Red cell distribution width as a significant indicator of medication and prognosis in type 2 diabetic patients

Whether red cell distribution width (RDW) can be a potential indicator for diabetic nephropathy (DN) is unknown. A total of 809 type 2 diabetes mellitus (T2D) patients were divided into 4 groups according to the quartiles (Q) of the RDW (%): Q1 ≤ 12.4 (n = 229), 12.4 < Q2 ≤ 12.9 (n = 202), 12.9 < Q3 < 13.5 (n = 168), Q4 ≥ 13.5 (n = 210). Results showed that the levels in Q4 group was higher in age, disease duration, systolic blood pressure, blood urea nitrogen, creatinine, uric acid and proteinuria but lower in hemoglobin, serum albumin and glycosylated hemoglobin compared to Q1 group. Furthermore, the incidences of DN, diabetic peripheral neuropathy, hypertension and coronary heart disease in the Q3 or Q4 group were higher compared to Q1 group. Medications including calcium channel blockers and antiplatelet therapy also showed higher frequencies in Q3 or Q4 group compared to Q1. Logistic regression indicated that the antiplatelet therapy (OR = 2.065), hypertension (OR = 2.819), creatinine (OR = 4.473) and proteinuria (OR = 2.085) were positively associated with level of Q4 group, but higher hemoglobin (OR = 0.021) and serum Ca2+ (OR = 0.178) were negatively associated with Q4. This data suggest that high level of RDW in T2D patients indicates a higher risk and a poor prognosis for DN.

Diabetes mellitus (DM) is a metabolic disorder caused either by the insufficient production of insulin in islet cells of the pancreas or by resistance against secreted insulin in tissues, leading to an elevation in the glucose concentration in the blood. In China, a national survey in 2010 showed a DM prevalence of 9.65%, with the total number of DM patients up to 92.4 million and accounting for a quarter of worldwide DM patients in the population aged 20 to 79 years. Several studies have shown that plasma cholesterol levels, blood pressure, microalbuminuria and hyperglycemia are closely associated with the progression of DM 1, 2 . Additionally, recent studies have revealed serum uric acid level, carboxy-terminal propeptide and retinal venular diameter as significant indicators of diabetic complications such as diabetic nephropathy or retinopathy [3][4][5] . RDW, defined as the heterogeneity of circulating erythrocytes (anisocytosis), was used to distinguish the variable pathogenesis of anemia together with the MPV 6 . Malnutrition, including Fe deficiency and lack of vitamin B 12 and folic acid, generates elevated RDW 7 . Recent studies have demonstrated that RDW may also be an effective predictor of morbidity and mortality in various diseases such as PH 8,9 , NAFLD 10 , CHD [11][12][13][14][15] , stoke 16 , atherosclerosis 17 , prevalent dementia 18 , IBD 19 , ESRD 20 and heart failure 21 . A study following 8175 adults for up to 6 years showed that the measurement of RDW may be used to predict mortality in CVD, cancer and other diseases 22 in the early stages. Moreover, in a 5.5-year follow-up of 13039 patients diagnosed with PAD, the 1% increase of RDW was accompanied by the increased 10% in all-cause mortality. Also, RDW was considered as a prognostic marker in PAD patients 23 , which was extraordinarily higher in metabolic syndrome (MS) patients compared to those without MS 24 .
Recently, some studies have shown that increased RDW is associated with the incidence of DM 6,[25][26][27] . However, the explicit relationship between RDW and the basic indexes, drug treatment and related complications (such as chronic heart disease and diabetic retinopathy) remains implicit in T2D patients. Here, we study the characteristics of RDW and its association with distributions of clinical indexes in T2D patients.
Laboratory results of blood and urine in type 2 diabetic patients. Compared to the Q1 group, in the Q4 group, the T-test or the non-parametric Kruskal-Wallis test showed that Hb (g/dL), Alb (g/L) and HbA1c (%) levels were significantly lower, but BUN (mmol/L), serum Ca 2+ (mmol/L), Cr (μmmol/L), UA (mmol/L), Upro (mg/L) and HbA1c (%) levels were significantly higher ( Table 2 and  Complications and medications of the Q1 to Q4 groups in type 2 diabetic patients. Compared to the Q1 group, the patients in the higher Q3 or Q4 groups had increasing morbidities of DR, HTN and using CCB. The Q3 group had a higher rate of DPN compared to the Q1 group (61.9% vs 46.7%, P < 0.05). Moreover, groups Q2 and Q4 had more morbidities of CHD compared to the Q1 group (P < 0.05), and Q4 group used less OHA than Q1 group (P < 0.05). (Table 3 and Fig. 3a-f). No significant differences were found in either category of patient treatment, with or without ACE-I or ARB, β-Blocker, lipid-lowering agents, or insulin (P > 0.05).  Binary Logistic regression models for RDW. As shown in

Discussion
Recently, the increasing prevalence of DM has become a global health problem. Disclosed risk factors include age, family history of diabetes, obesity, hypertension and high triglycerides 28 . In China, the crude and age-standardized prevalence of DM are 12.19% and 6.98%, respectively 29 . Several studies have focused on prognostic biomarkers that could indicate the incidences of coronary artery spasm (CAS), acute ischemic stroke (AIS), cardiovascular disease (CVD) and diabetic nephropathy (DN) in DM patients. Microalbuminuria is a remarkable biomarker for the diagnosis of DN 30 . Inflammatory biomarkers including WBC, TNF-α, matrix metalloproteinases (MMT) and dysglycemia are expressed concurrently to fit with the early stages of CVD in patients with diabetes 31 . Low high-sensitivity C-reactive protein (hs-CRP) levels in patients with DM are associated with a high risk of CAS 32 . The higher copeptin levels in the upper inter-quartile group (Q4 > 17.1 pmol/L) were associated with a higher death risk in short-term stroke prognosis in patients with T2D and stroke 33 . Currently, the significant indicators involved in the development or the prognosis of DM are still not fully understood.
In this study, we demonstrated that changes in RDW are associated with T2D. Disease duration and SBP showed positive associations with the higher Q4 group of RDW, but other clinical indexes, including serum Ca 2+ , Hb or HDL, were negatively associated with the Q4 group of RDW. One retrospective study 22 indicated that RDW is an age-associated biomarker in people >45 years old. Here, we also found that elderly patients had significantly higher Q4 values of RDW than younger patients, P < 0.05 (Fig. 1a). Furthermore, the regression model showed that age was positively associated with RDW (Table 4). One possible explanation involves the greater potential of seniors to be in a state of inflammation, nutritional deficiency and other complications. Another retrospective study 6 that included 260 T2D patients and 44 healthy control subjects found that RDW was correlated with BMI. However, our study showed that the changes of RDW value were not significantly associated with the BMI levels in diabetic patients. Additionally, in the present study, SBP in the Q4 group was obviously higher compared to the  Q1 group (Fig. 1c) and HTN was positively associated with RDW as detected by the regression model (Table 4), which was consistent with reports by Dada, O. A. et al. 34 . However, Malandrino N's study indicated that the duration of DM had no significant association with RDW, which contradicts our study that indicates that the higher levels of RDW are positively associated with the longer duration in T2D patients (Table 4 and Fig. 1b) 35 . The levels of RDW were measured in 26709 non-diabetic subjects over a 14-year follow-up period in a report by Engstrom G, showing that RDW was positively correlated to HbA1c, indicating that the HbA1c would increase by 0.10% per Standard Deviation (SD) elevation in RDW 26 . In our study, the HbA1c in the Q4 group showed a significant decrease compared to Q1 group (Fig. 2h). Recently, Lippi G et al. 36 showed a negative correlation between RDW and Hb in 4874 outpatients, which was consistent with our results (Fig. 2a). Additionally, a negative association was identified between RDW and HDL-cholesterol in the multivariate Logistic regression after adjustment for age, Hb, and MCV, which was different from our study, which showed no obvious discrepancy between HDL and RDW. The reason may be related to the fact that the patients in our study were receiving appropriate treatment, including control of lipidemia, because all indicators such as TG, TC and LDL in lipidemia showed no differences (P > 0.05). On the other hand, we also concluded that serum Ca 2+ was negatively   (Table 2). A possible explanation is that higher serum Ca 2+ may increase the deformability of red cells, leading to reduced RDW. For the relationship between RDW and proteinuria, the results of several studies showed the consistence with our study. For example, Zhang M et al. 27 assessed 320 patients who were newly diagnosed with T2D and indicated that RDW was a risk factor for microalbuminuria (MAU), with a value of 0.79 for the area under the curve as compared to a healthy group. In addition, Caroline J et al. 37 collected 196 T2D patients with DN (57%), diabetic neuropathy (46%) and peripheral arterial disease (26%) and found that RDW level is a high risk factor in DN (OR: 1.64, 95% CI: 1. 15-2.35). Similarly, our results indicate that RDW is positively associated with proteinuria ( Table 4) after adjustment of the potential confounders such as age and gender.
Furthermore, one study of 786 older women suggested that a higher quartile of RDW tended to be associated with a higher interleukin-6 level 38 . It was also reported that TNF-α and interleukin-6, which reflect pro-inflammatory conditions in DM patients, have significantly close relationships with proteinuria 39 . Moreover, chronic inflammatory cytokines displayed a key role in damaging and increasing the permeability of glomerular endothelial cells, resulting in proteinuria 40 . Therefore, proteinuria could be used to reflect the level of inflammation, providing a reasonable explanation for the close association between RDW and proteinuria. Interestingly, proteinuria was significantly related to oxidative stress, which is involved in the oxidation of the LDL fatty acids, and it was also associated with RBC fragments 41, 42 that give rise to incremental increases in RDW 43 .
Diabetes mellitus plays a pivotal role in recurrent atherothrombotic events, especially in patients with acute coronary syndrome (ACS) that underwent percutaneous coronary intervention (PCI) 44 due to risk factors including hyperglycemia, insulin deficiency, and metabolic dysregulation resulting in platelet dysfunction 45 . In addition, "prolonged" antiplatelet therapy could decrease vascular events by approximately 1/4 in diabetic and non-diabetic subjects 46 . However, antiplatelet therapy may be associated with the elevation of RDW. Here, we found that antiplatelet therapy, such as aspirin or clopidogrel, showed a higher proportion in Q2 or Q4 of RDW in T2D patients (Table 3 and Fig. 3a), suggesting these diabetic patients, characterized by chronic inflammation and oxidative conditions, might experience rearrangement of the cytoskeleton and loss of asymmetric lipids of the RBC47, resulting in high levels of RDW.
In conclusion, our results provide novel insight into the relationship between RDW and basic characteristics, blood and urine examinations, complications and medications in DM. First, a graded association between disease duration and RDW was obtained, showing that longer durations were correlated with increasing RDW, and that older patients had significantly higher RDW levels than younger patients. Second, serum Ca 2+ and Hb may be protective in decreasing the level of RDW. Third, an increased level of RDW was accompanied by elevated levels of serum Cr and proteinuria in type 2 diabetes patients. Finally, the elevated level of RDW with antiplatelet therapy indicates that RDW might be a new biomarker for evaluating the dosage of antiplatelet drugs. However, several limitations should be mentioned in our study. For example, the present study is a cross-sectional study, and the relationship between RDW and other indexes is temporal or casual. Altogether, RDW may be a significant and accessible biomarker in T2D patients relative to clinical detection and evaluation.  Laboratory examination. Hematologic testing was performed with the ADVIA 2120 automated hematology analyzer (Siemens Healthcare Diagnostics, Germany), measuring hemoglobin, white cell count, RDW, mean platelet volume, platelet crit, platelet large cell ratio, and platelet distribution width, as described previously 48 . The liver and renal function parameters, including BUN, Cr, UA, Alb, TG, TC, HDL, LDL and FBG levels, were analyzed using standard automated enzymatic methods (Hitachi 912 automated analyzer), as previously described 49 . Other biochemical examinations including CRP, serum Ca 2+ , and P were detected on the C8000 Abbott ARCHITECT Clinical Chemistry Analyzers (Abbott Diagnostics, USA). In addition, proteinuria was defined as a urine albumin excretion rate (UAER) of greater than 30 mg/24 h 50 . Urine concentrations of albumin were measured by the immunoturbidimetric method as described previously 48,49 . In addition, HbA1c evaluations were performed with automatic high-performance liquid chromatography (HPLC) (VARIANT-II Hemoglobin Testing System; Bio-Rad Laboratories, Hercules, CA) 51 .