Microalbuminuria and diabetic retinopathy in type 2 diabetic patients: From risk association to risk prediction

Type 2 diabetes has become a pandemic disorder and its alarming increase in prevalence raises worldwide concerns. In contrast to Western countries where the largest increase in the numbers of diabetic patients are the elderly population aged >64 years, the majority part of the increase in Asia will occur among the population aged 45–64 years. These patients have long disease duration and thus are associated with a higher risk of long‐term complications, including diabetic retinopathy. Diabetic retinopathy is the most common cause of preventable blindness in working‐aged adults. Despite low prevalence rates being reported in some Asian countries (such as China and India)1, because of the effect of urbanization, increasing obesity and longer lifespan, the frequency and severity of diabetic retinopathy is expected to increase in most Asian countries during the next decades.

Microalbuminuria is widely accepted as the first clinical sign of diabetic nephropathy. Current knowledge about the natural course of diabetic kidney disease is mostly derived from studies of patients with type 1 diabetes. As diabetic nephropathy progresses, the development of microalbuminuria eventually leads to macroalbuminuria and then to progressive loss of glomerular filtration rate (GFR). Among type 1 diabetic patients who have nephropathy, more than 95% will already have diabetic retinopathy. However, things are more complicated for patients with type 2 diabetes, because they are also susceptible to parenchymal renal disease other than classic diabetic glomerulosclerosis, which might include hypertensive atherosclerosis and lipid toxicity2. Despite many researchers suggesting the superior role of microalbuminuria in predicting adverse outcomes, including all‐cause mortality, cardiovascular end‐points and renal failure even among patients without diabetes, few studies have investigated the potential association of microalbuminuria and retinopathy in type 2 diabetic patients. In a recently‐published hospital‐based cohort study, Chen YH et al. compared the predicting capability of microalbuminuria and moderately reduced GFR on predicting the development and progression of retinopathy among 487 type 2 diabetic patients3. During the mean follow up of 6.6 years, they found that patients with microalbuminuria and estimated GFR >60 mL/min/1.73 m² had a threefold increase in risk compared with those with normoalbuminuria and estimated GFR 30–59.9 mL/min/1.73 m². They concluded that microalbuminuria has a greater impact on predicting diabetic retinopathy development and progression than moderate decline in renal function in patients with type 2 diabetes. This finding is interesting, but not unexpected, as microalbuminuria truly reflects systemic microvascular complications (which include diabetic retinopathy), but moderately reduced GFR might not be totally attributed to diabetic glomerulosclerosis.

However, the question must be asked, ‘does microalbuminuria have any additional role in predicting retinopathy as compared with other major risk factors?’ Although many cross‐sectional studies have found the apparent association between albuminuria and prevalent diabetic retinopathy that might be a result of enrolling patients with long disease duration, few cohort studies have investigated the incidence and progression of diabetic retinopathy in type 2 diabetic patients. Table 1 summarizes the major published cohort studies that examined independent risk factors for diabetic retinopathy incidence or progression in type 2 diabetes patients. Most of the studies found that in addition to age and diabetic duration, glycemic level and baseline blood pressure were the two important independent predictors for diabetic retinopathy incidence and progression. Among the few studies that evaluated the albumin excretion rate or albuminuria, many did not find an independent association after putting these variables into the multivariable regression models. Therefore, in a prior systemic review on urine albumin testing for early detection of diabetic complications, the author concluded that, ‘there was no evidence of any independent prognostic significance for the incidence of retinopathy, and little evidence for the progression of retinopathy or development of proliferative retinopathy’4. One speculation for the inconsistent results might be that the predicting power of microalbuminuria might vary among patients with different age, diabetic duration, retinopathy severity, glycemic and blood pressure level, and even GFR level. A more sophisticated prediction equation might be required in this complicated situation5. As risk association does not necessarily translate into risk prediction, further research is required to evaluate whether incorporating information on albuminuria will substantially increase the accuracy of clinical prediction in terms of improvement in c statistics, calibration or reclassification.

Table 1. Cohort studies that examined independent risk factors for diabetic retinopathy incidence or progression in type 2 diabetes patients.

Study	Country	Setting	Retinal screening method	n	Follow‐up duration	Independent risk factors for retinopathy incidence	Independent risk factors for retinopathy progression	Albumin excretion rate or albumiuria in multivaraible analysis
Agardh et al.a	Sweden	Hospital	Fundus photography	240	5 years	Not found	Systolic blood pressure	NS
Florkowski et al.b	New Zealand	Hospital	NR	153	6 years	Not found	NR	NS
Guillausseau et al.c	France	Hospital	Fluroescein angiography	64	7 years	Baseline A1c	NR	NS
Kim et al.d	Korea	Hospital	Ophthalmoscopy	130	5 years	Baseline A1c	NR	NS
Voutilaninen‐Kaunisto et al.e	Finland	Population	Fundus photography	80	10 years	Fasting plasma glucose	NR	NR
United Kindom Prospective Diabetes Study 50f	UK	Hospital	Retinal photography	1,919	6 years	Baseline A1c, higher blood pressure, non‐smoking	Higher A1c, older age, male, non‐smoking	NR
Looker et al.g	US (Pima Indians)	Population	Funduscopy Retinal photography	411	4 years	Baseline A1c, use of oral hypoglycemic agents	Macroalbuminuria, hyperglycemia, age	Macroalbuminuria P = 0.01
Tapp et al.h	Australia (Mauritius)	Population	Retinal photography	528	6 years	Fasting plasma glucose, diabetes duration	NR	NR
Blue Mountains Eye Studyi	Australia	Population	Fundus photography	139	5 years	NR	Fasting blood glucose, diabetes duration	NR
Tam et al.j	Hong Kong	Hospital	Stereo fundus photography	413	4 years	Baseline A1c	Macroalbuminuria, high mean A1c	Macroalbuminuria P = 0.001
Yamamoto et al.k	Japan	Hospital	Fundus photography, fluorescein angiography	1,173	6 years	Not found	Baseline systolic blood pressure	NS

A1c, glycated hemoglobin A1c; NR, non‐reported; NS, non‐significant.

^aAgardh et al. Diabetes Res Clin Pract 1996; 32: 35–44.

^bFlorkowski et al. Diabetes Res Clin Pract 1998; 40: 167–73.

^cGuillausseau et al. Diabet Med 1998; 15: 151–155.

^dKim et al. Diabetes Care 1998; 21: 134–138.

^eVoutilainen‐Kaunisto et al. J Diabetes Complications 2001; 15: 24–33.

^fStratton et al. Diabetologia 2001; 44: 156–163.

^gLooker et al. Diabetes Care 2003; 26: 320–326.

^hTapp et al. Diab Res Clin Pract 2006; 73: 298–303.

ⁱCikamatana et al. Eye 2007; 21: 465–471.

^jTam et al. J of Diabetes and its Complications 2009; 23: 185–193.

^kYamamoto et al. Geriatr Gerontol Int 2012; 12 (Suppl 1): 141–144.

Another question is, ‘if there is a role for microalbuminuria in diabetic retinopathy prediction, then what is the optimal cut‐off point?’ One USA study analyzing the National Health and Nutrition Examination Survey in 2005–2008 found that glycated hemoglobin levels at which the risk for prevalent retinopathy begins to increase are lower (5.5–5.9%) in African Americans than in Caucasians (6.0–6.4%)6. In a cross‐sectional study of 4,739 type 2 diabetic patients in Shanghai China, Chen H et al. reported that a microalbuminuria threshold of 10.7 mg/24 h, which was within the traditional ‘normal range’, can predict the increased risk for diabetic retinopathy development7. This result was further validated in a prospective cohort of 297 patients followed for 4.5 years showing that microalbuminuria 10.7 mg/24 h was still an independent factor for the development of retinopathy, even after adjusting for other risk factors. Further studies are required to find the optimal cut‐off points of urinary albumin excretion for identifying type 2 diabetic patients at highest or lowest risk of developing retinopathy.

Despite both studies being hospital‐based with relatively small sample sizes and susceptibility to substantial loss‐to‐follow up and detection bias, the findings by Chen YH et al. and Chen H et al. are inspiring, and encourage more researchers to explore the optimal utilization of microalbuminuria in detecting diabetic retinopathy in type 2 diabetes patients. As frequent tests of microalbuminuria and annual eye examinations become the standard of care in diabetes management, healthcare professionals could pay more attention to the relationship between them in order to optimize the screening frequency for diabetic retinopathy in patients with type 2 diabetes.