September 2011
Volume 52, Issue 10
Free
Retina  |   September 2011
Elevated Systemic Neutrophil Count in Diabetic Retinopathy and Diabetes: A Hospital-Based Cross-sectional Study of 30,793 Korean Subjects
Author Affiliations & Notes
  • Se Joon Woo
    From the Departments of Ophthalmology and
  • Seong Joon Ahn
    the Department of Ophthalmology, Seoul National University College of Medicine, Seoul National University Hospital, Seoul, Korea.
  • Jeeyun Ahn
    From the Departments of Ophthalmology and
  • Kyu Hyung Park
    From the Departments of Ophthalmology and
  • Kiheon Lee
    Family Medicine, Seoul National University College of Medicine, Seoul National University Bundang Hospital (SNUBH), Seongnam, Korea; and
  • Corresponding author: Se Joon Woo, Department of Ophthalmology, Seoul National University Bundang Hospital, 300, Gumi-dong, Bundang-gu, Seongnam, Gyeonggi-do 463-707, Korea; sejoon1@daum.net
Investigative Ophthalmology & Visual Science September 2011, Vol.52, 7697-7703. doi:10.1167/iovs.11-7784
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to Subscribers Only
      Sign In or Create an Account ×
    • Get Citation

      Se Joon Woo, Seong Joon Ahn, Jeeyun Ahn, Kyu Hyung Park, Kiheon Lee; Elevated Systemic Neutrophil Count in Diabetic Retinopathy and Diabetes: A Hospital-Based Cross-sectional Study of 30,793 Korean Subjects. Invest. Ophthalmol. Vis. Sci. 2011;52(10):7697-7703. doi: 10.1167/iovs.11-7784.

      Download citation file:


      © 2015 Association for Research in Vision and Ophthalmology.

      ×
  • Supplements

Purpose. To evaluate the association between systemic parameters including inflammatory markers and diabetic retinopathy (DR).

Methods. This was a cross-sectional study enrolling 30,793 persons who visited a health care center for a medical checkup. Diabetic patients were classified into five DR groups: no DR; mild, moderate, or severe nonproliferative DR, and proliferative DR. A full laboratory workup and comprehensive medical data on the subjects were obtained and used for analysis.

Results. The mean (SD) age of the participants was 47.4 (11.9) years (range, 18–90) and the male-to-female ratio was 55.7:44.3. The prevalence of diabetes and DR were 6.6% (2,020/30,793) and 5.3% (108/2,020), respectively. Among inflammatory markers, the mean absolute neutrophil count (ANC; per microliter) was significantly higher in the DR than in the non-DR group (3900 vs. 3566; P = 0.0143) and in diabetic than in nondiabetic subjects (3583 vs. 3262; P < 0.0001). Subjects in the fourth quintiles of ANC showed 2.7 odds of having DR by multivariate analysis, and there was a linear trend in the odds ratios according to increasing ANC levels. The level of ANC and the ANC/leukocyte ratio also demonstrated a linearly increasing trend with the severity of DR, even after adjustment for other clinical factors, including HbA1c. Among significant risk factors of DR, ANC showed the second strongest predictive power for DR (AUROC = 0.590) after HbA1C (0.624).

Conclusions. Elevated systemic neutrophil count is associated with the presence and severity of DR as well as diabetes. This result indicates that systemic subclinical inflammation is related with DR, and neutrophil-mediated inflammation may play an important role in the pathogenesis of DR.

Diabetic retinopathy (DR) is the leading cause of blindness among the working-age population in the United States. 1 Studies have estimated that nearly all patients with type 1 diabetes and more than 60% with type 2 diabetes have some degree of DR 20 years after the onset of diabetes. 2 The 14-year incidences of visual impairment and blindness in a diabetic population were reported to be 12.7% and 2.4%, respectively. 3 The early detection of DR and control of risk factors are crucial for the prevention of DR-related visual loss. 
There is increasing evidence that even established risk factors of DR, including hyperglycemia, hypertension, and duration of diabetes, explain only a limited proportion of the risk. 4,5 Inflammation and endothelial dysfunction have been hypothesized to be key mechanisms in diabetes-related vascular complications. 5 Inflammation is a nonspecific response to injury that includes a variety of functional and molecular mediators, including recruitment and activation of leukocytes. 6 Most of the molecular and functional changes that are characteristic of inflammation have been found in retinas of diabetic animals and humans. 6,7 Although the role of inflammatory processes in the development of DR has been consistently reported in animal studies, it has been neither well-proven nor strongly demonstrated in human clinical studies. 6, 8  
There have been attempts to find systemic biomarkers associated with DR, most of which have been related to inflammation and hemostatic disturbances. 9, 11 Systemic markers, such as C-reactive protein (CRP), fibrinogen, homocysteine, and vascular cell adhesion molecules, have shown inconsistent results and only limited associations with DR after controlling for other established risk factors. 10,12 Among them, CRP, the most well-known inflammatory marker, has been reported to be associated with DR by some research groups. 11,13,14 However, more recent studies failed to replicate the association between CRP and DR. 9,10 Moreover, Lim et al. 15 showed that a high level of CRP had a protective effect against DR. Thus, it remains unclear whether true inflammatory markers related to DR actually exist and whether there is strong clinical evidence that inflammation plays an important role in the pathogenesis of DR in humans. 
In this study, we evaluated the association of systemic inflammatory markers and clinical and laboratory data with DR after controlling for known and possible risk factors, to reveal the true association of inflammation with DR. 
Research Design and Methods
This was a hospital-based, cross-sectional study that focused on the clinical risk factors associated with DR. The study protocol was approved by the Institutional Review Board of Seoul National University Bundang Hospital and complied with the Declaration of Helsinki. 
Participants and Data Collection
Medical records of 30,797 patients who visited a health care center in Seoul National University Bundang hospital for a checkup from July 2004 to June 2008 were reviewed. The medical checkup consisted of a history of systemic diseases and health behavior—for example, smoking, physical examination by physicians, laboratory tests on peripheral blood and urine, and nonmydriatic retinal photography on each eye. Systolic and diastolic blood pressures, height, and weight were measured, and the body mass index (BMI, weight/height2) was calculated. Laboratory tests included complete blood cell count (CBC), hematocrit, white blood cell count (WBC), absolute neutrophil count (ANC), blood glucose, hemoglobin A1c (HbA1c), cholesterol, triglyceride, low-density lipoprotein (LDL), high-density lipoprotein (HDL), thyroid-stimulating hormone (TSH), serum creatinine, C-reactive protein (CRP), serum erythrocyte sedimentation rate (ESR), and urine microalbumin. ANC, ANC/WBC ratio, CRP, and ESR were the targeted inflammatory markers and the main variables of interest. Those who were identified in the health interview survey as having received a diagnosis of diabetes by a health care professional or taking glucose-lowering medicines were classified as known cases of diabetes. Only known cases of diabetes were included in the diabetes group, and the blood glucose and HbA1c levels were not used for diabetes diagnosis. The rationale for defining the diabetes group was that (1) it is difficult to diagnose diabetes from a single glucose measurement and that (2) inclusion of indefinite new-onset diabetic cases in the diabetes group could lead to bias in the analysis of DR, which is caused by chronic hyperglycemia. 
Exclusion criteria were (1) the general systemic diseases that could affect the inflammatory parameters; (2) WBC count >20,000. Extremely high WBC levels (>20,000) were seen in four subjects, of whom three had leukemia and one had pneumonia, and they were excluded from the study. Finally, medical data from 30,793 people were used for statistical analysis. Subjects were grouped according to the presence of diabetes and the severity of DR. 
Diabetic Retinopathy Grading
Retinal photography on each eye was performed by an experienced technician using a nonmydriatic 45° fundus camera (Retinal Camera CR6-45NM; Canon Inc., Tokyo, Japan). A single fundus photograph centered on the fovea was taken for each eye. All retinal photographs of diabetic patients were independently reviewed twice by two ophthalmologists, and the presence and severity of DR were graded based on international clinical DR severity scales proposed by the Global Diabetic Retinopathy Project Group. 16 The five-stage disease severity classification of DR consists of no apparent retinopathy (no DR); mild, moderate, or severe nonproliferative DR (NPDR); and proliferative DR (PDR). Subjects were assigned to the PDR group when retinal neovascularization or panretinal photocoagulation scars were visible on retinal photographs. Diabetic macular edema was diagnosed when hard exudates were observed in the macular area. Any DR was defined to include both NPDR and PDR. When there was disagreement regarding the DR stage between the two ophthalmologists, the photograph was reviewed again by the two ophthalmologists and finally graded after discussion. When two eyes in one patient had different DR severity stages, the more severe stage was allocated to the subject. 
Statistical Analysis
Univariate analyses were performed on demographics, clinical data, and laboratory test results, and the mean values and frequencies were compared according to the presence of diabetes and DR. The significant variables at the 0.1 level of significance on the univariate analyses were used for multiple logistic regression analysis to show independent association with DR or diabetes. ANC levels were categorized into five quintile groups, and we evaluated the relationships between ANC and DR and between ANC and diabetes through stepwise multiple logistic regression analysis. The mean ANC level in each DR severity group was calculated and compared by analysis of variance (ANOVA). The linear trend in ANC levels by the severity of DR was evaluated by using contrast. ANC values were transformed into logarithmic values, since its distribution was nonnormal. A multivariate analysis on ANC levels in DR severity groups was performed by analysis of covariance (ANCOVA). ANC levels between DR groups were compared by using adjusted P values to revise the type 1 error by multiple comparisons (Hochberg's estimation). The ANC/WBC ratios were also analyzed for association with DR. Stepwise multiple regression analysis was performed to find clinical and laboratory factors influencing the ANC and ANC/WBC ratio. The ROC curves were plotted for the prediction of DR. The areas under the ROC (AUROC) curves were calculated before and after the inclusion of ANC as a determining factor for DR. 
Results reaching P < 0.05 were considered statistically significant (SAS 9.1.3; SAS Institute Inc, Cary, NC). 
Results
The mean (SD) age of the participants was 47.4 (11.9) years (range, 18–90 years) and the male-to-female ratio was 55.7:44.3. Of 30,793 subjects, 2,020 (6.6%) were diabetic. The mean age (SD) of the diabetic patients was 55.5 (10.2) years (range, 27–86 years) and the male-to-female ratio was 69.3:30.7. All diabetic patients had type 2 diabetes. The prevalence of any DR among the diabetic patients was 5.3% (108/2020). 
Univariate analysis on clinical and laboratory data according to the presence of diabetes and DR are summarized in Table 1. Diabetes was associated with older age; male sex; smoking; obesity (BMI ≥25.0); high systolic and diastolic BP, ANC, ESR, WBC, blood glucose, HbA1c, hematocrit, cholesterol, triglyceride, LDL, serum creatinine, and urine microalbumin; and low HDL. DR was associated with high systolic BP, ANC, WBC, ANC/WBC ratio, and HbA1c and low hematocrit and triglyceride. Among the inflammatory markers (ANC, WBC, ANC/WBC ratio, CRP, and ESR), ANC, WBC, and ESR were associated with diabetes, and ANC, WBC, and ANC/WBC ratio were associated with DR. Regarding diabetic macular edema, no variables showed a significant association. The number of cases with diabetic macular edema was relatively small (18/108 [16.7%] diabetic patients) to yield statistically significant results. 
Table 1.
 
Clinical and Laboratory Data of 30,793 Subjects with or without Diabetes and 2,020 Diabetic Patients with or without DR
Table 1.
 
Clinical and Laboratory Data of 30,793 Subjects with or without Diabetes and 2,020 Diabetic Patients with or without DR
No Diabetes (n = 28,773) Diabetes (n = 2,020) P No DR (n = 1,912) DR (n = 108) P
Age, y 46.9 ± 11.8 55.5 ± 10.2 <0.0001 55.6 ± 10.3 55.4 ± 8.5 0.8237
Sex, % male 54.7 69.3 <0.0001 69.5 66.7 0.5409
Systolic BP, mm Hg 116.1 ± 14.8 124.0 ± 16.0 <0.0001 123.8 ± 16.0 127.0 ± 17.1 0.0450
Diastolic BP, mm Hg 72.2 ± 11.5 76.5 ± 11.1 <0.0001 76.5 ± 11.1 76.7 ± 11.3 0.8800
Smoking status, %
    Current or past 50.1 62.6 <0.0001 62.9 57.9 0.3259
    Nonsmoker 49.9 37.4 37.1 42.1
Obesity (BMI ≥25.0)
    Yes, % 30.5 47.7 <0.0001 48.2 39.3 0.0725
    No, % 69.5 52.3 51.8 60.7
ANC, count/μL 3,262.4 ± 1,269.3 3,583.8 ± 1,379.7 <0.0001 3,566.0 ± 1,385.0 3,900.2 ± 1,244.9 0.0143
CRP, mg/dL 0.13 ± 0.50 0.13 ± 0.55 0.9151 0.13 ± 0.55 0.18 ± 0.46 0.3749
ESR, mm/hr 9.35 ± 9.11 12.18 ± 11.43 <0.0001 12.14 ± 11.45 12.81 ± 10.98 0.5511
WBC, 103/μL 5.76 ± 1.64 6.31 ± 1.80 <0.0001 6.29 ± 1.80 6.65 ± 1.78 0.0435
ANC/WBC, % 55.74 ± 8.91 56.13 ± 9.52 0.0757 55.99 ± 9.53 58.46 ± 9.05 0.0087
Blood glucose, mg/dL 91.4 ± 10.9 145.9 ± 40.8 <0.0001 145.4 ± 40.1 154.4 ± 51.3 0.0774
HbA1c, % 5.56 ± 0.41 7.47 ± 1.42 <0.0001 7.43 ± 1.40 8.07 ± 1.70 0.0003
Hematocrit, % 43.6 ± 4.1 44.4 ± 4.0 <0.0001 44.5 ± 4.0 43.5 ± 4.0 0.0142
Cholesterol, mg/dL 201.7 ± 35.3 204.0 ± 40.7 0.0147 204.2 ± 40.9 200.7 ± 38.6 0.3957
Triglyceride, mg/dL 119.4 ± 75.8 166.2 ± 134.6 <0.0001 167.3 ± 136.7 147.1 ± 87.6 0.0261
LDL, mg/dL 106.6 ± 26.6 108.1 ± 29.1 0.0346 108.1 ± 29.1 107.1 ± 29.0 0.7336
HDL, mg/dL 58.2 ± 14.4 52.6 ± 12.5 <0.0001 52.5 ± 12.4 53.5 ± 14.0 0.4917
Serum creatinine, mg/dL 1.00 ± 0.19 1.03 ± 0.25 <0.0001 1.03 ± 0.25 1.06 ± 0.30 0.4279
Serum TSH, μIU/mL 1.78 ± 1.99 1.84 ± 2.35 0.2431 1.85 ± 2.40 1.69 ± 1.12 0.1915
Urine microablumin, mg/dL 2.95 ± 10.90 9.18 ± 40.71 <0.0001 9.02 ± 41.57 12.02 ± 20.98 0.3050
Stepwise multivariate logistic regression analysis was performed to find variables having independent association with DR (Table 2, n = 2020). ANC showed a linear trend in odds ratios in the higher ANC quintiles (P = 0.0194). Compared to quintile 1, quintile 4 was the only ANC group showing significant association with DR. The adjusted OR of the 4th quintile group was 2.65 (95% CI, 1.36–5.17), which is the highest value among those of other quintile groups. Other variables showing independent association with DR were HbA1c (OR = 1.312 per 1% increase; P < 0.0001), hematocrit (OR = 0.932 per 1% increase; P = 0.0033), and systolic BP (OR = 1.013 per 1-mm Hg increase; P = 0.0282). 
Table 2.
 
Multiple Logistic Regression Analysis Showing the Parameters with Significant Association with DR (N = 2020). Absolute Neutrophil Count (ANC) Groups Showed a Linear Trend in OR among Quintiles (P = 0.0194)
Table 2.
 
Multiple Logistic Regression Analysis Showing the Parameters with Significant Association with DR (N = 2020). Absolute Neutrophil Count (ANC) Groups Showed a Linear Trend in OR among Quintiles (P = 0.0194)
Variable Adjusted OR 95% CI P
ANC group
    Quintile 1 (≤2486.8) 1.0
    Quintile 2 (>2486.8 to ≤3061.4) 1.473 (0.708–3.066) 0.3000
    Quintile 3 (>3061.4 to ≤3634.2) 1.100 (0.507–2.387) 0.8098
    Quintile 4 (>3634.2 to ≤4486.0) 2.652 (1.360–5.172) 0.0042
    Quintile 5 (>4486.0) 1.851 (0.922–3.716) 0.0835
HbA1c, % 1.312 (1.163–1.480) <0.0001
Hematocrit, % 0.932 (0.889–0.977) 0.0033
Systolic BP, mm Hg 1.013 (1.001–1.025) 0.0282
Table 3 shows the association of diabetes with clinical factors including ANC groups by stepwise multiple logistic regression analysis (N = 30,793). ANC quintile group 5 showed the highest OR among quintile groups (OR = 1.456; P = 0.0294). The linear trend in OR with increasing quintile groups was also significant (P = 0.0226). 
Table 3.
 
Multiple Logistic Regression Analysis Showing the Association of Absolute Neutrophil Count and Various Parameters with Diabetes Mellitus (N = 30,793). ANC Groups Showed a Linear Trend in Odds Ratios among Quintiles (P = 0.0226)
Table 3.
 
Multiple Logistic Regression Analysis Showing the Association of Absolute Neutrophil Count and Various Parameters with Diabetes Mellitus (N = 30,793). ANC Groups Showed a Linear Trend in Odds Ratios among Quintiles (P = 0.0226)
Variable Adjusted OR 95% CI P
ANC group
    Quintile 1 (≤2285.0) 1.0
    Quintile 2 (>2285.0 to ≤2801.6) 1.121 0.791–1.587 0.5028
    Quintile 3 (>2801.6 to ≤3339.4) 1.123 0.799–1.578 0.5037
    Quintile 4 (>3339.4 to ≤4096.0) 1.275 0.904–1.797 0.1655
    Quintile 5 (>4096.0) 1.456 1.038–2.042 0.0294
Age, y 1.051 1.041–1.062 <0.0001
Smoking, current or past 1.373 1.078–1.750 0.0102
ESR, mm/h 0.986 0.974–0.998 0.0206
Blood glucose, mg/dL 1.126 1.117–1.135 <0.0001
HbA1c, % 6.202 5.234–7.349 <0.0001
Hematocrit, % 0.931 0.902–0.961 <0.0001
Cholesterol, mg/dL 0.984 0.981–0.987 <0.0001
The mean ANC and ANC/WBC ratios in the five DR groups (no diabetes; diabetes without DR; mild, moderate, and severe NPDR; or PDR) were calculated and plotted (Fig. 1). Mean ANC (per microliter) increased with increasing severity of DR and the linear trend was statistically significant (3050.9, 3336.8, 3491.9, 3851.8, and 3881.2, respectively, from no diabetes to severe NPDR or PDR; P for linear trend < 0.0001; Fig. 1, left). Intergroup comparisons revealed significant elevation of ANC in the diabetic group versus the nondiabetic group (P = 0.0004) and in the moderate NPDR versus the diabetes without DR group (P = 0.0246). For multivariate analysis, factors related to diabetes were considered for analysis and after stepwise multiple regression analysis, sex, age, smoking, HbA1c, systolic BP, triglyceride, blood glucose, BMI, WBC, cholesterol, serum creatinine, and urine microalbumin were used for adjustment. Mean ANCs also showed a linear trend, with increasing severity of DR (P for trend = 0.0384 by multivariate analysis). ANC/WBC ratios also showed a linear trend with increasing severity of DR (55.01%, 55.17%, 55.24%, 57.70%, and 62.97%, respectively, from no diabetes to severe NPDR or PDR; P for linear trend < 0.0001; Fig. 1, right). The linear trend was maintained after adjustment for sex, age, smoking, HbA1c, systolic BP, triglyceride, blood glucose, BMI, hematocrit, WBC, ESR, cholesterol, serum creatinine, and urine microalbumin (P for trend = 0.0145). Intergroup comparisons revealed a significant elevation of ANC/WBC ratio in severe NPDR or PDR group compared with diabetes group (P = 0.0016), whereas there was no difference in ANC/WBC ratio between the diabetes without DR and the no diabetes group (P = 0.8982). 
Figure 1.
 
The ANC and ANC/WBC ratio according to DR severity group. The logarithmic values for ANC (left) and ANC/WBC ratio (right) showed an increasing trend along with the severity of DR (P for trend < 0.0001). *The adjusted P value to revise the type 1 error from multiple comparisons (Hochberg's estimation).
Figure 1.
 
The ANC and ANC/WBC ratio according to DR severity group. The logarithmic values for ANC (left) and ANC/WBC ratio (right) showed an increasing trend along with the severity of DR (P for trend < 0.0001). *The adjusted P value to revise the type 1 error from multiple comparisons (Hochberg's estimation).
To investigate the systemic parameters influencing ANC and ANC/WBC ratio, stepwise multiple regression analysis was performed (Table 4). High ANC levels were associated with multiple clinical and laboratory factors: diabetes, severe DR, being of the female sex, nonsmoking, low HbA1c, high systolic BP, low triglyceride, high blood glucose, low BMI, high WBC, low cholesterol, high serum creatinine, and high urine microalbumin. High ANC/WBC ratios were associated with severe DR, young age, being female, nonsmoking, low HbA1c, high systolic BP, low triglyceride, high blood glucose, low BMI, low hematocrit, high WBC, high ESR, low cholesterol, high serum creatinine, and high urine microalbumin. 
Table 4.
 
Clinical and Laboratory Factors Related with the ANC and ANC/WBC Ratio by Stepwise Multiple Regression Analysis
Table 4.
 
Clinical and Laboratory Factors Related with the ANC and ANC/WBC Ratio by Stepwise Multiple Regression Analysis
Parameter ANC ANC/WBC Ratio
Estimate SE P Estimate SE P
DR Group
    No diabetes −0.117 0.0585 0.0454 −0.1304 0.0525 0.0131
    Diabetes without DR −0.103 0.0578 0.0741 −0.1137 0.0520 0.0286
    Mild NPDR −0.139 0.0670 0.0381 −0.1627 0.0602 0.0069
    Moderate NPDR −0.085 0.0666 0.2045 −0.1062 0.0598 0.0760
    Severe NPDR or PDR (reference)
Age, y NS −0.0003 0.0001 0.0183
Sex
    Male −0.021 0.0051 <0.0001 −0.0159 0.0052 0.0022
    Female (reference)
Smoking, current or past −0.022 0.0044 <0.0001 −0.0286 0.0040 <0.0001
Systolic BP, mm Hg 0.001 0.0001 <0.0001 0.0005 0.0001 <0.0001
BMI, kg/m2 −0.004 0.0005 <0.0001 −0.0059 0.0005 <0.0001
HbA1c, % −0.018 0.0033 <0.0001 −0.0194 0.0030 <0.0001
Hematocrit NS −0.0011 0.0005 0.0196
WBC, 103/μL 0.197 0.0009 <0.0001 0.0389 0.0008 <0.0001
ESR, mm/h NS 0.0004 0.0002 0.0281
Blood glucose, mg/dL 0.0003 0.0001 0.0074 0.0003 0.0001 0.0028
Cholesterol, mg/dL −0.0002 <0.0001 0.0002 −0.0002 <0.0001 <0.0001
Triglyceride, mg/dL −0.0001 <0.0001 <0.0001 −0.0001 <0.0001 <0.0001
Serum creatinine, mg/dL 0.026 0.0103 0.0119 0.0186 0.0093 0.0462
Urine microalbumin, mg/dL 0.0002 0.0001 0.0314 0.0002 0.0001 0.0077
Table 5 shows the AUROC for the prediction of DR with risk factors statistically significant on multivariate analysis. HbA1c was the strongest predictive factor for DR (AUROC = 0.624), whereas ANC was the second strongest (AUROC = 0.590). Hematocrit (0.573) and systolic blood pressure (0.554) followed ANC. The inclusion of ANC groups in the prediction model resulted in a 2.44% increase in AUROC, although the increase was not statistically significant (P = 0.373). 
Table 5.
 
AUROC for Diabetic Retinopathy of the Predictive Models
Table 5.
 
AUROC for Diabetic Retinopathy of the Predictive Models
AUROC (95% C.I.) Change (%) in AUROC Compared with Model 1 P
HbA1c, % 0.624 (0.602–0.646)
ANC, μL 0.590 (0.568–0.612)
Hematocrit, % 0.573 (0.550–0.595)
Systolic BP, mm Hg 0.554 (0.532–0.576)
Model 1: HbA1c+systolic BP+hematocrit 0.657 (0.635–0.678)
Model 2: HbA1c+systolic BP+hematocrit+ANC 0.661 (0.640–0.682) 0.61 0.602
Model 3: HbA1c+systolic BP+hematocrit+ANC quintile (1–5) 0.673 (0.652–0.694) 2.44 0.373
Discussion
This study showed that a high systemic neutrophil count, along with high HbA1c, high systolic blood pressure, and low hematocrit, have an association with DR. The association between the systemic neutrophil count and DR was independent of glucose control and showed a dose-dependent relationship. 
It is of note that our clinical study demonstrates that the neutrophil, not all leukocytes, has an association with DR. Figure 1 shows that a high neutrophil count is associated with both diabetes and DR. However, high neutrophil/leukocyte count ratios (high ANC/WBC ratio) were associated with DR, not with diabetes, which indicates that neutrophils are specifically associated with development of DR. Although there are numerous experimental studies supporting the role of leukocytes in the pathogenesis DR, there have been few studies indicating neutrophils among leukocytes as being associated with DR development. 6,17, , 20 The exact role of neutrophils in DR pathogenesis may be deduced from animal studies that have been reported so far. 21, 23 Joussen et al. 24 proposed that diabetic neutrophils, through a Fas-Fas ligand-dependent pathway, induce retinal endothelial cell apoptosis and blood–retinal barrier breakdown, which are the key mechanisms of DR. 
It is perplexing that the systemic neutrophil count, a very simple inexpensive laboratory parameter, has not been reported in association with DR or diabetes in the past. In the stepwise multiple regression analysis for absolute neutrophil count (Table 4), systemic neutrophil count, and the ratio of neutrophils to leukocytes are associated with numerous clinical parameters, including sex, age, smoking, glucose control, blood pressure, triglyceride and cholesterol, BMI, hematocrit, ESR, serum creatinine, and urine microalbumin. To elucidate the independent association of neutrophil counts with DR, a statistical analysis adjusted for all these factors and a large study population, as in our study, are necessary. The complicated nature of the association between neutrophil counts and various clinical parameters may have hindered the revelation of significant results in past clinical research involving a relatively small number of study participants. 
Although our data showed neutrophil count to be increased in patients having DR and diabetes and revealed the association between them, the causal relationship is not yet established. It is still unknown whether the elevation of systemic neutrophil count precedes the occurrence of DR and diabetes. However, from the results of the present study and past experiments, we can hypothesize that an increased systemic neutrophil count elevates the local number of neutrophils in retinal vessels, and the increased diabetic neutrophils damage the retinal vascular endothelial cells and the blood–retinal barrier resulting in DR. 18,25,26 On the other hand, chronic hyperglycemia itself could cause elevation of systemic neutrophil counts via proinflammatory mediators, such as NF-κB, IL-6, and TNF-α. 6 Therefore, further research is mandatory to elucidate whether the quantitative increase in the systemic neutrophil count is related to the qualitative abnormality of neutrophils and what causes the increase in systemic neutrophil count in DR and diabetes. 
Of note is that, among the five ANC quintile groups, the 4th quintile group showed the highest odds ratio in association with DR (Table 2). This result indicates that patients with DR are in a state of subclinical inflammation or para-inflammation rather than in an overt inflammatory state, which was recently suggested in type 2 diabetes, metabolic syndromes, Alzheimer disease, and cancer. 27,28 In diabetes, leukocytes and vascular endothelial cells are constantly under noxious stress by glycemic dysregulation. This stress triggers para-inflammation responses in leukocyte and endothelial cells by upregulating inflammatory cytokines and adhesion molecules, which in turn cause abnormal leukocyte–endothelial interactions and ultimately vascular damage. 28  
Previous studies showed CRP is related to diabetes, diabetic macrovascular complications, and neuropathy. 29,30 However, there have been inconsistent results on the association between CRP with DR. 9,11,12,15 In our study, CRP showed no association with DR or diabetes (P = 0.3749 and 0.9151, respectively; Table 1). As previous study subject samples were far smaller than ours and as we could control almost all possible risk factors of DR, we believe the association between CRP and DR is not yet established and that CRP should not be treated as a therapeutic or monitoring target in DR. Furthermore, another recent study showed higher CRP to be associated with a lower prevalence of DR. 15  
As people with advanced stages of DR show poor vision-specific function, detection, and preventive measures at the non-DR or earliest DR stage is important. In our prediction model for DR using parameters that were revealed to be associated with DR, ANC was the second strongest factor after HbA1c. However, inclusion of ANC with other risk factors resulted in only a 2.4% increase of prediction strength. Our result was consistent with the study by Nguyen et al., 12 which showed that inclusion of novel inflammatory and hemostatic markers in the prediction model of DR consisting of traditional risk factors increased the predictive power by only 2.2%. We believe ANC in itself has limited value as a predictive marker for DR and that its association with DR should be viewed as part of the pathogenic mechanism and a possible therapeutic target for DR. 
Our study also revealed additional important information on risk factors of diabetes. In addition to ANC, age, smoking, ESR, glucose control, and hematocrit showed associations with diabetes. History of smoking increased the incidence of diabetes by 37.3%. Interestingly, although diabetic subjects showed higher levels of ESR than nondiabetic subjects in univariate analysis (Table 2), ESR showed an inverse association (OR = 0.986), with diabetes after adjusting for HbA1c and/or blood glucose (Table 3). Therefore, ESR, a nonspecific measure of inflammation, showed no association with diabetes, as well as its microvascular complication, DR, which suggests a low possibility that it has a role in the disease process. 
The prevalence of DR (5.3%) among diabetic patients in our study is relatively lower than that of previous reports. The low prevalence of DR may be due to superior glucose control in the study population since, from the fact that they visited the hospital for health checkup, it may be inferred that these patients are very cautious of their overall health status as well as glucose level control. This notion is also supported by the measured HbA1c level of diabetic patients in our study (mean ± SD = 7.47 ± 1.42%). The second possible cause may be found in the technique of fundus photography acquisition. We took single fundus photographs for each eye rather than seven photographs covering the full peripheral retina. As we could not see the peripheral retina, subjects having mild DR may have been allocated into the no-DR group, with downgrading of the overall DR grade. However, compared to seven-field standard stereoscopic fundus photographs, 1-field 45° nonmydriatic retinal photography has showed perfect agreement for detecting the presence of DR and moderate agreement for DR grading. 31 Therefore, the current analysis on the presence and absence of DR based on single fundus photographs is convincing, and the association of clinical parameters including neutrophil counts and DR is unquestionable. As our study included a large number of subjects, we believe the suspected downgrading of DR did not influence the trend of association and significance of our data. 
There are several limitations to our study. First, the study design was a cross-sectional one, enrolling hospital-based cohort groups, and therefore, we cannot define the causal relationship between neutrophil count and DR and its clinical implication as a marker for the progression of DR. Further longitudinal studies should be performed to identify ANC as a clinical marker for the progression of DR and as a key player in its pathogenesis. In addition, the inclusion of participants who came for a health checkup has the potential of selection bias, and this should be considered when interpreting the result. Second, we performed single-field, nonmydriatic retinal photography for assessing the presence and grading of DR. As explained above, we believe it may have had a limited influence on our result. Third, information on other traditional risk factors for DR such as duration of diabetes and diabetes medication could not be obtained and analyzed in this study. Finally, as we included only Korean subjects, there is a limitation in the generalization of our result to other ethnic groups. Despite these limitations, we believe our study is valuable in that we analyzed many clinical and laboratory parameters in a large number of subjects and showed a new clinical risk factor for DR. 
In conclusion, this clinical study revealed that the elevated systemic neutrophil count is associated with the presence and severity of DR as well as diabetes. Our result indicates that patients with DR are in a subclinical inflammatory state and that neutrophil-mediated inflammation may play an important role in the pathogenesis of DR. 
References
Mohamed Q Gillies MC Wong TY . Management of diabetic retinopathy: a systematic review. JAMA. 2007;298(8):902–916.
Klein R Klein BE Moss SE . The Wisconsin epidemiological study of diabetic retinopathy: a review. Diabetes Metab Rev. 1989;5(7):559–570.
Moss SE Klein R Klein BE . The 14-year incidence of visual loss in a diabetic population. Ophthalmology. 1998;105(6):998–1003.
Frank RN . Diabetic retinopathy. N Engl J Med. 2004;350(1):48–58.
Nguyen TT Wong TY . Retinal vascular manifestations of metabolic disorders. Trends Endocrinol Metab. 2006;17(7):262–268.
Kern TS . Contributions of inflammatory processes to the development of the early stages of diabetic retinopathy. Exp Diabetes Res. 2007;2007:95103.
Elalamy I Chakroun T Gerotziafas GT . Circulating platelet-leukocyte aggregates: a marker of microvascular injury in diabetic patients. Thromb Res. 2008;121(6):843–848.
Lutty GA Cao J McLeod DS . Relationship of polymorphonuclear leukocytes to capillary dropout in the human diabetic choroid. Am J Pathol. 1997;151(3):707–714.
Spijkerman AM Gall MA Tarnow L . Endothelial dysfunction and low-grade inflammation and the progression of retinopathy in type 2 diabetes. Diabet Med. 2007;24(9):969–976.
Klein BE Knudtson MD Tsai MY Klein R . The relation of markers of inflammation and endothelial dysfunction to the prevalence and progression of diabetic retinopathy: Wisconsin epidemiologic study of diabetic retinopathy. Arch Ophthalmol. 2009;127(9):1175–1182.
Izuora KE Chase HP Jackson WE . Inflammatory markers and diabetic retinopathy in type 1 diabetes. Diabetes Care. 2005;28(3):714–715.
Nguyen TT Alibrahim E Islam FM . Inflammatory, hemostatic, and other novel biomarkers for diabetic retinopathy: the multi-ethnic study of atherosclerosis. Diabetes Care. 2009;32(9):1704–1709.
van Hecke MV Dekker JM Nijpels G . Inflammation and endothelial dysfunction are associated with retinopathy: the Hoorn Study. Diabetologia. 2005;48(7):1300–1306.
Schram MT Chaturvedi N Schalkwijk CG Fuller JH Stehouwer CD . Markers of inflammation are cross-sectionally associated with microvascular complications and cardiovascular disease in type 1 diabetes: the EURODIAB Prospective Complications Study. Diabetologia. 2005;48(2):370–378.
Lim LS Tai ES Mitchell P . C-reactive protein, body mass index, and diabetic retinopathy. Invest Ophthalmol Vis Sci. 2010;51(9):4458–4463.
Wilkinson CP Ferris FL3rd Klein RE . Proposed international clinical diabetic retinopathy and diabetic macular edema disease severity scales. Ophthalmology. 2003;110(9):1677–1682.
Schroder S Palinski W Schmid-Schonbein GW . Activated monocytes and granulocytes, capillary nonperfusion, and neovascularization in diabetic retinopathy. Am J Pathol. 1991;139(1):81–100.
McLeod DS Lefer DJ Merges C Lutty GA . Enhanced expression of intracellular adhesion molecule-1 and P-selectin in the diabetic human retina and choroid. Am J Pathol. 1995;147(3):642–653.
Miyamoto K Khosrof S Bursell SE . Vascular endothelial growth factor (VEGF)-induced retinal vascular permeability is mediated by intercellular adhesion molecule-1 (ICAM-1). Am J Pathol. 2000;156(5):1733–1739.
Joussen AM Poulaki V Le ML . A central role for inflammation in the pathogenesis of diabetic retinopathy. FASEB J. 2004;18(12):1450–1452.
Sharma NK Gardiner TA Archer DB . A morphologic and autoradiographic study of cell death and regeneration in the retinal microvasculature of normal and diabetic rats. Am J Ophthalmol. 1985;100(1):51–60.
Freedman SF Hatchell DL . Enhanced superoxide radical production by stimulated polymorphonuclear leukocytes in a cat model of diabetes. Exp Eye Res. 1992;55(5):767–773.
Barouch FC Miyamoto K Allport JR . Integrin-mediated neutrophil adhesion and retinal leukostasis in diabetes. Invest Ophthalmol Vis Sci. 2000;41(5):1153–1158.
Joussen AM Poulaki V Mitsiades N . Suppression of Fas-FasL-induced endothelial cell apoptosis prevents diabetic blood-retinal barrier breakdown in a model of streptozotocin-induced diabetes. FASEB J. 2003;17(1):76–78.
Kelly LW Barden CA Tiedeman JS Hatchell DL . Alterations in viscosity and filterability of whole blood and blood cell subpopulations in diabetic cats. Exp Eye Res. 1993;56(3):341–347.
Harris AG Skalak TC Hatchell DL . Leukocyte-capillary plugging and network resistance are increased in skeletal muscle of rats with streptozotocin-induced hyperglycemia. Int J Microcirc Clin Exp. 1994;14(3):159–166.
Medzhitov R . Origin and physiological roles of inflammation. Nature. 2008;454(7203):428–435.
Xu H Chen M Forrester JV . Para-inflammation in the aging retina. Prog Retin Eye Res. 2009;28(5):348–368.
Goldberg RB . Cytokine and cytokine-like inflammation markers, endothelial dysfunction, and imbalanced coagulation in development of diabetes and its complications. J Clin Endocrinol Metab. 2009;94(9):3171–3182.
Lee CC Adler AI Sandhu MS . Association of C-reactive protein with type 2 diabetes: prospective analysis and meta-analysis. Diabetologia. 2009;52(6):1040–1047.
Vujosevic S Benetti E Massignan F . Screening for diabetic retinopathy: 1 and 3 nonmydriatic 45-degree digital fundus photographs vs 7 standard early treatment diabetic retinopathy study fields. Am J Ophthalmol. 2009;148(1):111–118.
Figure 1.
 
The ANC and ANC/WBC ratio according to DR severity group. The logarithmic values for ANC (left) and ANC/WBC ratio (right) showed an increasing trend along with the severity of DR (P for trend < 0.0001). *The adjusted P value to revise the type 1 error from multiple comparisons (Hochberg's estimation).
Figure 1.
 
The ANC and ANC/WBC ratio according to DR severity group. The logarithmic values for ANC (left) and ANC/WBC ratio (right) showed an increasing trend along with the severity of DR (P for trend < 0.0001). *The adjusted P value to revise the type 1 error from multiple comparisons (Hochberg's estimation).
Table 1.
 
Clinical and Laboratory Data of 30,793 Subjects with or without Diabetes and 2,020 Diabetic Patients with or without DR
Table 1.
 
Clinical and Laboratory Data of 30,793 Subjects with or without Diabetes and 2,020 Diabetic Patients with or without DR
No Diabetes (n = 28,773) Diabetes (n = 2,020) P No DR (n = 1,912) DR (n = 108) P
Age, y 46.9 ± 11.8 55.5 ± 10.2 <0.0001 55.6 ± 10.3 55.4 ± 8.5 0.8237
Sex, % male 54.7 69.3 <0.0001 69.5 66.7 0.5409
Systolic BP, mm Hg 116.1 ± 14.8 124.0 ± 16.0 <0.0001 123.8 ± 16.0 127.0 ± 17.1 0.0450
Diastolic BP, mm Hg 72.2 ± 11.5 76.5 ± 11.1 <0.0001 76.5 ± 11.1 76.7 ± 11.3 0.8800
Smoking status, %
    Current or past 50.1 62.6 <0.0001 62.9 57.9 0.3259
    Nonsmoker 49.9 37.4 37.1 42.1
Obesity (BMI ≥25.0)
    Yes, % 30.5 47.7 <0.0001 48.2 39.3 0.0725
    No, % 69.5 52.3 51.8 60.7
ANC, count/μL 3,262.4 ± 1,269.3 3,583.8 ± 1,379.7 <0.0001 3,566.0 ± 1,385.0 3,900.2 ± 1,244.9 0.0143
CRP, mg/dL 0.13 ± 0.50 0.13 ± 0.55 0.9151 0.13 ± 0.55 0.18 ± 0.46 0.3749
ESR, mm/hr 9.35 ± 9.11 12.18 ± 11.43 <0.0001 12.14 ± 11.45 12.81 ± 10.98 0.5511
WBC, 103/μL 5.76 ± 1.64 6.31 ± 1.80 <0.0001 6.29 ± 1.80 6.65 ± 1.78 0.0435
ANC/WBC, % 55.74 ± 8.91 56.13 ± 9.52 0.0757 55.99 ± 9.53 58.46 ± 9.05 0.0087
Blood glucose, mg/dL 91.4 ± 10.9 145.9 ± 40.8 <0.0001 145.4 ± 40.1 154.4 ± 51.3 0.0774
HbA1c, % 5.56 ± 0.41 7.47 ± 1.42 <0.0001 7.43 ± 1.40 8.07 ± 1.70 0.0003
Hematocrit, % 43.6 ± 4.1 44.4 ± 4.0 <0.0001 44.5 ± 4.0 43.5 ± 4.0 0.0142
Cholesterol, mg/dL 201.7 ± 35.3 204.0 ± 40.7 0.0147 204.2 ± 40.9 200.7 ± 38.6 0.3957
Triglyceride, mg/dL 119.4 ± 75.8 166.2 ± 134.6 <0.0001 167.3 ± 136.7 147.1 ± 87.6 0.0261
LDL, mg/dL 106.6 ± 26.6 108.1 ± 29.1 0.0346 108.1 ± 29.1 107.1 ± 29.0 0.7336
HDL, mg/dL 58.2 ± 14.4 52.6 ± 12.5 <0.0001 52.5 ± 12.4 53.5 ± 14.0 0.4917
Serum creatinine, mg/dL 1.00 ± 0.19 1.03 ± 0.25 <0.0001 1.03 ± 0.25 1.06 ± 0.30 0.4279
Serum TSH, μIU/mL 1.78 ± 1.99 1.84 ± 2.35 0.2431 1.85 ± 2.40 1.69 ± 1.12 0.1915
Urine microablumin, mg/dL 2.95 ± 10.90 9.18 ± 40.71 <0.0001 9.02 ± 41.57 12.02 ± 20.98 0.3050
Table 2.
 
Multiple Logistic Regression Analysis Showing the Parameters with Significant Association with DR (N = 2020). Absolute Neutrophil Count (ANC) Groups Showed a Linear Trend in OR among Quintiles (P = 0.0194)
Table 2.
 
Multiple Logistic Regression Analysis Showing the Parameters with Significant Association with DR (N = 2020). Absolute Neutrophil Count (ANC) Groups Showed a Linear Trend in OR among Quintiles (P = 0.0194)
Variable Adjusted OR 95% CI P
ANC group
    Quintile 1 (≤2486.8) 1.0
    Quintile 2 (>2486.8 to ≤3061.4) 1.473 (0.708–3.066) 0.3000
    Quintile 3 (>3061.4 to ≤3634.2) 1.100 (0.507–2.387) 0.8098
    Quintile 4 (>3634.2 to ≤4486.0) 2.652 (1.360–5.172) 0.0042
    Quintile 5 (>4486.0) 1.851 (0.922–3.716) 0.0835
HbA1c, % 1.312 (1.163–1.480) <0.0001
Hematocrit, % 0.932 (0.889–0.977) 0.0033
Systolic BP, mm Hg 1.013 (1.001–1.025) 0.0282
Table 3.
 
Multiple Logistic Regression Analysis Showing the Association of Absolute Neutrophil Count and Various Parameters with Diabetes Mellitus (N = 30,793). ANC Groups Showed a Linear Trend in Odds Ratios among Quintiles (P = 0.0226)
Table 3.
 
Multiple Logistic Regression Analysis Showing the Association of Absolute Neutrophil Count and Various Parameters with Diabetes Mellitus (N = 30,793). ANC Groups Showed a Linear Trend in Odds Ratios among Quintiles (P = 0.0226)
Variable Adjusted OR 95% CI P
ANC group
    Quintile 1 (≤2285.0) 1.0
    Quintile 2 (>2285.0 to ≤2801.6) 1.121 0.791–1.587 0.5028
    Quintile 3 (>2801.6 to ≤3339.4) 1.123 0.799–1.578 0.5037
    Quintile 4 (>3339.4 to ≤4096.0) 1.275 0.904–1.797 0.1655
    Quintile 5 (>4096.0) 1.456 1.038–2.042 0.0294
Age, y 1.051 1.041–1.062 <0.0001
Smoking, current or past 1.373 1.078–1.750 0.0102
ESR, mm/h 0.986 0.974–0.998 0.0206
Blood glucose, mg/dL 1.126 1.117–1.135 <0.0001
HbA1c, % 6.202 5.234–7.349 <0.0001
Hematocrit, % 0.931 0.902–0.961 <0.0001
Cholesterol, mg/dL 0.984 0.981–0.987 <0.0001
Table 4.
 
Clinical and Laboratory Factors Related with the ANC and ANC/WBC Ratio by Stepwise Multiple Regression Analysis
Table 4.
 
Clinical and Laboratory Factors Related with the ANC and ANC/WBC Ratio by Stepwise Multiple Regression Analysis
Parameter ANC ANC/WBC Ratio
Estimate SE P Estimate SE P
DR Group
    No diabetes −0.117 0.0585 0.0454 −0.1304 0.0525 0.0131
    Diabetes without DR −0.103 0.0578 0.0741 −0.1137 0.0520 0.0286
    Mild NPDR −0.139 0.0670 0.0381 −0.1627 0.0602 0.0069
    Moderate NPDR −0.085 0.0666 0.2045 −0.1062 0.0598 0.0760
    Severe NPDR or PDR (reference)
Age, y NS −0.0003 0.0001 0.0183
Sex
    Male −0.021 0.0051 <0.0001 −0.0159 0.0052 0.0022
    Female (reference)
Smoking, current or past −0.022 0.0044 <0.0001 −0.0286 0.0040 <0.0001
Systolic BP, mm Hg 0.001 0.0001 <0.0001 0.0005 0.0001 <0.0001
BMI, kg/m2 −0.004 0.0005 <0.0001 −0.0059 0.0005 <0.0001
HbA1c, % −0.018 0.0033 <0.0001 −0.0194 0.0030 <0.0001
Hematocrit NS −0.0011 0.0005 0.0196
WBC, 103/μL 0.197 0.0009 <0.0001 0.0389 0.0008 <0.0001
ESR, mm/h NS 0.0004 0.0002 0.0281
Blood glucose, mg/dL 0.0003 0.0001 0.0074 0.0003 0.0001 0.0028
Cholesterol, mg/dL −0.0002 <0.0001 0.0002 −0.0002 <0.0001 <0.0001
Triglyceride, mg/dL −0.0001 <0.0001 <0.0001 −0.0001 <0.0001 <0.0001
Serum creatinine, mg/dL 0.026 0.0103 0.0119 0.0186 0.0093 0.0462
Urine microalbumin, mg/dL 0.0002 0.0001 0.0314 0.0002 0.0001 0.0077
Table 5.
 
AUROC for Diabetic Retinopathy of the Predictive Models
Table 5.
 
AUROC for Diabetic Retinopathy of the Predictive Models
AUROC (95% C.I.) Change (%) in AUROC Compared with Model 1 P
HbA1c, % 0.624 (0.602–0.646)
ANC, μL 0.590 (0.568–0.612)
Hematocrit, % 0.573 (0.550–0.595)
Systolic BP, mm Hg 0.554 (0.532–0.576)
Model 1: HbA1c+systolic BP+hematocrit 0.657 (0.635–0.678)
Model 2: HbA1c+systolic BP+hematocrit+ANC 0.661 (0.640–0.682) 0.61 0.602
Model 3: HbA1c+systolic BP+hematocrit+ANC quintile (1–5) 0.673 (0.652–0.694) 2.44 0.373
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×