Clinical Care/Education/Nutrition/Psychosocial Research ORIGINAL ARTICLE

NSE, a Potential Biomarker, Is Closely Connected to Diabetic Peripheral Neuropathy

1,2 1 JIANBO LI, MD, PHD LINGFEI YAN, MD nerves is the endoneurium, analogous to 1 1 HONGMAN ZHANG, MD JIAWEI CHEN, MD – 1 3 the blood brain barrier to some extent. In MIN XIE, MD HONGXING WANG, MD, PHD this respect, endoneurial fluid is similar to cerebrospinal fluid in the central nervous system. During the development of pe- OBJECTIVEd fi To explore the relationship between serum neuron-speci c enolase (NSE) lev- ripheral nerve damage, the amount of en- els and diabetic neuropathy. doneurial fluid may increase at the site RESEARCH DESIGN AND METHODSdType 1 or 2 diabetic and healthy control sub- due to irritation or other deleterious stim- jects (n = 568) were randomly enrolled in a cross-sectional study. Diabetic neuropathy status was uli (4). However, unlike the central ner- documented by the presence of clinical symptoms or signs, electromyography, quantitative vous system with the protection of the sensory tests, and cardiac autonomic neuropathy tests. The severity of the neuropathy was staged blood–brain barrier, the peripheral ner- by composite scores. Serum NSE was measured using electrochemiluminescence immunoassay. vous system is more vulnerable and read- The demographic and clinical variables were obtained through an interviewer questionnaire. ily exposed to toxins (4). Hyperglycemic and ischemic or hypoxic environments in- RESULTSdSerum NSE levels increased slightly in diabetic subjects compared with normal duce oxidative stress in the nervous sys- subjects (9.1 [1.5] vs. 8.7 [1.7], P = 0.037), and the levels increased greatly in diabetic subjects – with neuropathy compared with those without (10.8 [2.8] vs. 9.1 [1.5], P = 0.000). The asso- tem (1 3). Oxidation inactivates several ciation of NSE with diabetic neuropathy was independent of the hyperglycemic state (fasting glycolytic enzymes, including enolase, in blood glucose, HbA1c, duration, and the type of diabetes) and other potential confounders neurons (4). To meet the fairly high-energy affecting NSE levels (e.g., age, sex, and renal status) (odds ratio 1.48 [1.13–1.74], P =0.001). requirements under such conditions, the In addition, NSE levels increased with and were closely correlated to the stages of neuropathy (r = glycolytic enzymes are compensatively 0.63 [0.52–0.74], P = 0.000). The optimal cutoff point for serum NSE levels to distinguish upregulated to increase survival of the patients with diabetic neuropathy from those without was 10.10 mg/L, with a sensitivity of fi neurons (5). Chronic exposure to hyper- 66.3% and a speci city of 72.5%. glycemia or its related ischemia/hypoxia CONCLUSIONSdSerum NSE levels are closely associated with peripheral neuropathy in with oxidative stress leads to an increased patients with diabetes. Future studies are warranted to clarify the relationship. risk for peripheral neuropathy (6), which is characterized by neurodegeneration Diabetes Care 36:3405–3410, 2013 that is often concomitant with neurore- generation (7). During this process, the europathy is one of most common g-Enolase, also known as enolase 2 rate of synthesis of the enolase in the af- chronic complications in patients (ENO2) or NSE (2), is a highly soluble fected neurons may change, and it is N likely to cause the NSE to leak into the with diabetes. A timely and accu- intracellular enzyme normally located in fl rate diagnosis of diabetic neuropathy is the cytoplasm in neuroendocrine cells. endoneurial uid and serum. We postu- essential to early interventions for de- The protein is principally located in neu- lated that in this circumstance, NSE lev- creasing the rate of the associated disabil- ronal tissues. After tissue injury, NSE is els in circulation may be indicative of ity and death. readily released into the cerebrospinal diabetic peripheral neuropathy (DPNP). A range of diagnostic tools are avail- fluid and blood where it has been shown Previous studies on NSE in associa- able (1). However, a biomarker specificto to have a biological half-life of 48 h (3). tion with DPNP are limited (8). There are neural damage is still not known. In this Neurons of the peripheral nervous system no meticulous analyses of this relation- regard, neuron-specific enolase (NSE), reside in the ganglia around the spine or in ship. In addition, few data are available which catalyzes the conversion of 2- close proximity to their target organs. Pe- for the Chinese who are plagued by an phosphoglycerate into phosphoenolpyr- ripheral nerves are bundled along with increasing incidence of diabetes (9). For uvate, may act as a new emerging biomarker blood vessels supplying the organs. Sur- this reason, we evaluated the relationship of peripheral neuropathy in diabetes. rounding each fiber of these peripheral between the concentration of NSE and di- abetic neuropathy among subjects with ccccccccccccccccccccccccccccccccccccccccccccccccc diabetes. From the 1Department of Endocrinology and Metabolism, First Affiliated Hospital of Nanjing Medical University, 2 fi Nanjing, Jiangsu Province, China; the Department of Endocrinology and Metabolism, Af liated Jiangsu RESEARCH DESIGN AND Shengze Hospital of Nanjing Medical University, Suzhou, Jiangsu Province, China; and the 3Department of Rehabilitation, First Affiliated Hospital of Nanjing Medical University, Nanjing, Jiangsu Province, China. METHODS Corresponding author: Jianbo Li, [email protected]. Received 11 March 2013 and accepted 6 May 2013. Subject selection and diabetes status DOI: 10.2337/dc13-0590 © 2013 by the American Diabetes Association. Readers may use this article as long as the work is properly Subjects with type 1 or type 2 diabetes cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/ registered consecutively as outpatients or licenses/by-nc-nd/3.0/ for details. inpatients with our hospital (the tertiary care.diabetesjournals.org DIABETES CARE, VOLUME 36, NOVEMBER 2013 3405 NSE and peripheral neuropathy care hospital) and healthy control sub- the neuropathy symptom score published conduction (NC) was considered abnor- jects registered with our hospital physical previously (11). For somatic neuropathy, mal (14). Local skin temperature was kept examination center between March 2009 symptoms were documented, such as constant (32–338C). Variables such as age, and November 2012 were randomly en- numbness, asleep feeling, burning, deep height, sex, and weight were also measured rolled into the study. They were mainly aching, and unsteadiness in walking. For and accounted for when reporting an NCV local, from seven districts in Nanjing. cardioautonomic neuropathy, symptoms as normal or abnormal as described previ- Diabetes status was biochemically con- were documented, including unexplained ously (15). firmed according to the World Health resting tachycardia and postural fainting. Organization diagnostic criteria for the Ascore$1 was positive for a symptom(s). Cardiac autonomic neuropathy tests classification of diabetes (10). All volun- Signs, including decreased distal sensations Testing was carried out based on standard teers signed informed consent. The study and unequivocally decreased or absent re- procedures (16,17). The resting heart rate was approved by the hospital and univer- flexes for somatic neuropathy, were re- wasmeasuredaftera20-minsupinerest. sity scientific and ethic committees. Pa- vealed through physical examination with The Valsalva maneuver was performed tients with severe renal dysfunction tools. Touch sensation was tested using a three times. The mean value was the ratio (endogenous creatinine clearance rate 10-g monofilament (on four sites per foot), of the longest R-R interval to the shortest [Ccr] , 50 mL/min), severe liver disease pain sensation was tested using a pin, re- R-R interval after breathing against a (e.g., aspartate aminotransferase or ala- flexes were tested using a tendon hammer, resistance of 40 mmHg for 15 s. The nine aminotransferase greater than three and vibration sensation was tested using a expiration-inspiration ratio was the ratio times the normal level), heart failure (New standard 128-Hz tuning fork. The signs of the mean of the three longest R-R in- York Heart Association class III or IV), were documented based on the neurologi- tervals on expiration to the mean of the known peripheral vascular diseases, cen- cal disability score: 0 for no deficit, 1 for a three shortest R-R intervals on inspiration tral nervous system disorders, and neuro- mild deficit,2foramoderatedeficit, 3 for a during 1 min of deep breathing at six endocrine cell-derived hyperplasia or severe deficit, and 4 for a complete absence breaths per minute. The lying-to-standing tumors were excluded. The enrolled pa- of function or the severest deficit (11). ratio was the ratio of the longest R-R in- tients continued on their previous regi- terval around the 30th beat after standing mens of hypoglycemic agents or insulin Distal peripheral neuropathy to the shortest R-R interval around the during this study: 263 on metformin (bi- quantitative tests 15th beat after standing. The blood pres- guanide), 317 on sulfonylureas, 124 on Quantitative thermal sensory test. A sure response was tested. The blood pres- insulin, and 375 on more than one of quantitative sensory thermotest was per- sure was measured in the supine position at these agents. In addition, they continued formed on each subject using a thermo- 1 min after standing up and at 1-min inter- the use of antihypertensives (149 on ACE stimulator (TSA-2001 II; Medoc, Ramat vals thereafter for a 5-min period. A de- inhibitor or angiotensin receptor blocker Yishai, Israel). The test was performed crease in systolic pressure of more than agents) and lipid-regulating agents (98 on by placing a 30 3 30 mm thermode at least 20 mmHg was considered abnor- statins or fibrates) if necessary. over the skin at the site on both sides of mal. The results were evaluated using the body (e.g., the dorsolateral border of age-related reference values. Two or more Neuropathy assessment the foot or the hypothenar eminence) to abnormal tests were required for a con- Testing was performed on each partici- be tested (12). The laboratory temperature firmed diagnosis of cardiac autonomic neu- pant by the same experienced physician was maintained at 23–248C. A cluster of ropathy (CAN). according to standard procedures. The stimuli for each condition (cold sensation, results were evaluated using age-related warm sensation, heat-induced pain, and Diagnosis and stages of reference values. Before (e.g., 24 h prior cold-induced pain) was tested. All thresh- polyneuropathy to) and during neuropathy testing, the olds were obtained with ramped stimuli Diabetic neuropathy was classified accord- influence of confounding variables was (18C/s) that were terminated when the sub- ing to the American Diabetes Association minimized; medications (e.g., b-blockers), ject pressed a button. Cutoff temperatures recommendation (18). The peripheral food (e.g., spicy food), or drinks (e.g., cof- were 0 and 508C. The baseline temperature nerve deficit of nondiabetic origin (e.g., fee) affecting heart rates or nerve functions was 328C. Three consecutive recordings compression due to vertebral disk her- were discontinued or avoided if possible; were averaged. Based on the values ob- niation, vitamin B deficiency, alcoholism, and antecedent activities (e.g., physical ac- tained from the control reference, the or primary amyloidosis) was excluded tivities) and emotional stress were avoided mean threshold temperature 62.5 SD for through a careful medical history review, or minimized. Patients with symptomatic each modality was considered as the upper a differential test, or both. hyperglycemia, hypoglycemia, and ketonuria or lower limit of normal (13). Polyneuropathy was further staged were excluded or these conditions were according to components of composite precorrected at testing. All tests were con- Nerve conduction velocity tests scores similar to that previously described ducted in a quiet laboratory. Ulnar, median, sural, and peroneal nerve by Dyck (14): 1)NC,2) neurological ex- conduction velocities (NCVs) were per- amination (NE), 3) quantitative thermal Neurological symptoms and formed on the right side of each subject sensory test (QTST) or autonomic exam- physical examination using EMG (key point, medronet 4, Min- ination (QAE), and 4) neuropathic symp- All patients first had a complete history of neapolis, MN). All stimulations were per- toms (NS). Stage 0 (no neuropathy) was neurological symptoms taken and were formed supramaximally. The threshold defined as fewer than two abnormalities given a physical examination. Symptoms for the slowed NCV was set at .2.5 SD among the above parameters. Stage 1 were documented through the question- of the control NCV. When two or more (asymptomatic neuropathy) was two or naire concerning the symptoms listed in nerves were tested abnormal, nerve more abnormalities among 1)NC,2)NE,

3406 DIABETES CARE, VOLUME 36, NOVEMBER 2013 care.diabetesjournals.org Li and Associates or 3) QTST or QAE, in the absence of Urinary creatinine concentrations were status (urinary ACR and Ccr) (insignificant 4) NS. Stage 2 (symptomatic neuropathy) measured using the alkaline picrate P values not shown). Retinopathy (46 non- was two or more abnormalities among 1) method. The individual urinary albumin- proliferative diabetic retinopathy and 12 NC, 2)NE,or3) QTST or QAE, in the pres- to-creatinine ratio (ACR) was then calcu- proliferative diabetic retinopathy) or ne- ence of 4) NS judged to be due to diabetic lated as albumin (mg)/creatinine (g). The phropathy was identified in 58 and 49 di- neuropathy but not disabling. Stage 3 endogenous Ccr was calculated to estimate abetic patients, respectively. Neuropathy (disabling neuropathy) was two or more the glomerular filtration rate according to was identified in diabetic subjects with in- – abnormalities among 1)NC,2)NE,or the Cockcroft equation: Ccr = {[140 creased levels of blood glucose and HbA1c, 3) QTST or QAE, in the presence of age (years) 3 body weight (kg)]/[0.818 3 longer duration of diabetes, increased 4) NS (disabling NS). The diagnosis of serum creatinine (Scr, mmol/L)]} for male prevalence of retinopathy, and increased disabling NS is based on the occurrence and the result 3 0.85 for female. All mea- use of antidiabetics, antihypertensives, of any of the following neuropathic con- surements were repeated twice. and lipid-regulating agents compared ditions: motor disability (e.g., muscle with the patients without neuropathy. weakness sufficient for the patient not Statistical analysis Serum NSE levels increased slightly in to be able to walk independently), sensory We used SPSS version 13 for Windows the diabetic subjects, and the levels disability (e.g., sensory ataxia sufficient (SPSS Inc., Chicago, IL) software for statis- were significantly elevated in diabetic sub- for the patient not to be able to walk inde- tical analysis. The data were expressed as jects with neuropathy in contrast to those pendently), or severe autonomic (e.g., con- median with 25th and 75th quartiles for without (10.8 [2.8] vs. 9.1 [1.5], P =0.000) tinuous weekly fainting for at least 6 weeks skewed data or as the mean (SD) for nor- (Table 1). In addition, NSE was shown to due to orthostatic hypotension [.30 mally distributed data. Nonnormally dis- be higher in neuropathic patients with ret- mmHg systolic]) (14). tributed variables were log transformed inopathy than those without (11.5 [1.7] vs. before analysis. The multiple compari- 9.6 [1.4], P = 0.034). The two sexes had Clinical feature measurement sons among groups were assessed using similar NSE levels (9.0 [1.6] [male] vs. 9.2 Body weight was measured on the same ANOVA for variables. Percentages were [1.7] [female], P =0.14).TheserumNSE scales in light clothing and no shoes before compared using the x2 test. The variables, level in relation to neuropathy was further breakfast, and upright height was measured including NSE, were first assessed in uni- assessed in a multivariate model (Table 2), on the same wall-mounted stadiometer. variate analyses. Serum NSE was later controlling for retinopathy and other co- Individual BMI was then calculated as added to a logistic regression model, con- variables that may potentially influence weight (kg)/height (m)2.Theright-arm trolling for possible confounders for serum the NSE level or neuropathy, which in- blood pressure of each seated subject was NSE. The relation of the NSE levels to the cluded age, sex, the type of diabetes, blood obtained after 10 min of rest using a mer- stages of neuropathy was also examined. glucose, HbA1c, duration of diabetes, vita- cury sphygmomanometer. The last two of Receiver operating characteristic (ROC) min B12, and renal status. After adjustment, three consecutive readings with 1-min in- analysis was conducted with MedCalc the serum NSE level was still independently tervals were averaged as the blood pressure. Software version 12.0 for Windows associated with diabetic neuropathy (odds Retinal conditions were assessed by oph- (Mariakerke, Belgium) to assess the accu- ratio 1.48 [1.13–1.74], P = 0.001). The thalmologists using a combination of clin- racy of serum NSE levels in distinguishing diabetic neuropathy identified included ical examination, stereoscopic retinal between patients with diabetic neuropa- 172 distal sensorimotor peripheral neurop- photographs, and fluorescein angiography. thy and without. The optimal cutoff point athies (DSMPNPs) without CANs, 14 CANs was identified by calculating the area un- without DSMPNPs, and 28 DSMPNPs Blood NSE and other parameter der the curve (AUC). P , 0.05 was con- mixed with CANs. The overall severity of measurement sidered statistically significant. the neuropathies was stage 1, 2, or 3 in 95, Fasting serum NSE concentration was All procedures were conducted ac- 75, and 44 patients, respectively. The se- collected and measured using electroche- cording to the 19th revision of the Decla- rum NSE levels increased with the neurop- miluminescence immunoassay automatic ration of Helsinki. athy stages from the asymptomatic to the analyzer (Analytics E170; Roche, Basel, disabling (stage 1–3) when compared with Switzerland) according to the instruction of RESULTSdA total of 17 subjects with- diabetic subjects without neuropathy the manufacturer. Hemolysis was avoided drew from the study for the following (stage 0) (P = 0.004–0.000, respectively) during the procedure. Fasting plasma glu- reasons: withdrawal of consent, medical (Table 3). Correspondingly, the NSE lev- cose, plasma cholesterol and triacylglyc- care affordability, deterioration of medi- els were closely correlated to the stages of erol, and serum creatinine were measured cal conditions, or other difficulties com- neuropathy (r =0.63[0.52–0.74], P = using an automatic analyzer (AU5400; pleting the required tests. The study was 0.000). Using ROC analysis, the optimal Olympus, Shinjuku, Japan). HbA1c was completed by 568 subjects, 21–77 years cutoff point for the serum NSE level in measured using high-performance liquid of age, including 136 healthy control sub- order to distinguish patients with dia- chromatography (D10; Bio-Rad, Berkeley, jects, 218 diabetic subjects without neu- betic neuropathy from those without CA). Values of HbA1c were first reported ropathy, and 214 diabetic subjects with was 10.10 mg/L, with a sensitivity of as percent, followed by the mmol/mol neuropathy (Table 1). Among the three 66.3%, a specificity of 72.5%, and a highest equivalent. Serum vitamin B12 was deter- groups of subjects, there were no obvious AUC equal to 0.73 (0.68–0.77, P =0.000) mined using automated test assays (Access; differences between any given two groups (Fig. 1). Beckman, Brea, CA). Urinary albumin con- in the following variables: age, sex ratio, di- centration was measured using immunone- abetes type ratio, BMI, blood pressure, lipid CONCLUSIONSdThe results of this phelometry (DCA 2000; Bayer, Leverkusen, profile (total cholesterol, HDL cholesterol, study demonstrated that serum NSE is a North Rhine-Westphalia, Germany). and triglycerides), vitamin B12, and renal potential biomarker of DPNP. In this study, care.diabetesjournals.org DIABETES CARE, VOLUME 36, NOVEMBER 2013 3407 NSE and peripheral neuropathy

Table 1dThe demographic, clinical, and laboratory profiles of all the subjects

Healthy control Diabetes without neuropathy Diabetes with neuropathy Variables (n = 136) P value1 (n =218) P value2 (n =214) P value3 Age (years) 52.5 (38.6–61.9) N 51.4 (43.3–60.6) N 52.3 (40.0–64.8) N Sex, male/female (%) 65.4 N 64.7 N 67.2 N Type 1/type 2 diabetes (%) dd 13.2 d 12.6 N Fasting glucose (mmol/L) 5.3 (0.5) 0.000 7.4 (2.3) 0.000 7.9 (2.6) 0.032 HbA1c (%) (mmol/mol) 5.4 (0.3) (36 [4.2]) 0.000 6.8 (2.0) (51 [6.4]) 0.012 7.5 (2.5) (58 [9.3]) 0.006 Duration of diabetes (years) dd5.9 (2.4) d 7.2 (3.1) 0.000 Retinopathy (%) dd 10.2 d 18.5 0.024 BMI (kg/m2) 22.8 (3.3) N 23.0 (2.1) N 23.4 (2.5) N Systolic blood pressure (mmHg) 130 (12) N 129 (11) N 130 (14) N Diastolic blood pressure (mmHg) 78 (10) N 78 (9) N 79 (12) N Plasma total cholesterol (mmol/L) 5.0 (0.8) N 5.1 (1.0) N 5.0 (0.7) N Plasma HDL cholesterol (mmol/L) 1.06 (0.10) N 1.05 (0.10) N 1.06 (0.08) N Plasma triglycerides (mmol/L) 1.52 (0.19) N 1.52 (0.26) N 1.51 (0.28) N Serum B12 (pmol/L) 389 (106) N 402 (108) N 390 (112) N Urinary ACR (mg/g) 24 (0.1) N 24 (6.2) N 25 (7.2) N Ccr (mL/min) 84.1 (10.4) N 81.9 (7.3) N 82.5 (18.4) N Serum NSE (mg/L) 8.7 (1.7) 0.000 9.1 (1.5) 0.037 10.8 (2.8) 0.000

All values are mean (SD) for normally distributed data and median (interquartile range 25–75%) for skewed data. HbA1c is first reported as percent, followed by the mmol/mol equivalent in parentheses. Sex, retinopathy, and type 1/type 2 diabetes ratio data are percentage (%). N, insignificant, P . 0.05. 1P, healthy control vs. diabetes with neuropathy. 2P, diabetes without neuropathy vs. healthy control. 3P, diabetes with neuropathy vs. diabetes without neuropathy.

NSE levels were increased in diabetic sub- general population in our hospital labora- diabetic patients without detectable CNS jects. More important, NSE was observed to tory. This result suggests that the upper disorders and peripheral neuropathies. be significantly elevated in those with neu- limit, if applicable as a reference value, Like the blood–brain barrier, the per- ropathy. The elevated NSE levels were may have been too high, at least for these meability of the structurally similar endo- closely related to diabetic neuropathy, subjects with diabetic neuropathy. This is neurium may also be increased in patients and this relationship was independent of most likely attributed to subject selection. with diabetes. Further increased levels covariables. In addition, enolase levels A larger study is needed to confirm our of NSE were shown to be highly associ- increased with stages of neuropathy, and findings. ated with confirmed DPNPs. In our they were significantly correlated. The High glucose levels have been reported study, neuropathy severity was assessed threshold of the serum NSE level for the to be linked to elevated serum enolase using the composite components: quan- presence of DPNP was found to be ;10.10 concentrations in stroke patients (19). In titative neuroelectrophysiological tests, mg/L. This cutoff value was higher than the addition to central nervous system (CNS) quantitative somatic and autonomic mean value of NSE in either diabetic or disorders, hyperglycemia-induced peri- tests, and clinical manifestations. There- healthy control subjects of the study but cyte loss contributes to blood–brain bar- fore, elevated NSE levels may be indic- did not exceed the normal upper limit rier disruption (20). We found a slightly ative of neuropathy and the level of (16.5 mg/L) of the reference range for the increased level of serum enolase in involved nerve fiber damage, which are likely to be associated with changes in the synthesis and release of the enolase. d Table 2 Multiple regression analysis of the relation of NSE to neuropathy During the process of the pathological changes, including demyelination and Covariables OR (95% CI) P value remyelination, associated with DPNP, NSE may be released not only from af- – Age (years) 1.02 (0.97 1.08) 0.461 fected neurons but also from affected – Male/female (%) 0.83 (0.54 1.10) 0.145 Schwann cells forming myelin, as it was – Type 1/type 2 diabetes (%) 0.97 (0.32 2.61) 0.953 suggested in one report that NSE was de- – Fasting glucose (mmol/L) 1.01 (0.89 1.11) 0.890 tected in oligodendrocytes as well as in – HbA1c (%) 1.13 (1.02 1.24) 0.041 neurons (21). This may explain why NSE – Duration of diabetes (years) 1.31 (1.06 1.50) 0.005 levels were higher in large-fiber neuropathy – Retinopathy (%) 0.57 (0.28 1.13) 0.107 than small-fiber neuropathy in the spec- – Serum B12 (pmol/L) 0.95 (0.87 1.05) 0.768 trum of neuropathy, although both were – Urinary ACR (mg/g) 0.98 (0.92 1.03) 0.613 closely correlated to NSE from our obser- – Ccr (mL/min) 0.93 (0.78 1.14) 0.832 vation (data not shown). Interestingly, NSE – Serum NSE (mg/L) 1.48 (1.13 1.74) 0.001 was shown to be more elevated in neurop- Sex, retinopathy, and type 1/type 2 diabetes ratio data are percentage (%). N, insignificant, P . 0.05. athy with retinopathy than without in our Negelkerke R2 = 0.23. OR, odds ratio. study. Even early diabetic retinopathy

3408 DIABETES CARE, VOLUME 36, NOVEMBER 2013 care.diabetesjournals.org Li and Associates

Table 3dThe relation of levels of NSE to Ccr .50 mL/min). In addition, type 1 and H.W. collected and researched data and con- stages of neuropathy 2 diabetes had similar NSE levels at the given tributed to discussion. L.Y. is the guarantor of levels of severity of neuropathy. These vari- this work and, as such, had full access to all the data in the study and takes responsibility for Neuropathy ables were considered among the studied groups during the evaluation of the relation the integrity of the data and the accuracy of the stage n NSE P value data analysis. of NSE to neuropathy. Furthermore, we did The authors thank Jiawei Chen (First Affil- 0 218 9.1 (1.7) not find that antidiabetics, antihyperten- 1 iated Hospital of Nanjing Medical University) 1 95 9.7 (1.9) P =0.004 sives, and lipid-regulating agents (used for help in the preparation of the manuscript. 2 75 10.9 (2.1) P =0.0002 in the subjects) affected NSE levels signif- 3 3 44 11.4 (3.2) P =0.000 icantly (data not shown). In general, the References fi fi Stage 0, diabetes with no neuropathy; stage 1, diabetes coef cient of variation for intrasubject day- 1. Boulton AJ. Diabetic neuropathy: classi - with asymptomatic neuropathy; stage 2, diabetes with to-day reproducibilities of NSE analysis for cation, measurement and treatment. Curr symptomatic neuropathy; stage 3, diabetes with dis- our normal control and diabetic subjects Opin Endocrinol Diabetes Obes 2007;14: abling neuropathy. 1P, stage 1 vs. stage 0. 2P, stage 141–145 3 was 7.9 and 9.5%, respectively. 2 vs. stage 0. P, stage 3 vs. stage 0. There were some limitations of our 2. Kaiser E, Kuzmits R, Pregant P, Burghuber study. We assessed the relationship be- O, Worofka W. Clinical biochemistry of neuron specific enolase. Clin Chim Acta tween NSE levels and overall diabetic – involves a retinal neurodegenerative change fl 1989;183:13 31 (22). Therefore, a fraction of this NSE may neuropathy, and the result may be in u- 3. Marangos PJ, Schmechel DE. Neuron be due to nerve damage in the retina. After enced by the number and ethnicity of the specific enolase, a clinically useful marker retinopathy was considered, NSE was still subjects enrolled in this study. In addition, for neurons and neuroendocrine cells. correlated to peripheral neuropathy. thedifferenceinthelevelofNSEbetween Annu Rev Neurosci 1987;10:269–295 Apart from retinopathy, the NSE level diabetic patients without neuropathy and 4. Marieb EN, Hoehn K. Human Anatomy & associated with neuropathy was indepen- with asymptomatic neuropathy as opposed Physiology. 7th ed. London, UK, Pearson, 2007, p. 388–602 dent of other variables. Hemolysis, a ma- to the later stages of neuropathy was rela- tively small although significant. This, along 5. Mielke R, Schröder R, Fink GR, Kessler J, jor interfering factor, was avoided so that Herholz K, Heiss WD. Regional cerebral the analysis would not be affected during with a relatively low sensitivity of the cutoff value of NSE (through ROC analysis) for the glucose metabolism and postmortem pa- the study. We excluded CNS disorders and thology in Alzheimer’s disease. Acta Neu- other diseases reported to cause elevated indication of neuropathy, may be attributed ropathol 1996;91:174–179 serum enolase (23–27). Potential confound- to the detection limit of the NSE assay 6. Fernyhough P, Roy Chowdhury SK, ers that may influence the serum enolase lev- currently used by our hospital. This low Schmidt RE. Mitochondrial stress and the els were also considered, such as age, sex, sensitivity for NSE may limit the clinical and pathogenesis of diabetic neuropathy. Ex- – and renal status, as suggested in previous research usefulness of the test. Future, pert Rev Endocrinol Metab 2010;5:39 49 reports on Alzheimer disease and cancer larger-scale studies that include more sub- 7. Dyck PJ, Zimmerman BR, Vilen TH, et al. Nerve glucose, fructose, sorbitol, myo- markers (28,29). In our subjects, we did categories of diabetic neuropathy along with possible improvements in the detec- inositol, and fiber degeneration and re- not observe obvious differences in the serum generation in diabetic neuropathy. N Engl NSE levels between the two sexes. Enolase tion of NSE are warranted. This may clarify – the NSE-neuropathy relationship observed J Med 1988;319:542 548 levels increased slightly but not significantly 8. Sandhu HS, Butt AN, Powrie J, with age. There was no significant enolase in this cross-sectional study, thus providing Swaminathan R. Measurement of cir- level correlation with renal function (by an additional approach to guide the man- culating neuron-specificenolasemRNA agement of this diabetes complication. in diabetes mellitus. Ann N Y Acad Sci In conclusion, we have observed for 2008;1137:258–263 the first time that serum NSE levels are 9. Yang W, Lu J, Weng J, et al.; China Na- elevated in diabetes and are related to tional Diabetes and Metabolic Disorders diabetic neuropathy. This may provide a Study Group. Prevalence of diabetes among potential blood biomarker for diabetic men and women in China. N Engl J Med – neuropathy. If future studies confirm our 2010;362:1090 1101 results, an increase of serum NSE as an 10. World Health Organization. Diabetes Mel- litus: Report of a WHO Study Group.Geneva, indicator of diabetic neuropathy would aid World Health Org., 2006 the timely prediction, diagnosis, and treat- 11. Dyck PJ, Sherman WR, Hallcher LM, et al. ment of the diabetic population. Human diabetic endoneurial sorbitol, fructose, and myo-inositol related to sural nerve morphometry. Ann Neurol 1980;8: AcknowledgmentsdThis study was sup- 590–596 ported by the National Natural Science Foun- 12. Yarnitsky D, Sprecher E. Thermal testing: dation of China (81070655) and the Jiangsu normative data and repeatability for vari- Provincial Natural Science Foundation of China ous test algorithms. J Neurol Sci 1994; Figure 1dROC plot. Serum NSE levels were (BK2009441). 125:39–45 shown in distinguishing between patients with No potential conflicts of interest relevant to 13. Shukla G, Bhatia M, Behari M. Quantita- and without diabetic neuropathy. The optimal this article were reported. tive thermal sensory testingdvalue of cutoff point of the serum NSE levels was 10.10 J.L. designed the study, researched data, testing for both cold and warm sensation mg/L, with a sensitivity of 66.3%, a specificity and wrote, reviewed, and edited the manu- detection in evaluation of small fiber neu- of 72.5%, and the highest AUC equal to 0.73 script. H.Z., M.X., and L.Y. collected and re- ropathy. Clin Neurol Neurosurg 2005;107: (0.68–0.77, P =0.000). searched data. J.C. reviewed the manuscript. 486–490 care.diabetesjournals.org DIABETES CARE, VOLUME 36, NOVEMBER 2013 3409 NSE and peripheral neuropathy

14. Dyck PJ. Detection, characterization, and brain infarction. J Clin Neurosci 2005;12: brain pH, agonal state and inter-lobe staging of polyneuropathy: assessed in 542–547 mRNA variance. Brain Res Mol Brain Res diabetics. Muscle Nerve 1988;11:21–32 20. Wunderlich MT, Ebert AD, Kratz T, 2003;118:60–71 15. Jianbo L, Yuche C, Ming S, et al. Association Goertler M, Jost S, Herrmann M. Early 25. Xiaofang Y, Yue Z, Xialian X, Zhibin Y. of homocysteine with peripheral neuropa- neurobehavioral outcome after stroke is Serum tumour markers in patients with thy in Chinese patients with type 2 diabetes. related to release of neurobiochemical chronic kidney disease. Scand J Clin Lab Diabetes Res Clin Pract 2011;93:38–42 markers of brain damage. Stroke 1999;30: Invest 2007;67:661–667 16. Feldman EL, Stevens MJ, Thomas PK, 1190–1195 26. Pandey A, Saxena K, Verma M, Bharosay Brown MB, Canal N, Greene DA. A practical 21. Clegg N, Ferguson C, True LD, et al. A. Correlative study between neuron- two-step quantitative clinical and electro- Molecular characterization of prostatic specific enolase and blood sugar level in physiological assessment for the diagnosis small-cell neuroendocrine carcinoma. ischemic stroke patients. J Neurosci Rural and staging of diabetic neuropathy. Di- Prostate 2003;55:55–64 Pract 2011;2:50–54 abetes Care 1994;17:1281–1289 22. Kasprzak A, Zabel M, Biczysko W. Se- 27. Price TO, Eranki V, Banks WA, Ercal N, Shah 17. Ewing DJ. Analysis of heart rate variability lected markers (chromogranin A, neuron- GN. Topiramate treatment protects blood- and other non-invasive tests with special specific enolase, synaptophysin, protein brain barrier pericytes from hyperglycemia- reference to diabetes mellitus. In Auto- gene product 9.5) in diagnosis and prog- induced oxidative damage in diabetic mice. nomic Failure. Vol 16, 3rd ed. Bannister R, nosis of neuroendocrine pulmonary tu- Endocrinology 2012;153:362–372 Mathias CJ, Eds. Oxford, Oxford Univer- mours. Pol J Pathol 2007;58:23–33 28. Sensenbrenner M, Lucas M, Deloulme JC. sity Press, 1992, p. 312–333 23. Nemeth J, Galian A, Mikol J, Cochand- Expression of two neuronal markers, 18. Summary of revisions for the 2006 Clini- Priollet B, Wassef M, Lavergne A. Neuron- growth-associated protein 43 and neuron- cal Practice Recommendations. Diabetes specific enolase and malignant lymphomas specific enolase, in rat glial cells. J Mol Med Care 2006;29(Suppl. 1):S3 (23 cases). Virchows Arch A Pathol Anat (Berl) 1997;75:653–663 19. Selakovic V, Raicevic R, Radenovic L. The Histopathol 1987;412:89–93 29. van Dijk HW, Verbraak FD, Kok PH, et al. increase of neuron-specific enolase in ce- 24. Preece P, Cairns NJ. Quantifying mRNA in Early neurodegeneration in the retina of rebrospinal fluid and plasma as a marker postmortem human brain: influence of type 2 diabetic patients. Invest Ophthalmol of neuronal damage in patients with acute gender, age at death, postmortem interval, Vis Sci 2012;53:2715–2719

3410 DIABETES CARE, VOLUME 36, NOVEMBER 2013 care.diabetesjournals.org