Circulating Angiopoietic Cells and Diabetic Retinopathy in Type 2 Diabetes Mellitus, with or without Macrovascular Disease

Abstract

Purpose.: To investigate different types of circulating angiopoietic cells, such as vasculogenic circulating progenitor cells (CPCs), endothelial progenitor cells (EPCs), and mature EPCs (matEPCs) in patients with type 2 diabetes mellitus (T2DM), with or without diabetic retinopathy (DR) and with or without macrovascular disease (MVD).

Methods.: One hundred twenty-six patients with T2DM—66 with MVD and 60 without MVD—were enrolled in a case–control study. MVD comprised coronary heart disease, peripheral arterial disease, stroke, or various combinations of those conditions. By a modified Early Treatment of Diabetic Retinopathy Study (ETDRS) classification, 55 patients were classified without DR (CO), 19 with mild nonproliferative DR (mNPDR), 16 with moderate-severe NPDR (msNPDR), 19 with early proliferative diabetic retinopathy (ePDR), and 17 with high-risk PDR (hrPDR). CPCs (CD34/CD133), EPCs (CD34/CD133/CD30), and matEPCs (CD34/CD133/CD309/CD31) were enumerated by flow cytometry.

Results.: Patients with MVD CPCs, EPCs, and matEPCs showed a significant, stepwise decline with advancing stages of retinopathy. In contrast, in the patients without MVD, EPCs and matEPCs reached up to 56% of CPCs and 37% of EPCs. On the other hand, the percentage of EPCs and matEPCs was reduced to 5% of CPCs and EPCs each in MVD patients. Thus, the percentage of EPCs and matEPCs in comparison with that of CPCs and EPCs represented an 11- and 7-fold difference.

Conclusions.: The circulating angiopoietic CPCs, EPCs, and matEPCs in T2DM patients with DR had a different regulations, with increasing relative differences occurring in proliferative DR, apparently depending on the macrovascular comorbidities. Patients with MVD showed a strong retinopathy-stage–dependent depletion of all angiopoietic cells.

In patients with type 2 diabetes mellitus (T2DM), macrovascular and microvascular complications, including coronary (atherosclerotic) heart disease and diabetic retinopathy (DR), are both the result of long-term diabetes and insufficient glucose management. ¹ Approximately 81% of patients with T2DM will develop macrovascular disease (MVD) and 58% will develop peripheral arterial or microvascular disease. ² Approximately 25% of all T2DM patients will develop DR. ^3,4  

Advanced DR, especially diabetic macular edema and proliferative diabetic retinopathy (PDR), are among the leading causes of blindness in elderly people, ⁵ and MVD is the main cause of death in this age group. The pathogenetic mechanisms or interactions of active MVD and DR in T2DM are not yet fully understood. ⁶  

Hyperglycemia, hyperlipidemia, insulin resistance, and hypertension correlate positively with both progressive MVD and DR. ^{7 –9} The first visible changes in DR are the formation of microaneurysms and increased vascular permeability due to the loss of pericytes (nonproliferative diabetic retinopathy [NPDR]). ^3,10 Furthermore, progressive microvascular closures and ischemia cause tissue hypoxia with macular edema and/or retinal and iris neovascularization (proliferative diabetic retinopathy, PDR) that may be triggered by local proangiogenic factors. ^11,12  

DR's evolution and its conversion from NPDR to PDR are assumed to involve recruitment and proliferation of retinal vascular endothelial cells, eventually promoted by locally activated cytokines, such as vascular endothelial growth factor (VEGF). ^12,13 VEGF provokes endothelial cell growth and permeability and is thought to mobilize endothelial progenitor cells (EPCs) ¹³ from bone marrow by acting as a chemoattractant protein. 

Various co-risk factors (e.g., diabetes, obesity, hypertension, lipids, metabolic syndrome, and smoking) and medications (e.g. statins, angiotensin-converting enzyme inhibitors [ACEI], angiotensin-1 receptor blockers [ARBs], thiazolidinedione [TZD], and insulin) have been demonstrated to influence formation of EPCs in the bone marrow, release from the bone marrow, their number in the peripheral blood, and their survival in the periphery. ^{14 –17} However, after release from the bone marrow, circulating EPCs are assumed to go directly to the sites of ischemia or neovascularization and initiate new vessel formation, ¹⁸ finally leading to all the severe, late complications of DR, such as macular edema, secondary glaucoma, vitreous bleeding, and retinal detachment. 

Preliminary studies of T2DM patients have shown that EPCs and other vasculogenic factors were elevated in vivo in DR versus non-DR ¹⁹ or differentially altered in the presence of DR and peripheral arterial disease. ²⁰ However, previous studies demonstrated only up to three different stages of DR (e.g., none versus NPDR versus PDR), had rather small sample sizes, and only partly reported the co-risk factors or the influencing medications. 

Therefore, we investigated the association of CPCs, EPCs, and matEPCs adjusted for different stages of NPDR and PDR in a large sample of patients with T2DM. Furthermore, we looked for the possible influence of MVD on the regulation of those angiopoietic cells by comparing cohorts with microvascular disease (DR) alone with combined microvascular disease and MVD cohorts. 

Material and Methods

Patients

One hundred twenty-six patients with T2DM, 66 having MVD and 60 without MVD, were enrolled in our case-control study. Twenty-three of those in the MVD group had coronary heart disease, 43 had peripheral arterial occlusive disease, and 15 had had a stroke. 

All the patients were studied and staged for DR according to the Early Treatment of Diabetic Retinopathy Study (ETDRS) classification. ^{21 –23} For recruitment, patients with T2DM were consecutively enrolled at the outpatients‘ care departments of Internal Medicine and the Department of Ophthalmology at the Rudolf Foundation Clinic, Vienna, Austria. Individuals with other retinal pathologies than DR, including panretinal laser treatment, clinically significant macular edema, vitrectomy, or any anti-VEGF therapy in the past 12 months, were excluded. Patients with a history of systemic other than cardiovascular diseases, malignant diseases, hematologic disorders, an estimated glomerular filtration rate of less than 60 mM or treatment with erythropoietin were excluded. Smokers or former smokers were significantly more prevalent in patients with MVD: 35% vs. 25%. 

The study was approved by the institutional ethics committee and complied with the Declaration of Helsinki, ²⁴ including current revisions, and the Good Clinical Practice guidelines. ^25,26 The procedures followed were in accordance with institutional guidelines, and all subjects gave written informed consent before the beginning of the study. 

Ophthalmic DR Grading

Assessment of DR was performed in all patients according to the ETDRS grading scales. ^{21 –23}  

Patients' pupils were dilated with tropicamide 1%, to obtain a minimum pupil diameter of 5 mm. Color pseudostereoscopic slides were then taken of two fields of both eyes with a 40° retinal camera in a standardized protocol. ⁴ For the temporal field the optic disc was placed at the nasal part of the horizontal meridian of the field of view; for the nasal field, the optic disc was placed one disc diameter inside the temporal edge of the field on the same meridian. Two images of each field in both eyes were then taken with a horizontal shift of 3° to 5°, to get a stereoscopic effect. In sum, eight images of each patient were taken. The overlap of the two fields allowed for a retinal view of 75° to 80° horizontally by 35° to 40° vertically, so that clinically significant pathologic changes in the DR were easy to recognize. In all cases of DR and PDR, fluorescein angiograms (FAs) were performed (Heidelberg Retina Angiograph [HRA-1]; Heidelberg Engineering, Heidelberg, Germany). Several images were taken, after 15 seconds and at 1, 2, 5, and 10 minutes after the injection of 5 mL fluorescein 10% into the antecubital vein. 

Fundus images and angiograms were separately screened by two experienced graders in the same center, using a validated protocol. ⁴ The following stages of DR were differentiated: no retinopathy, meaning no specific lesions of DR (CO; n = 55); mild nonproliferative DR, defined by small hemorrhages and/or hard exudates (mNPDR; n = 19); moderate-severe NPDR, defined by moderate-severe bleeding, hard exudates, venous beadings, or intraretinal microvascular abnormalities (msNPDR; n = 16); early proliferative DR, defined by small neovascularizations on optic disc (NVD) or elsewhere (ePDR; n = 19); and high-risk PDR, defined by large NVD or vitreous hemorrhage with retinal neovascularizations (hrPDR; n = 17). Respective examples of fundus and FA images are shown in Figure 1. 

Figure 1.

 

Fluorescein angiograms of all respective DR stages are depicted. (A) CO, (B) mNPDR, (C) msNPDR, (D) ePDR, and (E) hrPDR. (A) Normal perfusion, with apparently no signs of DR (CO); (B) mNPDR with small hemorrhages and singular hard exudates in the macula. (C) msNPDR, with larger hemorrhages and hard exudates in three to four quadrants, as well as small cotton-wool spots as a sign for mild ischemia. (D) ePDR, shows leakages as signs of neovascularization outside the optic disc as well as panretinal laser scars. (E) hrPDR: the image is blurred by recurrent vitreous hemorrhages due to leakage of neovascularizations covering more than one third of the optic disc area.

Flow Cytometry of Bone Marrow–Derived Circulating Angiopoietic Cells

Various combinations of cell surface markers have been acknowledged to identify progenitor cells involved in vascular biology. ^{27 –29} To be more specific, bone marrow–derived angiopoietic progenitor cells have been demonstrated to be different from circulating endothelial cells, which are mobilized from damaged vessel walls. ³⁰ Most probably, putative EPCs are CD34, CD133, and CD309 positive. ^28,29 Further differentiation can be monitored by time-related higher or lower expression of these latter markers. Indeed, EPCs are further distinguished as early (resting) and late (activated) depending on their expression of adhesion molecules, such as CD31, enabling them to participate in vascular repair. Thus, we differentiate CPCs, EPCs, and mature EPCs (matEPCs) due to their surface receptors. For practical reasons and to be able to study different subsets, we used CD34/CD133 co-expression for all progenitor cells in peripheral blood with possible angioblastic potential—the circulating progenitor cells (CPCs). We used CD34/CD133/CD309 co-expression for the classic EPCs and CD34/CD133/CD309/CD31 for identification of EPCs that express surface adhesion molecules. We chose CD31 on the basis of literature that reported expression on EPCs, which are not yet fully matured endothelial cells. We additionally enumerated CD34/CD133/CD309 triple-positive cells, which are CD31 negative, thus similar to a nonmature state (non-matEPCs). 

One hundred microliters of whole heparinized blood were stained with saturating concentrations of monoclonal antibodies (mAb): fluorescein isothiocyanate (FITC)–conjugated anti-CD31 mAb (clone WM59; BD Biosciences, San Jose, CA), phycoerythrin (PE)-conjugated anti-CD309 (clone 89106; R&D, Minneapolis, MN), PE-cyanin dye 5 (PECY5)-labeled anti-CD34 (clone 581; BD Biosciences), and allophycocyanin (APC)-conjugated anti-CD133 mAb (clone AC133; Miltenyi Biotec, Bergisch Gladbach, Germany) for 45 minutes immediately after blood sampling. We may have found a different number of cells compared with other groups because of the different antibody conjugations that we used. Anti-CD34 mAb is used by most groups FITC, although the general recommendation for stem cell enumeration by ISHAGE (International Society of Hematotherapy and Graft Engineering) ^{31 –34} is to use anti-CD34 in a strong (reddish) color, such as phycoerythrin (PE), to prevent the cells' slipping. All experiments were performed with a 30-minute cleaning of the flow cytometer before acquisition (standard operating procedure at our institution). The lyse-and-wash method reduces debris. All the acquisitions obtained in our analysis accounted for up to 14% of all events. We performed FACS (FACS Calibur; BD Biosciences) of white blood cells (WBCs) after the lyse-and-wash period. The acquisition goal was 1 × 10⁶ events. Results were only processed statistically if at least 5 × 10⁵ events were obtained. Progenitor cells were counted by flow cytometry and expressed in absolute numbers per 10⁶ white blood cells. A gating strategy including four gates was applied to identify the angiopoietic progenitor cells, ³⁵ as reported previously. The negative control was phosphate-buffered saline (PBS) without antibody. All patients were in a fasting state after an overnight fast and did not perform excessive sports within the prior 72 hours. Blood was obtained between 8 and 9 AM. Data were included in the analysis only if the whole procedure (blood sampling, transport, staining, lyse-and-wash procedure, and acquisition on the flow cytometer) had taken less than 5 hours. 

Statistics

Data are shown as the mean ± SD. Differences between the patient groups (two or more) were analyzed by Student's unpaired t-test (two) or by ANOVA (more) or χ²-test. A univariate correlation analysis was used to identify determinants of levels of different subsets of angiopoietic progenitor cells. The variables that were significantly associated with the latter cells were included in multivariate regression analysis. An α-level of P < 0.05 (two-tailed) was statistically significant (all statistical analysis: SPSS 17.0; SPSS Inc., Chicago, IL). 

Results

The patients were well matched. Baseline characteristics were as follows (mean ± STD): age, 63.0 ± 10.2 years; diabetes duration, 12.9 ± 8.3 years; and HbA1c, 8.0% ± 1.6%. 

Baseline Characteristics

No significant differences between DR without MVD and DR with MVD concerning age (63.3 ± 10.2 years vs. 63.4 ± 10.3 years; P = 0.975), diabetes duration (12.8 ± 7.2 years vs. 13.3 ± 9.3 years; P = 0.733), HbA1c level (8.2 ± 1.9% vs. 7.8 ± 1.3%; P = 0.142), systolic blood pressure (144 ± 20 mm Hg vs. 147 ± 23 mm Hg, P = 0.454), diastolic blood pressure (84 ± 13 mm Hg vs. 83 ± 13 mm Hg, P = 0.605), body mass index (BMI; 30.5 ± 5.4 vs. 29.7 ± 5.0, P = 0.459), total cholesterol (185 ± 41 mg/dL vs. 188 ± 40 mg/dL; P = 0.791), LDL cholesterol (97 ± 38 mg/dL vs. 94 ± 36 mg/dL; P = 0.717), HDL cholesterol (47 ± 10 mg/dL vs. 47 ± 15 mg/dL, P = 0.913) and triglyceride (197 ± 109 mg/dL vs. 230 ± 173 mg/dL, P = 0.315) were found. Detailed baseline characteristics of the study population are depicted in Table 1. The patients without MVD with different stages of retinopathy did not differ in any quantitative baseline parameter, whereas those with MVD obtained the same findings apart from their age. Patients with mPDR and msPDR in the MVD group were older, and the age difference was significant (P = 0.026). 

Table 1.

 

Baseline Characteristics According to Different Retinopathy Stages

	CO	mNPDR	msNPDR	ePDR	hrPDR	P *
Retinopathy stages without MVD
Age, y	61.2 ± 10.3	68.3 ± 10.8	69.4 ± 8.9	61.3 ± 10.1	61.0 ± 8.2	0.143
Diabetes duration, y	10.2 ± 7.9	15.7 ± 4.6	17.3 ± 6.6	13.0 ± 7.2	12.3 ± 1.9	0.096
HbA1C, %	8.4 ± 2.1	7.5 ± 1.0	8.6 ± 2.3	7.9 ± 1.2	7.7 ± 1.3	0.751
BMI, kg/m2	30.4 ± 5.1	30.6 ± 3.9	30.5 ± 6.0	32.0 ± 9.0	28.0 ± 3.3	0.917
Systolic blood pressure, mm Hg	145 ± 20	145 ± 16	145 ± 28	137 ± 11	139 ± 21	0.926
Diastolic blood pressure, mm Hg	86 ± 14	83 ± 8	80 ± 12	76 ± 11	87 ± 19	0.498
Total cholesterol, mg/dL	186 ± 44	188 ± 21	201 ± 47	132 ± 0	160 ± 34	0.447
LDL cholesterol, mg/dL	102 ± 39	90 ± 35	105 ± 47	55 ± 0	79 ± 26	0.640
HDL cholesterol, mg/dL	43 ± 8	50 ± 11	55 ± 13	51 ± 0	49 ± 7	0.118
Triglycerides, mg/dL	190 ± 100	239 ± 131	202 ± 143	131 ± 0	161 ± 67	0.819
Retinopathy stages with MVD
Age, y	60.3 ± 9.5	70.9 ± 12.3	68.1 ± 9.7	60.1 ± 4.6	64.6 ± 11.5	0.026
Diabetes duration, y	12.7 ± 10.0	16.5 ± 8.2	6.9 ± 3.9	12.2 ± 8.6	19.7 ± 8.3	0.178
HbA1C, %	7.9 ± 1.4	7.5 ± 0.7	7.7 ± 1.8	7.3 ± 1.0	8.0 ± 1.3	0.677
BMI, kg/m2	29.6 ± 4.7	29.6 ± 2.6	31.0 ± 9.7	29.3 ± 4.5	30.3 ± 5.5	0.978
Systolic blood pressure, mm Hg	146 ± 17	144 ± 22	152 ± 44	156 ± 28	137 ± 21	0.549
Diastolic blood pressure, mm Hg	83 ± 10	84 ± 16	80 ± 21	86 ± 14	75 ± 12	0.633
Total cholesterol, mg/dL	178 ± 39	191 ± 36	215 ± 50	186 ± 42	197 ± 27	0.346
LDL cholesterol, mg/dL	92 ± 33	92 ± 38	111 ± 50	89 ± 46	97 ± 25	0.872
HDL cholesterol, mg/dL	44 ± 14	57 ± 23	49 ± 13	42 ± 6	56 ± 21	0.323
Triglycerides, mg/dL	227 ± 211	209 ± 135	298 ± 158	209 ± 139	209 ± 60	0.896

*  Determined by ANOVA.

Medication with a Possible Influence on Angiopoietic Progenitor Cells

Our patients were carefully treated for diabetes duration of approximately 13 years and an age of at least 6 years. Various medications were used, of which some have been demonstrated to influence the number of angiopoietic progenitor cells, to reach those goals. The use of statins, ARBs, ACEIs, glitazone, and insulin is depicted in Table 2. Although the patients with MVD and more advanced retinopathy needed statins, ARBs or ACEIs more often than patients with lower grades of DR or no MVD, no significant difference was found (Table 2). Only insulin treatment showed a significant difference (P = 0.02), which was attributed to the significant difference of insulin treatment of patients without retinopathy, with (62.1%) or without (23.1%) MVD. 

Table 2.

 

Patient's Medication According to Different Stages of Diabetic Retinopathy and MVD

Retinopathy Stage	No MVD					With MVD					P *
	CO	mNPDR	msNPDR	ePDR	hrPDR	CO	mNPDR	msNPDR	ePDR	hrPDR
Statins	9	5	1	3	1	16	5	2	6	2	0.347
ATBs	5	1	2	0	1	2	1	2	4	2	0.317
ACEIs	10	6	4	1	1	18	6	4	6	2	0.499
Glitazone	2	1	0	0	0	2	0	0	0	2	0.180
Insulin	6	6	4	3	3	18	7	3	4	3	0.020

 Data are the number of patients affected.

*  Determined by χ² test.

Different Progenitor Cells According to Different DR Stages

CPCs, EPCs, and matEPCs were enumerated as described. In Figure 2, respective examples of the identification and gating strategy of vascular progenitor cells are shown. Identification and gating strategy is demonstrated in respective dot plots. The amount of angiopoietic progenitor cells, expressed as number per million white blood cells, is depicted in Table 3. As already published, ²⁰ angiopoietic progenitor cells in patients with MVD but without retinopathy showed a reduction of approximately 30% compared with patients without MVD and retinopathy. 

Figure 2.

 

Respective examples of the identification and gating strategy of vascular progenitor cells. FSC, forward scatter characteristics; SSC, side scatter characteristics; FL2-FL-4, background noise of unstained cells with PBS; FITC, fluorescein isothiocyanate; PE, phycoerythrin; PECY5, phycoerythrin-cyanin dye 5; APC, allophycocya-nin. Plots 1–4: negative control with PBS but without antibody (10.000 events acquired). Plots 5–8: ungated dot plots of 1 million events with the respective monoclonal antibodies. Plots 9–11: all cells, Life Gate and then Leuco Gate on the respective cells. Plot 12: cells used for further analysis resulting after Life and Leuco Gate. Plot 13: identification of CD34/PECY5-positive cells. Plot 14: identification of CD133- and CD34-positive cells, marked by the frame. Plot 15: presentation of CD34/CD133 double positive cells. In this example, virtually all cells are CD309 positive, one part of them is CD31 negative, the other (depicted in bold points) is CD31/FITC positive.

Table 3.

 

Number of Different Progenitor Cells According to Retinopathy Stage

	CO	mNPDR	msNPDR	ePDR	hrPDR	P *
Retinopathy stages without MVD
CPC, n	1079 ± 958	322 ± 121	451 ± 313	559 ± 473	269 ± 141	0.008
EPC, n	155 ± 86	135 ± 52	117 ± 71	139 ± 113	145 ± 79	0.840
MatEPC, n	35 ± 30	67 ± 56	51 ± 49	80 ± 102	64 ± 71	0.275
EPC, % of CPCs	20 ± 12	45 ± 18	38 ± 23	34 ± 32	56 ± 26	<0.001
MatEPC, % of all EPCs	20 ± 11	45 ± 18	40 ± 28	42 ± 33	37 ± 30	0.017
Retinopathy stages with MVD
CPC, n	679 ± 285	521 ± 276	510 ± 329	407 ± 241	398 ± 329	0.037
EPC, n	99 ± 69	69 ± 14	59 ± 24	37 ± 16	21 ± 18	<0.001
MatEPC, n	22 ± 25	14 ± 11	8 ± 6	4 ± 3	1 ± 1	0.010
EPC, % of CPCs	17 ± 12	18 ± 12	14 ± 7	11 ± 5	5 ± 2	0.028
MatEPC, % of all EPCs	18 ± 12	18 ± 12	14 ± 7	11 ± 5	5 ± 4	0.008

*  Determined by ANOVA.

The T2DM patients without MVD, CPCs declined from CO to mNPDR, then increased in PDR, and finally decreased in hrPDR (P = 0.008). EPCs showed a decline from CO to msNPDR and then increased stepwise. MatEPCs showed an increase from CO to mNPDR, a decrease in msNPDR, an increase in ePDR, and a decrease in hrPDR. However, changes in EPCs and mature EPCs did not reach statistical significance. Changes in subfractions of angiopoietic progenitor cells are depicted in Table 3. 

In T2DM patients with MVD, the CPCs, EPCs, and matEPCs showed a significant, stepwise decline with advancing stages of retinopathy. In contrast to the results in our previous paper ³¹ and to the findings in the patient group without MVD, percentages of more differentiated (EPCs, matEPCs) subfractions of angiopoietic progenitor cells showed a dramatic decrease. In comparison to patients without MVD, in which EPCs and matEPCs reached up to 56% and 37% of CPCs and EPCs, percentages of EPCs and matEPCs were reduced to 5% each, which is an 11- and 7-fold difference, respectively. 

Influence of (Ex-)Smoking on Angiopoietic Progenitor Cells

We did not find a significant difference in angiopoietic progenitor cells in patients without MVD (P = 0.251) in the five stages of retinopathy (CO, mNPDR, msNPDR, ePDR, and hrPDR). The presence of the cells in the MVD patient group reached weak significance (P = 0.045). A multivariate regression was applied to exclude the influence of smoking on various numbers and percentages of angiopoietic progenitor cells (Table 4). For each parameter of angiopoietic progenitor cells, the observed associations with stages of retinopathy were challenged by integrating the parameter (ex)-smoking into the models. Associations of angiopoietic progenitor cells with stages of retinopathy remained unchanged, apart from matEPCs and the percentage of matEPCs in patients without MVD (Table 4). 

Table 4.

 

Influence of Smoking on Various Numbers of Angiopoietic Cells

	β	P *	β	P *	β	P *
No MVD
CPC	−0.373	0.003	−0.287	0.045	−0.236	0.126
EPC	−0.076	0.565	−0.110	0.434	−0.091	0.571
MatEPC	0.234	0.072	0.193	0.162	0.211	0.172
% EPCs	0.459	<0.001	0.320	0.017	0.349	0.017
% MatEPCs	0.309	0.017	0.211	0.127	0.206	0.175
Adjustment	None		(Ex-)Smoking		Insulin treatment
MVD
CPC	−0.374	0.002	−0.310	0.017	−0.310	0.020
EPC	−0.532	<0.001	−0.504	<0.001	−0.510	<0.001
MatEPC	−0.432	<0.001	−0.446	0.001	−0.426	0.001
% EPCs	−0.371	0.002	−0.369	0.005	−0.333	0.013
% MatEPCs	−0.435	<0.001	−0.450	0.001	−0.393	0.004
Adjustment	None		(Ex-)Smoking		Insulin treatment

*  Determined by ANOVA.

Influence of Insulin Treatment on Angiopoietic Progenitor Cells

Since insulin treatment was significantly different in the patient groups (Table 2), and insulin was supposed to influence the number of angiopoietic progenitor cells, we analyzed insulin treatment in our patient groups in detail. The observed overall significant difference was attributed to the significant difference in insulin treatment of patients without retinopathy with (62.1%) or without (23.1%) MVD. Thus, insulin treatment should not have affected the number of angiopoietic progenitor cells in patients with retinopathy. To exclude any influence of insulin treatment on the various numbers and percentages of angiopoietic progenitor cells, multivariate regression was applied (Table 4). For each parameter of angiopoietic progenitor cells, the observed association with stage of retinopathy was adjusted by insulin treatment. Our observed associations were only modified from significant to nonsignificant in the patients without MVD for CPCs, matEPCs, and the percentage of matEPCs. The significant associations in patients with MVD remained unchanged (Table 4). 

Discussion

We demonstrated that different circulating angiopoietic cells such as CPCs (CD34/CD133), EPCs (CD34/CD133/CD309), and matEPCs (CD34/CD133/CD309/CD31) in T2DM patients with DR had very different regulations in number, apparently depending on co-existing MVD. Patients with MVD showed a dramatic retinopathy-stage-dependent depletion of all angiopoietic cells. In contrast, in patients without MVD, EPCs, and matEPCs reached up to 56% and 37% of CPCs and EPCs. 

Recent studies in patients with T2DM patients have investigated the potential role of bone marrow–derived progenitor cells in proliferative DR ^25,26,36 versus nonproliferative DR. ^12,25 Lee et al. ¹⁹ measured CD117 and CD34 single-positive cells in the peripheral blood by flow cytometry. They were the first to show that CD34-positive cells may be increased in NPDR and PDR versus diabetic patients without DR. 

In another study, Fadini et al. ²⁰ demonstrated that angiopoietic cells were differentially altered in the presence or absence of DR and peripheral artery disease (PAD). ¹⁶ They showed an enhanced endothelial differentiation, together with a high CD34/CD309 cell proportion in patients with DR and a poor endothelial differentiation in patients with PAD. 

Conversely, macrovascular complications were characterized by reduced angiogenesis and exhausted EPCs levels. ³⁷ In fact, an EPC depletion in patients with MVD may be taken as an alternative explanation for the so-called diabetic paradox: the simultaneous appearance of reduced vessel formation in the ischemic macrovascular compartments, together with increased neovascularization in the microvascular beds, such as the retina. ²⁰  

In our recent paper ³¹ we demonstrated a reduction of EPCs in nonproliferative retinopathy. In contrast, patients with hrPDR showed a dramatic increase in EPCs, reflecting a massive amount of neovascularization, leading to visual impairment and loss. It is noteworthy that, in this previous study, patients with type 1 diabetes mellitus (T1DM), were younger and more homogenous (nearly no coincidental cardiovascular risk factors), with no clinical signs of macrovascular disease. 

In our present study, we also found nonsignificant changes of EPCs in T2DM patients without MVD. In contrast, in patients with MVD, we found a significant stepwise reduction in EPCs with advancing stages of retinopathy. According to the “exhaustion theory,” it may well be that the depletion of EPCs is caused by MVD burden. Clinical experience has shown that patients with advanced MVD are often those with advanced microvascular disease (nephropathy, retinopathy) and vice versa. ^2,38  

In addition, the different regulation of EPCs in T2DM may be simply related to the greater heterogeneity of T2DM, compared to T1DM, being influenced by various cofactors, such as coincidental diseases and adequate medication. Therefore, retinopathy in T2DM with MVD could principally be a more serious disease than in T2DM without MVD. Unfortunately, to our knowledge, placebo-controlled, randomized, double-blind trials that investigated the influence of MVD, other co-morbidities, and/or co-medication on the evolution of DR, have not yet been published, either in T1DM or in T2DM. 

Further complicating the matter is that all factors influencing recruitment, proliferation, migration, mobilization, and propagation of EPCs may influence the clinical course of this devastating disease. Maybe the apparent differences in the number of EPCs in patients with MVD and DR do reflect a different cell cycle regulation and function and follow the “bone marrow neuropathy” hypothesis. ³⁹  

Finally, many factors are involved in EPC mobilization from the bone marrow. The most important are VEGF, SDF-1, and NO. Erythropoietin ⁴⁰ could be a chemoattractant of EPCs into the diabetic retina. Any prolonged disturbance of the “balance” of those cytokines may as well be a major cause of the pathogenetic cascade of MVD and DR. 

Although our study is the largest reported study so far, we were unable to directly associate advancing stages of retinopathy with MVD. Neither burden of disease attributed to various affected sites (coronary heart disease, peripheral arterial disease, stroke) nor time from the last cardiovascular event was associated with the measured number of EPCs. 

In T2DM patients with only microvascular disease, we had the impression on reviewing all our data on the various subsets of angiopoietic progenitor cells (CD34⁺/CD133⁺; CD34⁺/CD133⁺/CD309⁺; CD34⁺/CD133⁺/CD309⁻; CD34⁺/CD133⁺/CD309⁺/CD31⁺; CD34⁺/CD133⁺/CD309⁺/CD31⁻; percentage of EPCs/CPCs; percentage of non-EPCs/CPCs; percentage of matEPCs/EPCs; and percentage of non-matEPCs/EPCs) that overall, the kinetics were similar to our findings in T1DM patients. As depicted in Figure 3, we had a high level of interindividual variance in T2DM patients without MVD; thus, the differences did not achieve statistical significance. Our patients with T2DM without MVD did not show clinically relevant atherosclerosis, nor did they have positive stress tests. Nevertheless, the higher level of variance might be attributable to the fact, that T2DM per se is a much more heterogeneous disease than T1DM. 

Figure 3.

 

The number of matEPCs (identified by CD31 expression on EPCs) per 1 million white blood cells (WBC) correlated with the advancing stages of DR, depending on co-existing MVD.

Alternatively, co-medication for cardiovascular risk factors and diseases may affect the number of EPCs, T2DM patients with MVD and more advanced retinopathy more often needed statins, ARBs, or ACEIs than did patients with lower grades of DR or no MVD; however, the differences were not significance (Table 2). More patients with MVD than without MVD needed insulin treatment, but this effect was only produced in the CO group (n = 7 vs. 18; P = 0.009). In patients with MVD, insulin treatment affected only the association of non-EPCs to different patient groups. In contrast, in the patient group without MVD, insulin treatment affected the association of most of the angiopoietic progenitor cells (CPCs, non-EPCs, matEPCs, percentage of matEPCs, and percentage of non-matEPCs). Likewise, Humpert et al. ⁴¹ have demonstrated an augmentation of EPCs by insulin treatment. The joint application of various medications cannot be a major factor for the observed decrease in EPCs in patients with MVD, as it has been shown that insulin, ⁴¹ antihypertensive, or lipid-lowering medications increase EPCs in nondiabetic and diabetic patients. 

Recent advances further extend the hypothesis of neovascularization caused by vascular progenitors: multiple cell types that formerly competed in human and animal studies with each other for being the “true” EPCs, could play important roles in the numerous stages of neovascularization. ⁴² Not all those cell types may in fact be the cells that are ingrown in the vessel walls, becoming new endothelial cells. In addition, not only does the diversity of different endothelial or endothelial-hematopoietic lineages complicate the analysis, but also the different stages of differentiation of those cell types. A limitation of our study is that, due to the use of four-color flow cytometry, we could investigate only one of those putative vascular progenitor lineages, but we analyzed them on three distinct differentiation steps. Another limitation of our study is the nondetermination of the functionality of the progenitor cells analyzed in our patients. An ideal protocol for studying EPC biology would be a starting population of human, cell-sorted, >99% pure, ex vivo EPCs that are cultured under different conditions and subjected to different experiments. 

Recently, interesting findings from two recent large studies ^43,44 indicate that treatment with fenofibrate reduces the progression of retinopathy ⁴³ or the need for laser treatment ⁴⁴ in patients with DR. In the ACCORD study, the preventive effect of fenofibrate occurred in concert with a decrease in triglyceride levels by 20 mg/dL, whereas in the FIELD study, the positive effect of fenofibrate was not related to lipid levels. In the patients in our study, LDL cholesterol levels were relatively low in most subgroups (Table 1), and 39.7% of all patients received statins. Fenofibrate was not used in our patients, since in all guidelines for CVD risk reduction statins and not fibrates were recommended. 

In summary, we and others have demonstrated that angiopoietic progenitor cells have a role in nonproliferative and proliferative DR. Thus, medications that influence recruitment, proliferation, migration, mobilization, and propagation of EPCs may influence the clinical course of this devastating disease. Unfortunately, to our knowledge, placebo-controlled, randomized, double-blind trials that investigated a positive influence of such a strategy have not been published. 

However, although medication studies are not available, our and other studies evoke the hypothesis that retinopathy disease monitoring could be supported by repetitive, regular measurements of angiopoietic progenitor cells. State-of-the-art clinical management of retinopathy is available in limited places in the world (mostly, in industrialized countries). With the ongoing simplification of flow cytometry, the technology is now available in out-of-hospital community care laboratories. Thus, with all published data in mind, joint-efforts should be undertaken to establish enumeration of EPCs as a diagnostic tool that could enable frequent disease monitoring, possibly saving time and costs. International multicenter studies are needed to obtain standard values and/or standard changes in values, respectively, for clear identification of advanced DR. 

To summarize, in our sample of T2DM patients with DR, a dramatic stage-related reduction in EPCs was observed in patients with MVD. In the patients' group without clinically detectable MVD, a different regulation of EPCs was found. That may be related to the heterogeneity of T2DM, influenced by various co-factors, such as co-occurring diseases and adequate medication. 

Footnotes

 Supported by Austrian Diabetes Association, Price Research Grant 2006.

Footnotes

 Disclosure: S. Brunner, None; F. Hoellerl, None; K.E. Schmid-Kubista, None; F. Zeiler, None; G. Schernthaner, None; S. Binder, None; G.-H. Schernthaner, None

The authors thank the Department of Ophthalmology for expendable items, especially photographic articles, and Peggy D. Lamar for her careful editing of the English.