Keywords

Hypertension causes alterations in vascular structure and function. Structural changes in small arterioles can be diverged to two different patterns: first eutrophic remodeling characterized by a rearrangement of the smooth muscle cells around a narrowed lumen but without growth response (meaning that media cross-sectional area remains unchanged) and second hypertrophic remodeling, a growth response with increment of media cross-sectional area observed in patients with long-standing and/or severe hypertension [1]. Regardless of the pattern, both are characterized by an increased wall-to-lumen ratio (WLR).

The analysis of retinal vessels offers the exceptional opportunity to assess directly and noninvasively human microvasculature in vivo. In the last years several methods have been introduced for the assessment of retinal changes. Since this book is proposed as a practical approach guiding the reader in the assessment, the focus is on the most established methods, namely, funduscopy and scanning laser Doppler flowmetry (SLDF).

1 Assessment of the Retinal Arterioles

For a long time, direct ophthalmoscopic examination using the traditional four-grade classification system with increasing severity (Table 11.1, Fig. 11.1) introduced by Keith, Wagener, and Barker [1], modified by Scheie [3], was regarded as part of standard evaluation of patients suffering from hypertension [4]. Nowadays, its clinical usefulness in current clinical practice has been questioned due to its unreliable reproducibility [5], and hence routine funduscopic examination is no longer recommended [6]. Reliable assessment was only shown for advanced alterations like hemorrhages and exudates referring to at least grade 3 [7].

Table 11.1 Keith-Wagener-Barker classification
Full size table
Fig. 11.1
figure 1

Funduscopic changes (e.g., cotton wool) (Republished from Ott and Schmieder [2])

Full size image

1.1 Funduscopy

In the last decade several approaches have been developed, assessing more sensitive and quantitative alterations of retinal microvascular changes. Although protocols may differ in some minor points, the principles are similar. According to standardized protocols, one (e.g., 45°) nonstereoscopic color retinal photograph centered between the optic disk and the macula and approximately two disk diameters nasal to the optic disk has to be done in a darkened room. Hence, due to dark adaption, mydriatic agents are no longer necessary. However, in some studies (e.g., Rotterdam Study), pharmacological mydriasis was routinely done. For quantitative assessment of retinal vessels, the photographs have to be converted to digital pictures and analyzed by specific imaging software, e.g., the “Interactive Vessels Analysis” (IVAN) (University of Wisconsin, Madison, WI, USA). This software analysis provides semiautomated measurement of retinal arterioles and venules. Using formulas (e.g., Parr and Spears [8] or Knudtson et al. [9]), a single “central retinal artery equivalent (CRAE)” and a “central retinal vein equivalent (CRVE)” are calculated. Subsequently, arteriole-to-venule ratio (AVR) can be computed; for details see Hubbard et al. [10]. However, by this method, it is not possible to evaluate the retinal vascular wall thickness or vessel diameter directly.

1.2 Scanning Laser Doppler Flowmetry

SLDF, introduced by our study group about 10 years ago, allows the dynamic assessment of both functional (i.e., vascular tone) and structural parameters (i.e., wall and lumen diameter). In brief, SLDF is performed in the juxtapapillary area of the right eye, 2–3 mm temporal superior of the optic nerve at 670 nm (Heidelberg Retina Flowmeter, Heidelberg Engineering, Germany). A retinal sample of 2.56 × 0.64 × 0.30 nm is scanned within 2 s (at least one full systolic and one diastolic phase) and measured every 10 μm of this specific length of the retinal arteriole (80–140 μm). The confocal technique of the device ensures that only capillary flow of the superficial layer of 300 μm is measured. No pupil dilation is necessary (i.e., no constriction of patient daily routine) [11].

For assessment of functional parameters, mean retinal capillary flow (RCF) is assessed in the area of interest, and for further dynamic analysis, non pharmacological and pharmacological tools can be applied. Flicker light increases RCF at least in part via a nitric oxide (NO)-dependent mechanism and represents a non pharmacological tool to investigate vasodilatory capacity of retinal arterioles. It is noteworthy to mention that flicker light exposure has no effects on systemic blood pressure (BP), thereby minimizing potential systemic hemodynamic influences on RCF. Moreover, basal NO activity can be assessed by administration of the NO synthase inhibitor NG-monomethyl-l-arginine (l-NMMA).

For assessment of structural parameters, the outer arteriole diameter (AD) is measured by reflection images, and the lumen diameter (LD) is measured by perfusion images. From the raw parameters, wall thickness (WT, [AD−LD]/2), WLR ([AD−LD]/LD) (Fig. 11.2), and wall cross-sectional area (WCSA, п/4 × [AD2–LD2]) can be calculated.

Fig. 11.2
figure 2

Scanning laser Doppler flowmetry (SLDF) (Republished from Ott and Schmieder [2]). (a) Differentiation between retinal arteriole and venule (SLDF live image before measurement). (b) Scanned area – reflection image. (c) Scanned area – perfusion image. (d) Scanned area – corrected and analyzed flow image. (e) Pulse curve run as mean retinal capillary flow (RCF) and time plot. (f) Localization of systolic and diastolic RCF on the image (d). (g) Localization of systolic and diastolic RCF on the image (c). (h) Calculation of wall-to-lumen ratio

Full size image

Importantly, also individual pulsatile pattern of functional (RCF) and structural (e.g., WT) parameters of retinal arterioles in systole and diastole can be reliably assessed (Fig. 11.2).

All analyses are performed offline with automatic full-field perfusion imaging analysis (AFFPIA) (SLDF Version 4.0 by Welzenbach with improved resolution) [11].

2 Prevalence and Incidence (General Population, Hypertension)

2.1 Funduscopy

Several population-based studies have provided data on prevalence of retinal signs using standardized funduscopic photographs in the general population, partly with subsequent categorization according to (among others) hypertension status. In general, retinal signs are common in people aged 40 years or older, even in those without arterial hypertension. However, these findings can only be respected with caution, since different definitions of arterial hypertension have been used. Moreover, the reported prevalence of retinal signs depends also largely on the assessed parameter, e.g., retinopathy per se, AV nicking, or focal/generalized arteriolar narrowing (for details see Table 11.2).

Table 11.2 Prevalence/incidence of retinal signs in normotensive and hypertensive populations using funduscopy
Full size table

In the Cardiovascular Health Study (CHS) (aged ≥65 years), 16.6 % (men 19.7 %; women 14.3 %) of normotensive participants and 25.4 % (men 30.0 %; women 23.0 %) of hypertensive patients (defined as BP ≥140/90 mmHg or history of hypertension with use of antihypertensive drugs) were reported to have generalized arteriolar narrowing (defined as the lowest twentieth percentile of AVR). In contrast, retinopathy was by far less frequently documented in this study, i.e., in 5.6 % (men 4.7 %; women 6.3 %) of normotensive participants and in 10.4 % (men 7.5 %; women 11.9 %) of hypertensive patients [16]. Confirmatory results were found in the Beijing study (aged ≥ 40 years), which also used the accepted criteria of hypertension (BP ≥140/90 mmHg or history of hypertension with use of antihypertensive drugs) [19]. Moreover, in the former study another important point was found, namely, differences in the prevalence of retinal signs according gender [16].

Other influencing factors are age and ethnicity. In the Blue Mountains Eye Study (BMES) [14] and the Atherosclerosis Risk in Communities (ARIC) Study [20], the prevalence of retinopathy increased with advancing age, whereas in the CHS only some retinal signs revealed an age-dependent relationship [16]. Regarding ethnicity, an enhanced prevalence of retinopathy was suggested in Afro-Caribbeans compared to Europeans, but in this study the use of standardized protocols was not clearly outlined [12]. In the ARIC study higher prevalence of retinopathy has been documented in blacks compared to whites [17]; however this difference was largely explained by the severity of hypertension.

Much less data are available addressing the frequency of new retinal signs. In the Beaver Dam Eye Study (BDES), the 5-year incidence of focal arteriolar narrowing was 7.7 % and of retinopathy 4.6 %, respectively, in normotensive (BP <160/95 mmHg) participants. Both incidences were about doubled in hypertensive patients [13]. In contrast, in the BMES the 5-year incidence for retinopathy was numerically higher in normotensive (BP <140/90 mmHg) subjects (8.2 %) but not clearly increased in hypertensive subjects (10.4 %) [15]. On the other hand, there is good evidence from several epidemiological studies that retinal alterations (i.e., generalized arteriolar narrowing) precedes the development of hypertension (Table 11.3), as a preclinical marker of hypertension.

Table 11.3 Large-scale, population-based studies (in alphabetical order) assessing associations between retinal vascular caliber (based on retinal photography) and blood pressure, target-organ damage, and cardiovascular risk (in chronological order)
Full size table

2.2 Scanning Laser Doppler Flowmetry

Large epidemiological studies addressing prevalence and incidence of functional and structural microvascular alterations assessed with SLDF are lacking. Regarding RCF, similar values were found in young hypertensive patients compared to normotensive controls, which was confirmed by findings of an unaltered RCF between middle-aged patients with and without hypertension [49, 50]. Moreover, mean RCF was found to be similar in patients with hypertension stage 1–2 compared to patients with advanced stage of hypertensive disease, e.g., patients with treatment-resistant hypertension (TRH). However, by analyzing the individual pulsatile pattern of RCF in latter both groups, we were able to demonstrate a different pattern. RCF in systole was higher, whereas RCF in diastole was lower, and hence an exaggerated pulsed RCF (difference in RCF between systole and diastole) in patients with TRH was observed compared to patients with hypertension stage 1–2 [51].

Regarding structural parameters, small (monocentric) studies point toward similar findings as seen with funduscopy, namely, an increment of retinal alterations (e.g., WLR) with increased BP [52]. Moreover, a pooled analysis comprising ≥500 patients suggests an increased WLR with aging (Schmieder RE, Ott C, unpublished data). Prevalence and incidence rates of retinal alterations assessed by SLDF are difficult to describe since no thresholds values for the parameters are yet established.

3 Change with Treatment (Criteria for Significant Change, Incidence During Treatment)

3.1 Funduscopy

It was repeatedly shown that initiating effective antihypertensive therapy resulted in disappearance of severe (grade III and IV) hypertensive retinopathy [53, 54]. In a case report of a 34-year-old woman with a short history of hypertension, headache, and blurred vision, all indicative of malignant hypertension (her BP was 240/150 mmHg), funduscopic examination found swelling of the optic disk, widespread hemorrhages, and soft and hard exudates, consistent with grade IV or malignant hypertensive retinopathy, respectively. Antihypertensive treatment was initiated, and 10 months follow-up revealed a good BP control (110/70 mmHg). In accordance, funduscopy demonstrated an improvement of hypertensive retinopathy [55]. In a small cohort (n = 28), comprising previously untreated men with hypertension stage 1–2, scored (0–4) funduscopic changes were evaluated before and after 26 weeks of treatment with enalapril or hydrochlorothiazide, respectively. Both treatments resulted in a significant BP reduction but numerically higher after enalapril (−14.3 vs. −7.1 mmHg) compared to hydrochlorothiazide without reaching significant difference. Treatment with enalapril reduced numerically but nonsignificantly the frequency of arteriolar narrowing and arteriovenous crossing, whereas no changes were seen after treatment with hydrochlorothiazide [56].

Whether these observed changes are associated with improved cardiovascular (CV) and cerebrovascular prognosis remains to be determined. Indirect evidence comes from epidemiological studies. For example, in the BMES prevalence of hemorrhages and/or microaneurysm was comparable between normotensive and controlled (BP <160/95 mmHg) hypertensive men but not in women [14]. A subsequent sub-analysis of BMES revealed that prevalence of focal arteriolar narrowing was similar between normotensive (4.6 %) and controlled (BP <160/95 mmHg) hypertensive subjects (6.5 %), whereas its prevalence was more than doubled in treated uncontrolled (14.5 %) and untreated (15.3 %) hypertensive patients. In contrast, generalized arteriolar narrowing (narrowest quintile of AVR) was similarly prevalent in treated and controlled (22.0 %), treated and uncontrolled (22.5 %), and untreated hypertensive (27.2 %) patients but significantly greater compared to normotensive subjects (17.0 %) [57].

Thus, in contrast to data on improvement or even disappearance of qualitative hypertensive retinal abnormalities (e.g., papilledema, exudates), data are much less clear for quantitative retinal signs (e.g., arteriolar narrowing). On the other hand, generalized retinal arteriolar narrowing and AV nicking appear to be (irreversible) markers of mild to moderate hypertension, related not only to current and past BP levels but to cerebrovascular diseases as well [58].

3.2 Scanning Laser Doppler Flowmetry

Again data are limited with SLDF compared to funduscopy and are based on small studies only. Data with SLDF revealed that endothelial function (basal NO activity) was impaired in young hypertensive patients and improved after treatment with the angiotensin receptor blocker (ARB) candesartan [49], whereas no improvement was demonstrated in elderly hypertensive men after treatment with ARB valsartan [59]. Whether this discrepancy is related to the different ARB, different duration of therapy, or a potential irreversibility of vascular changes in the elderly patients is subject of ongoing investigations. Moreover, we were able to demonstrate that vasodilatory capacity (magnitude of vasodilation to flicker light) was lower in untreated hypertensive patients compared to normotensive controls, and systolic BP was inversely related to the percent increase of RCF due to flicker light exposure, independently of other CV risk factors [60]. Another study of our group suggests that BP and hence pulse pressure (PP) changes have an impact on pulsed RCF. In hypertensive patients with TRH, we observed a decrease of systolic and pulsed RCF 6 and 12 months after renal denervation (RDN), in parallel to decreases of BP and heart rate (HR). The reduction of pulsed RCF after RDN transfers into less shear stress on the vascular wall and, thereby, suggests an improvement of retinal (and potentially cerebral) microcirculation [61].

In a cross-sectional study, we observed that in treated hypertensive patients with BP control <140/90 mmHg, WLR was at the same level as observed in normotensive subjects but significantly lower than in treated hypertensive subjects with BP >140/90 mmHg [62]. Previously, two small prospective studies assessed the effect of antihypertensive treatment on retinal structural parameters using SLDF. In one study, hypertensive patients with non-insulin-dependent diabetes mellitus were treated with either aliskiren (n = 9) or ramipril (n = 7) for one year. To achieve equivalent BP control, open-label hydrochlorothiazide could be added, and hence only one patient in each group had BP ≥140/90 mmHg. Both treatment regimes resulted in a significant regression of retinal WLR after 1-year treatment without a difference between the groups [63]. In a second small unblinded study, hypertensive patients were treated with lercanidipine for 4 weeks and thereafter randomized to additional antihypertensive therapy with either enalapril (n = 10) or hydrochlorothiazide (n = 10) for 24 weeks. There was an improvement of WLR already after 4 weeks of treatment with lercanidipine alone, and only enalapril on top further reduced WLR (but not hydrochlorothiazide) [64]. Of note, both studies had small sample sizes, and surprisingly high values of WLR (>0.5) at baseline were reported. The reduction of WLR of about 50 % in both studies is very high in comparison to changes observed after treatment in analyses relying on the assessment in vascular remodeling of subcutaneous small arteries [65]. In a double-blind randomized study comprising in total of 40 patients with mild to moderate hypertension, treatment with manidipine or amlodipine for 4 weeks resulted not in any significant changes in WLR compared to baseline values (Ott et al., unpublished data). Overall data are sparse, and to clarify the effects of various antihypertensive agents on reversal of WLR, multicenter double-blind randomized studies with large number of patients are required.

4 Prognostic Value of Change

4.1 Funduscopy

No data are available whether treatment-induced regression of retinal alterations is related with reduction of other target-organ damages (e.g., left ventricular hypertrophy) or incident CV outcomes. So far, only epidemiological studies uniformly found that qualitative retinal signs of hypertensive retinopathy are related with incidence of CV disease. In accordance, quantitative retinal vascular caliber was associated with BP, target-organ damage, and CV disease (Table 11.2). Cross-sectional studies also indicated that the prevalence in retinal signs differs between normotensive, treated and controlled, treated and uncontrolled, and never-treated hypertensive patients. Thus, in addition to our pathophysiological understanding of vascular remodeling and its consequences, there seems to be a strong rationale that treatment-induced changes may also result in an improved CV outcome.

4.2 Scanning Laser Doppler Flowmetry

No prospective study analyzing the effects of treatment changes of retinal alterations assessed by SLDF, and hence its prognostic significance, is available. Again, at least indirect evidence exists that is based on findings of media-to-lumen ratio of subcutaneous small arterioles (measured with a myograph ex vivo) and cross-sectional studies.

In an Italian study (n = 126) with an averaged follow-up of 5.4 years including both patients with primary and secondary hypertension (e.g., pheochromocytoma) as well as normotensive subjects, an increment of subcutaneous media-to-lumen ratio was predictive of a diminished event-free survival [66]. Subsequent analysis, with an increased study population (n = 303) and a mean follow-up of 6.9 years, revealed that increased media-to-lumen ratio was of prognostic significance with regard to adverse CV and cerebrovascular outcome [67]. Accordingly, a Danish study comprising 159 patients with primary hypertension and moderate CV risk reported that media-to-lumen ratio was an independent predictor for the incidence of CV outcome even after adjustment of the Heart Score level over the follow-up of 4.6 years [68]. Recently, it was shown in a long-term follow-up survey (comprising 124 hypertensive patients) that after 9–12 months of antihypertensive treatment, SBP was reduced from 164 ± 15 to 134 ± 14 mmHg, which was accompanied by an regression of media-to-lumen ratio of subcutaneous small arteries (0.084 ± 0.03 vs. 0.075 ± 0.02, p < 0.01). Importantly, in the subsequent follow-up period of 15 years, the extent of the reduction in the media-to-lumen ratio of subcutaneous small arteries was demonstrated to be an independent predictor of CV events [69].

The relevance of these data on subcutaneous small arteries for retinal arteriolar changes comes from the nowadays recognized concept that changes seen in the small subcutaneous small arterioles are reflecting alterations seen also in other vascular beds. Indeed, Rizzoni et al. have previously demonstrated that WLR assessed by SLDF (retinal arterioles in vivo) and media-to-lumen ratio measured with myograph (subcutaneous small arteries taken from a biopsy) showed close correlation in hypertensive subjects (r = 0.80, p < 0.001), suggesting that SLDF may provide similar information about microcirculation alterations compared to subcutaneous small arteries [68].