Improving the Prediction of Hypertensive Target Organ Damage Using Novel Markers
Lessons From Retinal Vascular Imaging Research
See related article, pp 338–346
Retinal arteriolar narrowing is a classic sign of hypertensive retinopathy and has long been known to be a marker of hypertensive target organ damage (TOD), such as stroke, heart failure, and chronic kidney disease.1 However, despite the acceptance of the importance of this sign in the medical literature, the value of a retinal assessment in patients with hypertension remains unclear and continues to be debated in clinical guidelines.2
In this issue of the journal, Baumann et al3 examined whether the combination of retinal arteriolar narrowing and albuminuria as markers of microvascular disease may predict the progression of chronic kidney disease. The study prospectively recruited 164 men and women with early and intermediate chronic kidney disease (stages 2–4) who had retinal photographs and computer-assisted retinal vessel diameter measurement, and an assessment of albuminuria. After a median follow-up of ≈4 years, persons with retinal arteriolar narrowing had a higher risk of renal end points, defined in the study as 50% renal function loss and start of renal replacement therapy, than those without retinal arteriolar narrowing. Persons with albuminuria similarly had a higher risk of renal end points than those without albuminuria. Importantly, persons with both markers of microvascular damage (ie, retinal arteriolar narrowing and albuminuria) had substantially higher risk of renal end points compared with those who had no evidence of microvascular damage (ie, wider retinal arterioles and no albuminuria). These are new and exciting results with potential for clinical translation.
One of the long-standing hurdles in the translation of retinal vascular imaging research is the accurate documentation and measurement of retinal arteriolar narrowing. Traditionally, this clinical sign is difficult to detect even for experienced ophthalmologists. Significant improvements in digital retinal vascular imaging technologies now allow more objective documentation of this sign, with the introduction of computer software methods to measure retinal blood vessel (both arteriolar and venular) diameter from retinal photographs. This research was first demonstrated in the Atherosclerosis Risk In Communities (ARIC) study in the late 1990s.4 The ARIC study reported good inter- and intraobserver reliability in measurement of retinal vessel diameter, strong correlation of retinal arteriolar narrowing with higher current, past, and future blood pressure levels, and prospective association of retinal arteriolar narrowing with clinical outcomes such as stroke.5,6 Retinal vascular imaging is now used widely in many epidemiological and clinical studies to examine the relationship of quantitatively measured retinal vessel diameter as an indicator of early microvascular disease to the risk of hypertension and clinically significant TOD. It has been estimated that >100 000 retinal vessel diameter measurements have been performed globally.
Despite these studies, unanswered questions remain regarding whether retinal vascular imaging may provide additional clinical prediction of TOD beyond traditional measures, including blood pressure levels, and whether the extent of this prediction is sufficiently convincing to change management for translation into clinical practice. For example, in a meta-analysis, McGeechan et al7 showed that in the general adult population, after accounting for traditional risk factors, inclusion of retinal vessel diameter assessment in prediction models of stroke reclassifies 10% of people at intermediate risk into a different, mostly lower, risk categories. Thus, 10 retinal vascular imaging scans need to be performed in the community to reassign a single person’s stroke risk, beyond knowing other risk profile indicators such as blood pressure and diabetes mellitus status. These results are not particularly compelling for physicians or healthcare groups to obtain retinal photographs in the general community with an intention to predict and, therefore, prevent stroke.
One approach to improve clinical translation of novel markers or tests such as retinal vascular imaging is to identify a more specific clinical need and target the test to a more specific subgroup of patients who could benefit from more intensive investigations. Neurologists, for example, have long been interested in searching for evidence of microvascular disease in patients who present with stroke, and thus, retinal vascular imaging has been suggested to be a useful investigation in patients with stroke as a means to differentiate lacunar (which shows strong association with retinal arteriolar narrowing) compared with those with large-vessel stroke.8 Such an approach focuses the novel test on a particular clinical need (in this case, the difficulty of diagnosing lacunar stroke) and a subgroup of patients (in this case, patients who present with clinical signs suggestive of microvascular stroke) in which current investigations (in this case, MRI) are either expensive or inconclusive.
Another approach is based on the concept of examining multiple markers or tests to predict TOD. In this scenario, retinal vascular imaging is an adjunctive test that adds to and increases the value of a more widely accepted test. This approach was taken by Baumann et al3 in their article, which examined whether a combination of retinal vessel diameter measurement and albuminuria as markers of microvascular disease may predict the progression of chronic kidney disease. The study found that persons with 2 markers of microvascular damage (ie, retinal arteriolar narrowing and albuminuria) were more likely to have progressive renal disease than persons who had no evidence of microvascular damage. The increased risk was sufficiently large (relative risk ranging from 9 to 16, comparing persons with retinal arteriolar narrowing and albuminuria versus to those with wider arterioles and no albuminuria) and could lead to reclassification of >80% of patients into either a higher or a lower kidney risk category. Such strong effects are impressive and are important convincing evidence for translation into clinical applications. The study had some limitations, including the modest sample size and the small number of participants who had renal events; this complicates statistical analysis in terms of the robustness of regression models to control for traditional risk factors.
Other studies have also examined the value of such an approach—the use of multiple markers to improve prediction of diseases. In the ARIC study, a combination of cerebral and retinal scans to assess the extent of microvascular disease improved the ability to predict stroke.9 Although participants with signs of retinopathy and those with white matter lesions on cerebral MRI were individually more likely to develop a clinical stroke event compared with those who do not have retinopathy or white matter lesions, the risk of stroke was substantially higher (with an adjusted relative risk of 18) among participants who had both retinopathy and white matter lesions compared with those without either marker of microvascular disease. In the same vein, the ability of using other novel biomarkers such as C-reactive protein to predict cardiovascular disease is improved when used in conjunction with other markers. In the Uppsala Longitudinal Study of Adult Men, a combination of biomarkers that reflected myocardial cell damage, left ventricular dysfunction, renal failure, and inflammation (ie, troponin I, N-terminal pro–brain natriuretic peptide, cystatin C, and C-reactive protein, respectively) substantially improved the prediction of cardiovascular disease death compared with the use of a single biomarker alone.10
Clearly, physicians will welcome more precise methods to determine which patient with hypertension is more likely to have TOD such as renal impairment and conversely which patients are at lower risk. Thus, it can be envisioned that in practice, risk stratification of patients with hypertension could have as an initial assessment, already routinely practiced, of the presence or absence of albuminuria. Patients with albuminuria may further have retinal vascular imaging scans, and management is focused only in this subgroup of patients with evidence of early microvascular damage in both the kidney and eye.
There remain some key gaps in translating retinal vascular imaging research into clinical practice. First, the current software to measure retinal vessel diameter, such as the one used in Baumann et al’s study, are not fully automated and require input from a technician based on standardized protocols, training, and manipulation. Newer software with automated optic disc detection, retinal vessel identification, and differentiation (artery versus veins) are being developed and will improve precision and reliability. Second, although Baumann et al’s study has shown a prospective association of the combination of retinal vessel diameter and albuminuria to renal end points, and this approach reclassifies 80% of patients into different risk groups over traditional methods, the performance of this multiple marker strategy should be further assessed for its cost-effectiveness to the healthcare system. Such data are not readily captured in traditional clinical studies, and the types of analyses to demonstrate cost-effectiveness are daunting and difficult to do in isolation without health outcome, health economics, and policy implementation experts.
The exciting journey of retinal vascular imaging research, based on the landmark observations of the correlation of hypertensive retinopathy signs to mortality by Keith, Wagner, and Barker in the 1940s,1 continues.
Sources of Funding
This work was funded by grants from the National Medical Research Council (NMRC/STaR), Singapore.
The opinions expressed in this article are not necessarily those of the editors or of the American Heart Association.
- © 2014 American Heart Association, Inc.
- Mancia G,
- Fagard R,
- Narkiewicz K,
- et al
- Hubbard LD,
- Brothers RJ,
- King WN,
- Clegg LX,
- Klein R,
- Cooper LS,
- Sharrett AR,
- Davis MD,
- Cai J
- McGeechan K,
- Liew G,
- Macaskill P,
- Irwig L,
- Klein R,
- Klein BE,
- Wang JJ,
- Mitchell P,
- Vingerling JR,
- de Jong PT,
- Witteman JC,
- Breteler MM,
- Shaw J,
- Zimmet P,
- Wong TY