were available for 208 eyes of 109 subjects. The parameters include the following: 

• • Rising rate (RR)—RR is a measure of the integral of the waveform during the time period in which MBR is increasing. This is defined as 25 × Area B/(Area A + Area B), where areas A and B are the areas above and below the pulsatile waveform during this period, respectively, as seen in Figure 1. An elevated RR indicates that the MBR increases more abruptly at the onset of the pulse cycle, suggesting stiffer vessels.
• • Falling rate (FR)—FR is the equivalent measure of the integral of the waveform during the time period in which MBR is decreasing.
• • Blowout score (BOS)—BOS is a measure of the proportion of blood flow that is maintained in tArea B/(Area Ahe vessel between heartbeats. This is defined as 100 × (2 – AC/DC)/2, where AC is the range of MBR and DC is its mean value across the pulse cycle, as seen in Figure 1. It equals 100 × [1 − (MBR_max − MBR_min)/MBR_mean] and so is analogous to 100% minus half of the pulsatility index [(Velocity_systole − Velocity_diasole)/Velocity_mean]³¹ but based on MBR rather than blood flow velocity. Thus, lower BOS indicates greater pulsatility, indicative of stiffer vessels.
• • Blowout time (BOT)—BOT is the proportion of the pulsatile waveform in which the MBR is closer to its maximum than its minimum.
• • Flow acceleration index (FAI)—The FAI is the maximum increase in MBR between two image frames (at a frame rate of 1/30 seconds).
• • Acceleration time index (ATI)—The ATI is the time taken for the waveform to reach its peak as a proportion of the length of the waveform.
• • Resistivity index (RI)—The RI is the range of MBR values expressed as a proportion of the maximum, (Max − Min)/Max, as seen in Figure 1. This has a similar interpretation as BOS, but with the opposite sign. It is analogous to the widely used resistance index defined as (Velocity_systole − Velocity_diasole)/Velocity_systole⁴⁶ and depends on both vascular resistance and vascular compliance.⁴⁷

Figure 1.

 

Schematic representation of the definitions of some of the parameterizations of pulsatile waveform produced by the LSFG instrument software, based on the average pulse cycle within a 4-second window. The blowout score is defined as BOS = 100 × (2 − AC/DC)/2, where AC represents the range of fluctuation and DC represents the mean over the pulse cycle. The rising rate is defined as 25 × Area B/(Area A + Area B), and the resistivity index is defined as RI = (Max − Min)/Max.

Many of these parameters are correlated with one another. For example, in the full longitudinal dataset, the correlation between RI and BOS was −0.98 in vessels and −0.99 in ONH tissue. The optimal parameter for measuring vascular resistance is not yet known, but it is expected to be correlated with each of BOM, TCR, RR, BOS, and RI. 

Perimetry

Visual field tests were performed using the HFA IIi perimeter (Carl Zeiss Meditec, Dublin, CA, USA), with the 24-2 test pattern and SITA Standard test strategy.⁴⁸ Strict cutoffs on reliability indices were not imposed, on the basis that they have recently been shown to be uninformative about test performance and reliability of the results.^49–52 Instead, the technician performing the test observed the patient, provided audible reminders and encouragement as needed, and repeated the test when necessary; eyes with a GHT result of “abnormally high sensitivity” were excluded as mentioned above. For each visual field, a linear-scaled mean deviation (MD) was calculated, defined as MD_Lin = 10^(MD/10), where MD is the instrument's decibel-scaled value, because linear-scaled indices have been shown to be more closely correlated with structural loss than decibel-scaled indices.^53–55 All analyses were repeated using decibel-scaled MD, and results were consistently similar to those using MD_Lin but with increased nonlinearity and hence fewer significant P values (results not shown). The rate of functional loss for an eye was calculated by linear regression of MD_Lin against test date over the series of six visual fields in the series. 

Analysis

Many of the LSFG parameters change with normal aging.⁴⁵ Therefore, age-corrected parameters were created, adjusting each of the listed LSFG parameters to their equivalent value for a patient of age 60 years, based on linear regression against age within the healthy control eyes, under the null hypothesis that age-corrected LSFG parameters should not differ between cohorts and hence the same age correction is appropriate for both cohorts. For consistency, all parameters were age corrected in this manner, even if the relation with age in the healthy control eyes was not significant for that particular parameter. Normative limits for each age-corrected parameter were defined as the empirical 5th and 95th percentiles of the distribution in the healthy control cohort. In order to reduce variability, the mean parameter value from the six test dates in the series was used for further analysis. Secondary analyses were also performed using just the first test date in the series. 

For the longitudinal cohort, the rate of change of function over the six test dates (using MD_Lin) was plotted against the mean of the six values of each age-corrected LSFG parameter in turn, together with the normative limits from the healthy control eyes. Generalized estimating equation (GEE) models⁵⁶ were used to determine whether each parameter predicted the severity of loss (the average value of MD_Lin in the series) and the rate of functional change after adjusting for the current severity of functional loss, accounting for inter-eye correlations in the cohort. These GEE models were performed separately for “glaucoma” eyes (with functional loss at baseline, based on the GHT as defined above) and for “glaucoma suspect” eyes (without functional loss at baseline). Secondary analyses were performed adjusting for IOP (the average of the six measurements). 

Results

The average age in the longitudinal cohort was 72.6 years (range, 50–90). Among the 118 “glaucoma suspect” eyes, the average age was 72.8 years, and the average MD was −0.45 dB, with an average rate of change of −0.445 dB/y. By definition, all of these eyes were “within normal limits” by the GHT at baseline (i.e., on the first of the six visits). Of those eyes, 41 remained “within normal limits” by the GHT throughout their six visits, 36 had at least one subsequent “borderline” visual field, and 41 had at least one subsequent “outside normal limits” visual field. Among the 113 “glaucoma” eyes, the average age was 72.5 years, and the average MD was −4.3 dB, with an average rate of change of −0.528 dB/y. The healthy control eyes had average age of 58.9 years (range, 23–86). 

Figure 2 shows MD_Lin averaged across the six fields in the series (top) and its rate of change over those six fields (bottom), plotted against MBR_Ave averaged across the same six test dates within ONH tissue (left) and ONH major vessels (right). Consistent with previous literature,^23,57 in “glaucoma” eyes reduced function was associated with reduced MBR in both tissue (P < 0.0001, GEE linear regression) and vessels (P = 0.0001). Similarly, in “glaucoma suspect” eyes, reduced function was again associated with reduced MBR in both tissue (P = 0.0031) and vessels (P = 0.0052). The relation was significantly shallower for “glaucoma suspect” eyes than for “glaucoma” eyes in tissue (P = 0.028) but not in vessels (P = 0.403). Although lower MBR appears to be unrelated to more rapid functional loss in the “glaucoma” eyes in Figure 2, after adjustment for the current level of loss (average MD_Lin over the six visits) more rapid functional loss was associated with reduced MBR_Ave in ONH vessels, for both “glaucoma” eyes (P = 0.0393) and “glaucoma suspect” eyes (P = 0.0012), but associations with MBR_Ave in ONH tissue were weaker (P = 0.0898 and P = 0.0272, respectively). 

Figure 2.

 

Severity of functional loss (top) and its rate of change (bottom) over six visits, plotted against the average MBR in optic nerve head tissue (left) and major vessels (right). Blue points represent “glaucoma suspect” eyes without functional loss at baseline, according to the GHT; red points represent “glaucoma” eyes with functional loss at baseline. Linear regression lines are shown together with the shaded 90% prediction intervals. Vertical dashed lines represent the empirical 90% interval for healthy control eyes; note that these represent the normative range for a single scan, not the average of six test dates. All MBR values were age corrected to the equivalent value for age 60 years prior to plotting or calculation of regression lines, based on linear regression among the healthy controls.

Table 1 shows the statistical significance for whether the different LSFG parameters described in the Methods section are related to the severity of functional loss in “glaucoma suspect” and/or “glaucoma” eyes, using the mean value over the six visits for each to reduce variability. Figure 3 shows plots of the severity of functional loss against four LSFG parameters that are believed to reflect vascular resistance. Among the “glaucoma suspect” eyes, the only parameters correlated with severity of functional loss were MBR_Ave, as discussed above, and beat strength within the ONH tissue which is itself correlated with MBR_Ave (correlation 0.798). Among the “glaucoma” eyes, the severity of loss was also correlated with some of the parameters believed to be related to vascular resistance, including BOM, BOS, and RI in both tissue and vessels, although it should be noted that these relations were partly driven by two outlier points seen in Figure 3, representing the two eyes of a single subject. 

Table 1.

 

Statistical Significance of Relations Between Parameters From LSFG (Averaged Across Six Test Dates) and the Severity of Functional Loss (Average Value of Linearized MD Over the Same Dates)

 

Figure 3.

 

Severity of functional loss averaged over six visits plotted against the average value of parameters believed to reflect vascular resistance, in ONH tissue (left) and major vessels (right). Blue points represent “glaucoma suspect” eyes without functional loss at baseline; red points represent “glaucoma” eyes with functional loss at baseline. Linear regression lines are shown together with the shaded 90% prediction intervals. Vertical dashed lines represent the empirical 90% interval for healthy control eyes. All MBR values were age corrected to the equivalent value for age 60 years prior to plotting or calculation of regression lines, based on linear regression among the healthy controls.

Table 2 shows the statistical significance for whether the different LSFG parameters described in the Methods section are related to the rate of functional loss in “glaucoma suspect” and/or “glaucoma” eyes over the same six visits, after adjusting for the severity of functional loss. Figure 4 shows plots of these relations for the same four LSFG parameters as before that are believed to reflect vascular resistance. Among the “glaucoma” eyes, the LSFG parameters that relate to vascular resistance appear to be related to the rate of functional progression in Figure 4, but these relations were not significant after adjusting for the severity of loss. However among the “glaucoma suspect” eyes, more rapid functional loss was correlated with each of the different parameterizations that relate to higher vascular resistance within the vessels, even after adjusting for severity—namely, higher BOM, higher TCR, higher RR, lower BOS, and higher RI. Similar relations were found for vascular resistance within the ONH tissue (i.e., in the capillaries), but these were weaker and not always statistically significant. This is consistent with the finding that TCR was higher in eyes that progressed more rapidly, as this parameter reflects resistance occurring in the vessels after subtracting resistance in ONH tissue. 

Table 2.

 

Statistical Significance of Relations Between Parameters From LSFG (Averaged Across Six Test Dates) and the Rate of Functional Loss (From Linear Regression Over the Same Period), After Adjusting for the Current Severity of Loss

 

Figure 4.

 

Rate of functional loss averaged over six visits plotted against the average value of parameters believed to reflect vascular resistance, in ONH tissue (left) and major vessels (right). Blue points represent “glaucoma suspect” eyes without functional loss at baseline; red points represent “glaucoma” eyes with functional loss at baseline. Linear regression lines are shown together with the shaded 90% prediction intervals. Vertical dashed lines represent the empirical 90% interval for healthy control eyes. All MBR values were age corrected to the equivalent value for age 60 years prior to plotting or calculation of regression lines, based on linear regression among the healthy controls.

IOP can affect some LSFG measurements.⁵⁸ In multivariable models, higher average IOP significantly modulated predictions of the rate of functional change based on MBR_Ave in both tissue and vessels in the “glaucoma suspect” eyes but not in the “glaucoma” eyes (likely because IOP was being managed clinically). However, adjusting for IOP did not significantly alter the relations between the rate of functional change and any of the LSFG parameters related to vascular resistance. 

The results above use the average of the six LSFG measurements as the predictor. In secondary analyses using just the first LSFG measurement in the series, variability and hence P values increase. However, the rate of functional loss in the “glaucoma suspect” eyes was still significantly related (after adjusting for mean severity of loss as before) to MBR_Ave in both ONH tissue (P = 0.045) and vessels (P = 0.041); to BOM in tissue and vessels (P = 0.011 and P < 0.001, respectively); to TCR (P < 0.001); to RR in vessels (P = 0.010); and to RI in vessels (P = 0.009). 

Discussion

In this study, we found that higher vascular resistance and, by implication, higher vascular stiffness were related to more rapid functional loss in glaucoma suspects. If confirmed, this has significant implications for understanding the pathophysiologic processes involved in glaucoma. Stiffer vessels could be linked with more general differences in tissue stiffness,⁵⁹ which is an area of great interest in glaucoma research due to its interaction with aging and IOP,^18,60,61 the two biggest known risk factors for developing glaucoma. More directly, increased retinal vascular resistance would be expected to cause reduced retinal blood flow, which is a well-known facet of glaucoma.^21,62,63 It could also plausibly increase the likelihood of disc hemorrhages, which are known to be more common in glaucomatous eyes and related to the rate of disease progression.^26,64,65 

Among the “glaucoma suspect” eyes, higher vascular resistance was related to more rapid functional progression, even after adjusting for the current severity of functional loss. The range of existing loss was small; these eyes were “within normal limits” according to the GHT at the start of the series, which does not preclude the existence of visual field defects but does ensure that any such defects are very early and possibly undetectable. However it is still notable that none of the vascular resistance parameters was directly correlated with the current severity of loss among these eyes. This suggests that vascular resistance is more likely to be related to the risk of future loss, rather than something that is purely a consequence of prior loss. Indeed, in the secondary analyses, vascular resistance measured just at the first visit in the series was significantly related to the rate of functional loss over subsequent visits. 

Among the “glaucoma” eyes, none of the LSFG parameters that relate to vascular resistance was significantly related to the rate of functional progression, after adjusting for the existing level of loss. However, BOM, BOS, and RI were all related to the current severity of functional loss (Table 1), and eyes with worse loss must have undergone previous functional progression to get to that point. Bearing in mind that these eyes were under clinical management, these findings suggest at least two possible explanations. It is plausible that vascular resistance is a risk factor for conversion from “glaucoma suspect” to “glaucoma” but is largely irrelevant to disease course thereafter. In this scenario, eyes with higher vascular resistance would appear to progress more rapidly at early stages of the disease and would therefore acquire more severe functional loss prior to their progression being detected and slowed by more aggressive treatment. A second possibility is that vascular resistance would remain an important factor for predicting the rate of change even after detectable functional loss has occurred if the eyes were untreated but that this effect is obscured by the fact that the patient's clinician will treat more aggressively if rapid progression is observed. It should also be noted that these results were at least partly driven by the presence of two outliers, which were (in every case) the two eyes of a single subject. Therefore, these results should be treated with caution and need to be confirmed in another dataset. 

The parameters that were most closely related to the rate of functional progression were measurements of pulsatile blood flow within the major vessels passing through the ONH vessels rather than the ONH tissue (microvasculature), as seen in Table 2. Given the limited spatial resolution of LSFG,³³ it is possible that at least some of the pixels along the edges of the major vessels were designated “ONH tissue.” Therefore, although hemodynamics in ONH tissue remain of interest,^29,41 resistance and stiffness of the larger retinal vessels merit particular attention. These arteries primarily supply the retina, and so, although the parameters of pulsatility in this study were assessed within the ONH, they are strongly influenced by conditions throughout the retina. However, increased systemic vascular resistance is a feature of diabetes but the clinical manifestation of diabetic retinopathy is very different from glaucoma. Taken together, these points suggest that hemodynamic changes in glaucoma are at least partly localized within the neuroretinal rim and/or peripapillary region. The ONH and peripapillary tissues are subject to high biomechanical loads and can manifest high strains during IOP fluctuations,¹⁹ and mechanical strains have been found to be higher in the peripapillary vasculature than in adjacent neural tissue.¹⁹ This likely influences the vessels traversing the ONH in a manner that depends on their mechanical stiffness. In the case of veins, compelling evidence suggests that the increase in venous pulsatility observed in glaucoma could be caused by an increase in central retinal vein stiffness in the prelaminar and laminar regions.⁶⁶ Currently, LSFG does not distinguish between the major arteries versus veins; future studies should be designed that can evaluate hemodynamics of these vessel types separately. 

For the reasons discussed in the previous paragraph, the most likely cause of increased vascular resistance is increased vessel stiffness; however, the mechanism of such stiffening cannot be ascertained by this technique. It is possible that the vessel walls stiffen due to changes in thickness and/or changes in their material properties⁶⁷; this could be consistent with reports that shear strains in the peripapillary retina are higher in the vasculature than in the surrounding tissue.¹⁹ There is also evidence of aberrant pericyte constriction of retinal capillaries in a mouse model of experimental glaucoma⁶⁸; this would cause an effective increase in vascular resistance while also reducing the ability to maintain autoregulation including normal neurovascular coupling.⁶⁹ Alternatively, evidence from both clinical and experimental glaucoma studies suggest that some degree of loss of functional microvasculature occurs within the inner retina,⁷⁰ which would also manifest as increased resistance (much like aberrant pericyte constriction reduces the total cross-sectional area of capillary beds). 

These results represent the largest longitudinal study to date of hemodynamic alterations in human subjects with open-angle glaucoma. Although the results may be interesting, they also come with caveats. The group of LSFG pulsatility parameters ostensibly reflecting vascular resistance are only indirect measurements of retinal vessel stiffness, inferred from the shape of the pulsatile waveform along the trunks of these vessels within the ONH. The optimal parameter to use in analyses is not yet known, although it should be noted that results were very similar whether using BOM (analogous to the hemodynamic pulsatility index) or RI (analogous to the hemodynamic resistance index). It has been reported that elevating IOP using an ophthalmodynamometer decreased BOS and increased RI but did not alter RR,⁵⁸ although it should be noted that adjusting for average IOP did not alter the statistical significance of any of the main results in our study. The participants were under clinical management at the discretion of the patient's clinician, so animal models in which relevant variables can be directly controlled may aid our understanding of the contribution of vessel stiffness to the true natural history of the disease. This is especially true in eyes with existing functional loss, as they are more likely to be receiving treatment, and different medications may have different impacts on blood flow and vascular resistance.⁷¹ There is also considerable variability among eyes in the plots in Figure 4 and between test dates; conclusions were similar when using just the first LSFG scan instead of the mean of all six measurements, but they were weaker because the variability reduced the likelihood of achieving statistical significance. Testing was performed using clinical instruments, but further refinement is necessary before it would be possible to predict the likely disease course for an individual patient. This might be achievable with a better understanding of, and hence a better ability to reduce, all of the sources of intersubject variability.^72,73 Alternatively, the results of this study may serve as motivation to develop alternative methods of assessing vascular resistance and vessel stiffness for clinical use, perhaps using optical coherence tomography to measure pulsatile changes in vessel caliber.^74,75 

In conclusion, LSFG was used to measure and parameterize the pulsatile waveform in vessels within the ONH vessels. Increased vascular resistance and, by implication, stiffness of the major retinal vessels were associated with more severe loss among eyes with existing functional defects and were associated with more rapid functional progression among glaucoma suspect eyes. Further studies are warranted to explore these relations using more direct measurements of peripapillary and wider retinal vessel stiffness and to investigate the temporal and causative relations between vessel stiffness and functional loss in glaucoma. 

Acknowledgments

Supported by grants from the National Institutes of Health (R01 EY031686 and EY020922 to SKG) and by unrestricted research support from the Legacy Good Samaritan Foundation (Portland, OR). The sponsors and funding organizations had no role in the design or conduct of this research. 

Disclosure: S.K. Gardiner, None; G. Cull, None; B. Fortune, None