To investigate the relationship of circulating levels of soluble form of LR11 (sLR11; also called SorLA or SORL1), with the progression of proliferative diabetic retinopathy (PDR) in patients with type 2 diabetes mellitus.
Fifty-four patients with type 2 diabetes mellitus were divided into 2 sex- and age-matched groups: one with PDR (n = 29) and the other with nonproliferative diabetic retinopathy (n = 25). The serum sLR11 levels were measured with an immunodetection system followed by chemifluorescence quantification.
The serum sLR11 levels were higher in the PDR group than in the nonproliferative diabetic retinopathy group (5.8 ± 1.2 U vs 3.7 ± 1.3 U; P < .01). A multivariate regression analysis showed that circulating sLR11 is a factor contributing to the prediction of PDR independent of other classical risk factors, and an area under the receiver operating characteristic curve analysis revealed that the sensitivity and the specificity were equivalent to or more than those of other factors. Among the classical risk factors for PDR, glycosylated hemoglobin levels showed the highest correlation coefficient ( P < .01) for the sLR11 concentrations.
Serum sLR11 concentration may reflect the progression of PDR in patients with type 2 diabetes mellitus. sLR11, released from immature vascular cells and indicating the development of atherosclerosis, is expected to be a novel candidate biomarker indicating diabetic retinopathy in patients with type 2 diabetes mellitus.
Sustained hyperglycemia, even in the absence of other risk factors, can increase the risk of microvascular complications. Given the substantial quality-of-life burden that diabetic retinopathy can confer, the ability to detect early retinal vascular abnormalities sensitively in patients with diabetes mellitus is desirable. The detection of such markers of pathologic cell function in combination with treatment of hyperglycemia is needed.
LR11 (also called SorLA or SORL1), an low-density lipoprotein (LDL)-receptor family member, has been identified as a molecule expressed in intimal smooth muscle cells in the development of atherosclerosis and endothelial cells under the condition of dyslipidemia. The released soluble form of LR11 (sLR11) promotes pathologic infiltration of macrophages into the damaged vessels. We have shown that the circulating sLR11 levels were increased in patients with coronary artery disease and dyslipidemic subjects with carotid atherosclerosis. A multivariate analysis in these independent studies in patients with atherosclerosis indicated that the sLR11 levels were correlated distinctly with the glycemic level among the classical risk factors for atherosclerosis.
Diabetic retinopathy mainly is caused by diffuse endothelial damage at the microvascular level. However, the interesting observations are that the retinopathy is tightly associated with increased cardiovascular mortality, reduced coronary reactivity, and poorer prognosis of coronary revascularization procedures. Thus, high glucose levels may change the phenotype of endothelial cells as well as that of arterial smooth muscle cells; the pathologic cell phenotype in microvessels of the retina possibly is detected by the circulating sLR11 released from the damaged cells. In this analysis, we investigated the significance of circulating sLR11 with regard to proliferative diabetic retinopathy (PDR) in patients with type 2 diabetes mellitus. The factors contributing to the elevation of the serum sLR11 also were analyzed.
The subjects consisted of 56 consecutive Japanese patients with type 2 diabetes mellitus seeking treatment at the Department of Laboratory Vascular Function, Toho University Sakura Medical Center, who had already given blood samples. PDR was defined according to the international clinical classification of diabetic retinopathy as neovascularization in the retina. Vitreous surgeries had been performed to treat macular edema (n = 7), vitreous hemorrhage (n = 13), traction retinal detachment (n = 5), or neovascular glaucoma (n = 4). None of the nonproliferative diabetic retinopathy (NPDR) cases had retinal neovascularization. Patients with chronic heart disease with an ejection fraction of less than 50% or chronic renal failure with serum creatinine of more than 1.3 mg/dL were excluded from the study analysis.
Patient Data Analysis
Blood samples were collected in the morning after an overnight fast. Lipid variables and fasting blood glucose were measured using standard laboratory techniques. The potential risk factors for atherosclerosis were analyzed, including age, sex, body mass index (BMI), smoking, and history of hypertension and dyslipidemia. Hypertension was defined as systolic pressure of more than 140 mm Hg or diastolic pressure of more than 90 mm Hg. Diabetes mellitus was defined as a fasting blood glucose level of more than 126 mg/dL, glycosylated hemoglobin (HbA1c) of more than 5.8%, or both. Dyslipidemia was defined as serum total cholesterol of more than 220 mg/dL and triglycerides of more than 150 mg/dL in the fasting state, or both, and high-density lipoprotein (HDL) cholesterol of less than 40 mg/dL, or a combination thereof. The serum creatinine level was assayed by an enzymatic method. The estimated glomerular flow rate was estimated using a modified traceable Modification of Diet in Renal equation, as proposed by the Working Group of Japan Chronic Kidney Disease Initiative : estimated glomerular flow rate (mL/minute per 1.73 m 2 ) = 0.741 × 175 × age −0.203 × serum creatinine −1.154 (if female × 0.742).
Measurement of Serum sLR11
For the analysis of sLR11, fasting blood samples were collected and centrifuged immediately at 4000 g for 10 minutes, and the supernatant immediately was frozen in polypropylene tubes and stored at −80 C until use. Fifty microliters of serum was purified using 39-kDa receptor-associated protein-GST affinity beads (Cosmo Bio, Toyo city, Tokyo, Japan). For immunoblotting, equal amounts of protein extracted from pelleted beads were subjected to 10% sodium dodecyl sulfate poly-acrylamide gel electrophoresis (SDS-PAGE) after heating to 95 C for 5 minutes, as described previously under reducing conditions, and were transferred to a nylon membrane. Incubations were carried out with an antibody against LR11 (5-4-30-19-2 at 1:500 dilution), followed by peroxidase-conjugated antimouse immunoglobulin G. The development was performed with the ECL detection reagents (Amersham Pharmacia, Piscataway, New Jersey, USA). The signals were quantified by densitometric scanning using the NIH image software program (National Institutes of Health, Betheda, Maryland, USA). The sLR11 levels in each serum sample (50 μL) were determined as an averaged value of 3 quantified signal intensities resulting from independent assays using samples with blinded indications and were expressed as a ratio to that of standard serum. The immunologic estimation indicated that the signal of 1 U (in 50 μL serum) corresponded to approximately 50 ng/mL of recombinant sLR11.
The results are shown as means ± standard deviation or proportion (%) for each index. The statistical analyses were performed using the SPSS Statistical Package for Windows software program version 11.01.1. Comparisons between groups were performed using the Student t test. The data were subjected to a 1-way analysis of variance with the Dunnett multiple comparison of means. A Pearson correlation coefficient analysis was used to assess the associations between measured parameters. Subsequently, multiple linear regression analyses were used to calculate the odds ratio for PDR by controlling for all risk factors. These risk factors were scored as explanatory factors, and the subordinate variable was PDR = 1 and NPDR = 0. The sensitivity and specificity with respect to the presence of PDR were analyzed using a conventional receiver operating characteristic (ROC) curve. P values less than .05 were considered to be statistically significant.
The patient characteristics are shown in Table 1 . The age- and gender-matched NPDR and PDR groups comprised 25 and 29 subjects, respectively. There were no statistically significant differences in BMI, duration of diabetes, frequency of hyperlipidemia or dyslipidemia, or estimated glomerular flow rate between the NPDR and PDR subjects. There were also no statistically significant differences in HbA1c, fasting blood sugar, or lipid concentrations between the NPDR and PDR subjects. Although there was no significant difference in the use frequency of statin, angiotensin converting enzyme inhibitor (ACE-I), or angiotensin II receptor type 1 blocker (ARB) between the 2 groups, the frequency of patients using insulin was significantly higher in the PDR subjects than that in the NPDR subjects.
|NPDR Group||PDR Group||P Value|
|Age (y)||66.0 ± 8.6||62.4 ± 9.7||.15|
|Duration of diabetes (y)||11.4 ± 7.8||11.9 ± 7.8||.71|
|Body mass index (kg/m 2 )||23.8 ± 4.0||25.8 ± 3.7||.06|
|eGFR (mL/minute per 1.73m 2 )||60.2 ± 15.3||58.2 ± 28.0||.72|
|HbA1c (%)||6.5 ± 0.8||7.0 ± 1.4||.10|
|Fasting blood sugar (mg/dL)||124.6 ± 33.1||132.1 ± 38.2||.63|
|Total cholesterol (mg/dL)||183.8 ± 34.8||202.6 ± 40.7||.12|
|LDL cholesterol (mg/dL)||111.7 ± 30.6||124.1 ± 33.3||.26|
|HDL cholesterol (mg/dL)||47.7 ± 16.4||49.4 ± 10.6||.58|
|Triglyceride (mg/dL)||122.4 ± 43.1||122.7 ± 52.7||.93|
|Insulin therapy (%)||13.6||65.5||< .0001|
|Administration of statin (%)||45.5||24.1||.11|
|Administration of ACE-I or ARB (%)||54.5||41.4||.43|
The circulating sLR11 levels in the NPDR and PDR groups were 3.7 ± 1.8 U and 5.8 ± 2.7 U, respectively ( Figure 1 ) , indicating that the sLR11 levels in the PDR group were higher than those in the NPDR group ( P < .01). Note that we previously reported that the mean circulating sLR11 levels in 400 dyslipidemic subjects was 3.0 ± 1.0 U. The sLR11 analysis restricted for the patients not treated with insulin showed that the sLR11 levels again were higher in the subjects with PDR (4.8 ± 1.2 U; n = 10) than in those with NPDR (3.7 ± 1.3; n = 12; P < .05). Thus, circulating sLR11 levels were increased in type 2 diabetes mellitus patients with PDR regardless of medication with insulin therapy.
We analyzed the significance of the sLR11 concentration in comparison with other risk factors for PDR, including age, male gender, estimated glomerular flow rate, and the total cholesterol and HbA1c concentrations, in all subjects ( Table 2 ). The multivariate analysis using all variables for PDR showed that the circulating sLR11 level, as well as younger age, strongly associated with PDR independent of other variables.
|Odds Ratio (95% Confidence Interval)||P Values|
|Age, per 1-y increase||4.12 (0.78 to 0.996)||< .05|
|Male||0.01 (0.16 to 6.18)||.99|
|eGFR, per 1-U (mL/minute per 1.73 m 2 ) increase||0.80 (0.97 – 1.08)||.37|
|Total cholesterol, per 1-mg/dL increase||1.43 (0.99 to 1.04)||.34|
|HbA1c, per 1% increase||1.23 (0.24 to 2.04)||.51|
|sLR11, per 1-U increase||8.50 (1.63 to 12.25)||< .01|
The ROC curves of the various factors were examined for discriminating the probability of the type 2 diabetes mellitus patients with PDR from the NPDR patients based on the levels of sLR11, the levels of HbA1c, the BMI, or the estimated glomerular filtration rate ( Figure 2 ) . The curves showed the fraction of true-positive results (sensitivity) and false-positive results (1-specificity) for various cutoff levels of each parameter. The cutoff level of sLR11 that gave the maximum sensitivity and specificity for PDR was 4.2 U. At the cutoff level, the sensitivity of sLR11 for PDR was 78%, and the specificity was 77%, equivalent to or more than the other classical risk factors, HbA1c, BMI, or estimated glomerular flow rate ( Table 3 ).