Diabetes Diagnostic Criteria May Be Off Target for Retinopathy

By Crystal Phend
MELBOURNE, Australia, Feb. 28 -- Retinopathy occurs at substantial rates even when glucose levels are below those that would trigger a diagnosis of diabetes, according to a meta-analysis.
More than 60% of retinopathy cases were among patients with fasting plasma glucose levels below 7.0 mmol/L (126 mg/dL), reported Tien Y. Wong, M.D., Ph.D., of the University of Melbourne here, and colleagues in the March 1 issue of The Lancet.
WHO and American Diabetes Association criteria for diabetes diagnosis were based on increasing rates of retinopathy beyond the glycemic threshold at 7.0 mmol/L (126 mg/dL) found in three earlier studies.
However, these studies assessed retinopathy from direct clinical ophthalmoscopic examination or a single retinal photograph rather than the gold standard in clinical trials--multiple field retinal photographs.
The meta-analysis of trials using the more reliable method "point to the need to revisit current diagnostic criteria for diabetes," the researchers wrote.
There's no point in throwing out the current criteria, though, commented Quresh Mohamed, M.D., and Alison Evans, MBBCh, both of Cheltenham General Hospital in Cheltenham, England, in an accompanying editorial.
The meta-analysis offered no alternative threshold that would do the job, they noted, and there is limited evidence that further lowering the criteria would be beneficial.
"We perhaps should focus less on a single universal cut-off and instead target resources on the basis of standardized, evidence-based, individual risk scores in which measures of glycemia are combined with other risk factors," they suggested.
"But what would we tell our patients when they asked if they had diabetes?" they asked. "We are probably best sticking with what we know until a better alternative diagnostic tool becomes available."
Dr. Wong and colleagues analyzed cross-sectional data from three population-based cohorts that followed the same protocol for retinopathy grading using multiple field retinal photographs.
This included 3,162 participants from the Blue Mountains Eye Study with prospective five-year follow-up for 1,903 of these patients; 2,182 participants in the Australian Diabetes, Obesity, and Lifestyle Study; and 6,079 patients in the Multi-Ethnic Study of Atherosclerosis.
The overall prevalence of retinopathy in these studies was 11.5%,
9.6%, and 15.8%, respectively.
There was no consistent fasting plasma glucose threshold at which the prevalence of retinopathy increased.
In the Blue Mountains Eye Study, retinopathy prevalence was around 10% even at fasting plasma glucose concentrations below 5.6 mmol/L (100 mg/dL) -- the current definition of normal -- and increased further at concentrations beyond about 6.3 mmol/L (115 mg/dL), the researchers noted.
In the Australian Diabetes study, retinopathy likewise occurred at a similar rate among patients with low fasting plasma glucose concentrations as in the overall cohort. Prevalence increased above 7.1 to 7.8 mmol/L (128 to 140 mg/dL) but for moderate retinopathy there was no suggestion of a threshold.
The Multi-Ethnic Study of Atherosclerosis showed the same pattern of prevalence at low fasting plasma glucose levels but there was no threshold for retinopathy overall or for moderate cases.
Putting the three studies together, 7.4% to 13.4% of participants had retinopathy at glucose levels below 5.6 mmol/L (100 mg/dL).
Earlier studies had suggested that retinopathy was uncommon, at only 2% to 4% prevalence, among patients whose fasting plasma glucose was below 5.6 (100 mg/dL).
In the combined studies, prevalence of retinopathy was 17.8% to 34.7% above the 7.0 mmol/L (126 mg/dL) cutoff for a diagnosis of diabetes.
This two- to five-fold underestimation of prevalence at low glucose levels likely inflated the sensitivity of the cutoffs derived from them, the investigators said.
Whereas the earlier studies suggested a sensitivity of more than 80% for the diagnostic threshold of 7.0 mmol/L (126 mg/dL), Dr. Wong's group found a low sensitivity of only 14.8% to 39.1% for the same cutoff.
Although the specificity was high at 80.8 to 95.8% for detecting retinopathy, the positive predictive value was low at 17.8% to 34.7%.
Limiting the analyses to cases of moderate retinopathy, performance of the 7.0 mmol/L (126 mg/dL) cutoff "remained generally poor."
When the researchers tested a higher 7.8 mmol/L (140 mg/dL) diagnostic threshold, it had lower sensitivity but higher specificity and positive predictive values.
Lowering the cutoff to 5.6 mmol/L (100 mg/dL) increased sensitivity at the expense of specificity and positive predictive values.
Overall, the analysis showed a more gradual increase of retinopathy prevalence with fasting plasma glucose than seen in prior studies, which strongly suggests a continuous rise without a threshold, the researchers said.
"These results of fasting plasma glucose and retinopathy are therefore consistent with observations that the relation between glucose and macrovascular complications such as cardiovascular disease is continuous with no threshold," Dr. Wong and colleagues concluded, "and is analogous to that of end-organ damage found with other cardiovascular risk factors such as blood pressure and serum cholesterol levels."

Primary source: The LancetSource reference:Wong TY, et al "Relation between fasting glucose and retinopathy for diagnosis of diabetes: three population-based cross-sectional studies" Lancet 2008; 371: 736-743. Additional source: The LancetSource reference: Mohamed Q, Evans A "Retinopathy, plasma glucose, and the diagnosis of diabetes" Lancet 2008; 371: b700-702.