Journal of Diabetes and Its Complications 24 (2010) 306 – 312

WWW.JDCJOURNAL.COM

Transforming growth factor beta 1 as a biomarker of diabetic

peripheral neuropathy: cross-sectional study☆
Juan Ybarra a,⁎, Josep M. Pou b , June Hart Romeo c , Javier Merce d , Jeroni Jurado e
a
Instituto de Cardiología Avanzada (ICAMED), Centro Medico Teknon, Barcelona, Spain
b
Division of Endocrinology, Hospital de la Sta. Creu i Sant Pau, Autonomous University of Barcelona, Barcelona, Spain
c
Baldwin-Wallace College, Division of Health, Berea, OH, USA
d
Department of Biochemistry, Hospital de la Santa Creu i Sant Pau, Autonomous University of Barcelona, Spain
e
Primary Care Center, Catalan Health Institute, Olot, Girona, Spain

Received 6 April 2009; received in revised form 6 July 2009; accepted 31 July 2009

Abstract

Background: Simple and efficient screening methods are lacking for diabetic peripheral neuropathy (DPN), the most common and most
difficult to treat of the long-term diabetic complications. Increased levels of transforming growth factor beta 1 (TGFβ1) in type 2 diabetic
patients (T2DM) plays an immunomodulatory role in diabetic nephropathy and, possibly, in atherosclerotic evolution. Since preliminary
interrelationships between experimental DPN and TGFβ1 have been observed, we sought to assess whether TGFβ1 could be a biomarker
molecule for human DPN. Materials and Methods: Cross-sectional cohort study focused on the assessment of the interrelationships
between TGFβ1 levels, cardiovascular disease (CVD), diabetic nephropathy (DNF), and neuropathy (DPN) in a group of T2DM patients
(N=180; male 117, female 63) randomly selected from the North Catalonia Diabetes Study. DPN was diagnosed using clinical and
neurophysiology evaluation. Incipient DNF was assessed by microalbuminuria (MAU). Total TGFβ1 (without acidification) was measured
by immunoassay by ELISA (Promega). Results: DPN correlated with age, time of diabetes duration, MAU, CVD, and TGFβ1 (Pb.0001).
Log-transformed TGFβ1 (logTGβ1) was significantly higher in patients with DPN than in those without (Pb.0005). LogTGFβ1 (OR=7.5;
P=.006), age (OR=1.1; Pb.0005), and logMAU (OR=2.0; P=.016) appear as significant estimators of the occurrence of DPN in our series.
The integrated ROC curve evaluation with these three parameters expressed an important sensitivity (78.1%), specificity (76.0%), positive
predictive value (79.2%), and negative predictive value (70.3%) in relation to DPN presence. Discussion: TGFβ1 stands as an important
biomarker molecule for DFN and DPN screening in our series. Further prospective studies are warranted.
© 2010 Elsevier Inc. All rights reserved.
Keywords: Diabetic peripheral neuropathy; TGFβ1

1. Background factor to foot ulceration and also amputation (Boulton,

Vileikyte, Ragnarson-Tennvall, & Apelqvist, 2005). Thus,
Diabetic peripheral neuropathy (DPN) is the most screening, and appropriate treatment, for DPN is of
commonly reported long-term diabetic complication, affect- paramount importance. This is due particularly to its
ing up to 50% of type 2 diabetic patients (T2DM) (Boulton, socioeconomic burden, the impact this has on quality of
Malik, Arezzo, & Sosenko, 2004; Vinik, Maser, Mitchell, & life, and on associated anxiety and depression (Benbow,
Freeman, 2003). Additionally, DPN is a major contributory Wallymahmed, & Macfarlane, 1998). Simple and efficient
methods are needed to detect the diabetic population at high
risk of DPN, and progression of the disease should be
☆
Funding: Instituto de Salud Carlos III, FIS PI040181-PI070340, Plan prevented by controlling the modifiable risk factors (Smith
Nacional I+D+I 2004-2007 y Fondos FEDER. et al., 2006; Tesfaye et al., 2005). T2DM patients without
⁎ Corresponding author. Instituto de Cardiología Avanzada, Centro
Medico Teknon, C/ Vilana 12, 08017-Barcelona, Spain. Tel.: +34 93
any known cardiovascular disease (CVD) have the same
2906459; fax: +34 93 2906458. cardiovascular risk as individuals without diabetes who have
E-mail address: [email protected] (J. Ybarra). had a prior cardiovascular event (i.e., myocardial infarction

1056-8727/$ – see front matter © 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jdiacomp.2009.07.007
 J. Ybarra et al. / Journal of Diabetes and Its Complications 24 (2010) 306–312 307

or stroke) (Haffner, Lehto, Ronnemaa, Pyorala, & Laakso, complications (Jurado et al., 2009). The sample number of
1998); however, outcomes after cardiovascular events are the cohort was related to DPN prevalence, about 30%, with a
significantly worse among patients with diabetes, and precision of 5% and a confidence interval of 95%. One
approximately 70% of T2DM patients will die from a hundred ninety-two (N=192) T2DM patients (aged 30 to 69
cardiovascular event or its complications (Stamler, Vaccaro, years) were randomly recruited from our T2DM population
Neaton, & Wentworth, 1993; Turner et al., 1998). and 180 (N=180) met the inclusion criteria. No difference
A number of inflammatory cytokines mediate atheroscle- was observed between the sample and the general population
rosis. Transforming growth factor beta 1 (TGFβ1) modulates with T2DM concerning sex and age (confidence interval:
important events like macrophage and fibroblast chemotaxis, 95%). Selection criteria have been described extensively
suppression of lymphocyte function, collagen synthesis, and elsewhere (Jurado et al., 2009) Clinical data on CVD were
stimulation of extracellular matrix synthesis (Delarue et al., obtained as follows: coronary artery disease (CAD), stroke,
1998; Hong and Yang, 1994; Jiang et al., 1988; Krag et al., and peripheral vascular events were evaluated by medical
2000; Reeves and Androly, 2000). Nikol et al. (1992) records revision. CAD was diagnosed by one of the
demonstrated increased expression of TGFβ1 in human following criteria: (1) previous documented myocardial
atherosclerotic plaques. It was subsequently demonstrated infarction; (2) past medical history of symptoms of angina,
that diabetic patients presenting with an acute myocardial confirmed by a positive treadmill test or significant positive
infarction revealed decreased smooth muscle cells (SMCs) result on coronary angiograms or on thallium scan.
and increased macrophages and TGF-β1 into the culprit Cerebrovascular disease was defined by (1) previous
lesion (Fukumoto, Naito, Asano, & Aramaki, 1998). stroke and/or (2) evidence of stenosis N50% for carotid
Furthermore, TGFβ1 induces renal hypertrophy and vessels at Doppler echography.
fibrosis, and cytokines like tumor necrosis factor-alpha Peripheral vascular disease was defined by the presence
(TNFα), chemoattractant protein-1 (MCP-1), and regulated of (1) intermittent claudication, (2) previous vascular
upon activation and normal T cell expressed and secreted surgery, and/or (3) by Doppler echography or abdominal-
(RANTES) mediate macrophage infiltration into kidney. ilial-femoral angiograms.
Overexpression of these chemokines leads to glomerulo- Patients were studied by a team of 28 health care
sclerosis and interstitial fibrosis (Blobe, Schiemann, & providers including 16 nurses and 12 general practitioners
Lodish, 2000; Chantry, Turner, Abney, & Feldmann, 1989; between September 2005 and September 2007. All exam-
Fukumoto et al., 1998; Navarro and Mora, 2006). inations in each patient were performed by two health care
Biomarkers have emerged in the last decade as useful providers. All participants gave informed written consent to
tools for the clinician in the discrimination of cardiovascular participate, and the study was approved by the primary care
risk (Shlipak, Ix, Bibbins-Domingo, Lin, & Whooley, 2008). ethics committee and by the J. Gol i Gurina Foundation. The
Our group recently reported a novel significant relationship study was carried out in accordance with the principles of the
between DPN and NT-proBNP, independently of previous Declaration of Helsinki as revised in 2000.
CVD in a cross-sectional study involving T2DM patients. Diabetic peripheral neuropathy (DPN) diagnosis was
Increased NT-proBNP levels would presumably indicate the based on clinical neurological examination (CNE) as well as
presence of repetitive microhypoxic insults to nerve fibers' on the Neuropathy Symptoms Score which was described
arterial supply independently of CVD (Jurado, Ybarra, previously (Jurado et al., 2009; Tesfaye et al., 2005).
Ferrandiz, Comerma, & Pou, 2007). To confirm the validity of CNE for DPN diagnosis, in
Few data are available regarding the association between borderline subjects (i.e., those with dubious diagnosis;
DPN and TGFβ1 besides a recent report of TGFβ1 inducing n=77), nerve conduction studies (by Synergy T-EPEMG/
cellular injury in experimental diabetic neuropathy (Anja- EP) were performed in sural, median, and ulnar nerves for
neyulu et al., 2008). Thus, we hypothesized that TGFβ1 upper extremities and in peroneal and tibial nerves for lower
might serve as a novel biomarker for human DPN and/or it extremities (sensory and motor nerve conduction velocities),
could be implicated in its pathogenesis. as described previously (Jurado, Caula, & Pou, 2006).
Hence, we set forth a cross-sectional pilot study, to assess Incipient DNF was defined by MAU [urinary albumin
the hypothetical relationships between TGFβ1 levels, CVD, concentration between 20 and 300 mg/l (twice minimum) out
DNF [microalbuminuria (MAU)], and DPN in a group of of four early morning urine spots collected 12 months apart].
T2DM patients randomly selected from the North Catalonia Asymptomatic urinary tract infections as potential confoun-
Diabetes Study (Jurado, Ybarra, Romeo, & Pou, 2009). ders for MAU prevalence were taken into account. Thus,
urine cultures were performed simultaneously in all received
urine spots. Positive urine cultures invalidated the sample.
2. Materials and methods Patients with macroalbuminuria (≥300 mg/l) were excluded.
Biochemical analyses were performed according to the
A cross-sectional cohort study was performed in three protocols of a unique laboratory. Briefly, total cholesterol,
different regions of North Catalonia (Spain) as part of the high-density lipoprotein cholesterol (HDL-cholesterol), fast-
labors of the North Catalonian Diabetes Study on long-term ing glucose, creatinine, triglycerides, and uric acid were
308 J. Ybarra et al. / Journal of Diabetes and Its Complications 24 (2010) 306–312

analysed using spectrophotometry methods (Hitachi Modu- Table 1

lar-Roche Diagnostics). Low-density lipoprotein cholesterol Clinical and biochemical characteristics of the study population
(LDL-cholesterol) was analysed using Friedwald's formula N (gender) 180 (male 117, female 63)
for triglyceride concentrations below 400 mg/dl. Micro- Age (years) (range) 61.7±7.8 (36–74)
Diabetes duration (years) 9.4±0.7
albuminuria was assessed by immunoturbidimetry (Hitachi
Caucasians 100%
Modular-Roche Diagnostics, San Cugat del Valles, Barce- BMI (kg/cm2) 29.7±5.1 (20.7–47.6)
lona, Spain). HbA1c was analyzed using HPLC (Hi-AUTO Systolic BP (mmHg) 140.50±19.90 (100–203)
A1C HA-8140 A. Menarini Diagnostics, Barcelona, Spain). Diastolic BP (mmHg) 79.20±10.10 (50–120)
TGFβ1 was analyzed as a total peptide without acidification. Fasting glucose (mg/dl) 152±38.8 (34–360)
HbA1c (%) 7.1±0.1 (3.7–13.1)
Variability was higher for free than for total TGFβ1. Total
Microalbuminuria (mg/l) 76±33.6 (0.0–286)
TGFβ1 was measured by ELISA (Promega, Madison, WI, TGFβ1 (μg/ml) 28.6±2.6 (2.23–322.2)
USA). Intra-assay and interassay coefficients of variation Triglycerides (mg/dl) 152±128 (14–1057)
were 3–4% and 7–9%, respectively. Cholesterol (mg/dl) 194±39 (115–318)
HDL (mg/dl) 55.4±13.4 (12–98)
LDL (mg/dl) 111.8±40.9 (22–377)
2.1. Statistical analysis
Creatinine (mg/dl) 1.0±0.8 (0.4–2.2)
DPN prevalence (CI) 40.6±3.7 (33–47)
Qualitative variables were given in percentages and CVD prevalence (CI) 25.4±3.2 (0–100)
number of cases. Quantitative variables were expressed as
mean and standard deviation. The distribution of variables
was evaluated using Kolmogorov–Smirnov's test, while
Levene's test was used to assess variance homogeneity. range (BMI=29.7 kg/m2), while 25% of the patient cohort
Nonparametric tests were employed for variables not was obese (data not shown). Fifty percent of the cohort was
following normal distributions. on insulin therapy, while the other half was on oral agents.
Tests of reproducibility [intraclass correlation coeffi- Most patients (98%) were on statins (data not shown).
cients, reliability analysis], intra- and interobserver for The prevalence of DPN and nephropathy was 40.6% and
vibration perception threshold, and Semmes–Weinstein 22%, respectively (Table 1). No gender differences were
monofilament (SF-MF) were calculated. observed regarding DPN's prevalence [men: 21.69% (CI:
The relationship between DPN and other variables was 15.76–27.62); women: 25.42% (CI: 17.45–33.4)].
evaluated by contingency tables and chi square (χ2) with T2DM patients with DPN showed significantly higher
Fisher's correction if necessary as well as with Pearson's prevalence of CVD (39.4% vs. 21.5%; Pb.0005), diabetic
correlation coefficient. Student's t test was used when mean retinopathy (35.4% vs. 9.4%; Pb.0005), and MAU (36.2%
values were compared. vs. 2.9%; Pb.0005) than those without (Fig. 1).
Logistic regression was performed to predict variables Log-transformed TGFβ1 (logTGβ1) was significantly
related to DPN presence or absence. In our model, all higher in patients with DPN than in those without
introduced variables were related to significant values (1.42±0.28 vs. 1.18±0.31 μg/ml, respectively; Pb.0005)
(Pb.05) or when significant values manifested some (data not shown).
tendency (Pb.10). We used the forward method. The The reproducibility of several diagnostic tests employed
removal of variable from the model was based on the in DPN's diagnosis is demonstrated in that all five tests
significance of the change in the likelihood ratio. Since 98%
of the patient cohort was under statin treatment and the latter
was susceptible to influencing lipid profiles, these were
excluded whenever performing multiple regression analysis.
We calculated the Hosmer and Lemeshow test for good fit.
Area under the curve from ROC was calculated to assess
sensitivity and specificity. A value of 5% (α=0.05) was the
level of significance. Data were analyzed by the SPSS 11.5
statistical package (SPSS, Chicago, IL, USA).

3. Results

Baseline characteristics of the patients are depicted in

Table 1. As shown, our patient cohort was composed of 180
(N=180) T2DM patients in their 60s, with a 1.8-fold male
prevalence. Diabetes evolution averaged 9.4±0.7 years and Fig. 1. Prevalence of diabetic micro- and macroangiopathy in our type 2
their body mass index (BMI) was in the high-overweight diabetic population with (+) and without (−) DPN (DPN). ⁎⁎⁎Pb.0005.
 J. Ybarra et al. / Journal of Diabetes and Its Complications 24 (2010) 306–312 309

(graded tuning fork, neurothesiometer, SW-MF, Neuropathy noninvasively and detects earlier stages of nerve damage
Symptoms Score, and the Aquilean reflex) showed a high compared with intraepidermal nerve fiber pathology in
value of intraclass correlation coefficients. T2DM. Thus, CCM shows promise as an ideal technique to
Borderline subjects (those with dubious DPN diagnosis accurately diagnose and assess progression of human
by CNE; n=77) underwent nerve conduction studies (NCS). diabetic neuropathy (Quattrini et al., 2007).
Confronting DPN diagnosis by NCS and CNE showed an The most novel finding in our study is the significant
intraclass correlation coefficient of 0.469 (Pb.0001) (data prediction of categorical diagnosis of DPN by three
not shown). independent factors: logTGFβ1 levels, incipient DNF
Univariate regression analysis disclosed DPN being (logMAU), and age with notable sensitivity and specificity.
significant and positively related to age, duration of diabetes, Noteworthy, TGFβ1 appears as a new biomarker molecule
CVD, incipient DNF (logMAU), diabetic retinopathy, and for DPN's diagnosis. As detailed previously, the main
TGFβ1, while significantly negative correlations were purpose of this study was to assess the hypothetical
disclosed with total and LDL cholesterol (Table 3). Finally, relationships between TGFβ1 levels, CVD, and microangio-
no significant relationship appeared between DPN and pathic complications (i.e., incipient DFN, DPN) in a group of
gender, BMI, waist circumference, hypertension, fasting T2DM patients randomly selected from the North Catalonia
blood glucose, HbA1c, creatinine, HDL-cholesterol, and Diabetes Study. Our data show that the prevalence of
triglycerides. microangiopathic complications (Table 1) as well as the lack
Three parameters, selected by logistic regression analysis of gender bias falls in agreement with previous reports
(Table 3), were used to predict the diagnosis of DPN (log (Rubino et al., 2007; Tesfaye et al., 2005). Furthermore, the
TGF β1, logMAU, and age). The integrated ROC curve reproducibility of several diagnostic tests employed in
evaluation with these three parameters expressed an DPN's diagnosis in our patient cohort disclosed highly
important sensitivity (78.1%), specificity (76.0%), positive significant intraclass correlation coefficients, yielding proof
predictive value (79.2%), and negative predictive value of good internal consistency and in agreement with previous
(70.3%) in relation to DPN presence. reports (Dyck et al., 1991).
The significant positive correlations between DPN vs. age
and diabetes duration (illustrated in Table 2) are in
4. Discussion agreement with the literature (Partanen et al., 1995), whereas
those with MAU have also been the subject of several reports
Surrogate markers of diabetic neuropathy are being (Hadi and Suwaidi, 2007; Partanen et al., 1995; Pedersen and
actively sought to facilitate the diagnosis, measure the Gaede, 2003), some of them suggesting a common soil for
progression, and assess the benefits of therapeutic interven- the diabetic microvascular complications.
tion in patients with diabetic neuropathy.
Serum levels of neuron-specific enolase mRNA are
decreased in diabetic neuropathy. Additionally, this molec- Table 2
ular marker may also be useful in differentiating early from Univariate regression analysis
advanced eye disease in those diabetics diagnosed with DPN (r)
neuropathy (Sandhu, Butt, Powrie, & Swaminathan, 2008). Age 0.401 ⁎⁎⁎
On the other hand, an independent association between Duration of DM 0.320 ⁎⁎⁎
heat shock protein 27 (HSP27) and distal symmetric Sex 0.023
polyneuropathy has been reported in a large cohort of type CVD 0.194 ⁎⁎
BMI 0.086
1 diabetic subjects, suggesting that sHSP27 levels may be a WC 0.072
novel marker for diabetic neuropathy (Gruden et al., 2008). SBP −0.054
Serum albumin has been reported to be significantly DBP −0.114
associated with the severity of retinopathy and neuropathy in Fasting blood glucose −0.030
a large cohort of Japanese T2DM. Noteworthy, in relation to HbA1c 0.042
Creatinine 0.141
neuropathy, serum albumin was found to be significantly LogMAU 0.271 ⁎⁎
related to the absence of Achilles tendon reflex, motor nerve Cholesterol (mg/dl) −0.259 ⁎⁎
conduction velocity, and somatic nerve function parameters HDL (mg/dl) −0.051
[i.e., minimum F-wave latency (MFWL)]. The results of LDL (mg/dl) −0.207 ⁎⁎
multiple regression analysis with adjustment for three Triglycerides (mg/dl) −0.117
logTGFβ1 (μg/dl) 0.361 ⁎⁎⁎
variables (age, gender, serum C-reactive protein) revealed
that serum albumin was independently related to MNCV and Pearson's correlation coefficient was used for continuous variables and
Spearman's correlation coefficient for categorical variables (gender, CVD).
MFWL (Iwasaki, Togashi, & Terauchi, 2008). WC: Waist circumference; SBP: systolic blood pressure; DBP: diastolic
In vivo corneal confocal microscopy (CCM) allows blood pressure; RNP: diabetic retinopathy.
assessment of small nerve fiber pathological changes. This ⁎⁎ Pb.001.
novel technique quantifies small fiber damage rapidly and ⁎⁎⁎ Pb.0005.
310 J. Ybarra et al. / Journal of Diabetes and Its Complications 24 (2010) 306–312

Noteworthy, the second more robust correlation bet- An in vitro model qualitatively reproducing diabetic foot
ween DPN and the variables depicted in Table 3 concerns ulcer conditions [i.e., changes in relevant cell populations
TGFβ1 (r=0.361; Pb.0005). This particular correlation (macrophages, neutrophils, fibroblasts) and their key
(DPN vs. TGFβ1) has not, to the best of our knowledge, effector cytokines (TNF, interleukin-1beta, interleukin-10,
been described previously in human clinical settings; and TGFβ1)] has shown that the genesis of diabetic foot
nevertheless, recent experimental data seem to suggest a ulcers can be simulated in two ways, either by increasing
role for TGFβ1 in experimental diabetic neuropathy the rate of TNF production fourfold or by decreasing the
(Anjaneyulu et al., 2008). One likely explanation for this rate of TGFβ1 production by 67%. Both manipulations
correlation might be found in DPN's etiology which is resulted in increased inflammation (elevated neutrophils,
complex and multifactorial: hyperglycemia and dyslipide- TNF, and tissue damage) and delayed healing (reduced
mia give rise to oxidative stress and formation of TGFβ1 and collagen) (Mi, Rivière, Clermont, Steed, &
advanced glycation and lipoxidation end products. The Vodovotz, 2007).
inflammatory mediators (i.e., TGFβ1) lead to nuclear The negative significant correlation between DPN vs.
factor kappa B (NFkappaB) activation (Cameron and total and LDL-cholesterol levels in our series (illustrated in
Cotter, 2008). Moreover, genes found down-regulated in Table 2) is a novel and provocative one. It falls in
response to TGFβ1 are mainly those associated with disagreement with previous reports (Agrawal, Sharma, Pal,
oxidative stress (i.e., NFkappaB) and several other genes Kochar, & Kochar, 2006; Benjafield, Glenn, Wang,
implicated in glutathione production and maintenance Colagiuri, & Morris, 2001; Tsuzura et al., 2004). Notewor-
(Roberts et al., 2004). thy, Agrawal et al. (2006) have reported significant
correlations between diabetic retinopathy and nephropathy
Table 3
with low HDL-cholesterol and raised LDL-cholesterol,
Logistic regression analysis was performed to predict variables related to respectively; nevertheless, no correlations were reported
DPN presence or absence between DPN and lipid profiles. Conversely, Tsuzura et al.
(2004) have reported significantly higher ratios of oxidized
LDL-cholesterol corrected by ApoB in T2DM patients with
peripheral neuropathy compared to those without. Further-
more, genetic variation in or near the TNF receptor 2 gene
may predispose clinical neuropathy, reduced glycosylated
hemoglobin, and increased HDL cholesterol in T2DM
(Benjafield et al., 2001). In this particular study, lipid
profiles were indistinguishable between T2DM with or
without peripheral neuropathy.
The fact that almost the entire patient cohort was on statin
treatment is likely to alter the relationship between lipid
profiles and DPN's diagnosis and become a serious potential
confounder. Additionally, those patients with more severe
forms of DPN received more aggressive statin treatment. The
latter is likely to bias these negative correlations for which
we do not have any other physiopathological explanation.
Yet we believe it would be a breach of scientific integrity not
to report it.
DPN itself is reported to be a risk factor for cardiovascular
mortality in T2DM in addition to coexisting macroangio-
pathy, increased albumin excretion rate, and poor glycaemic
control (Gerstein, Mann, & Yi, 2001; Schneider, 2006).
Noteworthy, in our series, T2DM with DPN have signifi-
cantly higher prevalence of CVD, diabetic retinopathy, and
DFN than those without (Fig. 1), in agreement with these
previous reports.
The involvement of TGFβ1 as a key cytokine effector in
DFN has been widely reported in the literature (Delarue
et al., 1998; Hong and Yang, 1994; Jiang et al., 1988; Krag
et al., 2000; Reeves and Androly, 2000); while the
involvement of TGFβ1 is a novel one in clinical research,
ROC: Receiver operating characteristics curve; PPV: positive predictive only an experimental neuropathy study reports the involve-
value; NPD: negative predictive value. ment of this vasoactive peptide (Anjaneyulu et al., 2008).
 J. Ybarra et al. / Journal of Diabetes and Its Complications 24 (2010) 306–312 311

LogTGFβ1, age, and logMAU are all significant Dyck, P. J., Kratz, K. M., Lehman, K. A., Karnes, J. L., Melton, L. J., III,
estimators of the occurrence of DPN in our series. O'Brien, P. C., Litchy, W. J., Windebank, A. J., Smith, B. E., & Low,
P. A. (1991). The Rochester Diabetic Neuropathy Study: Design,
Furthermore, with control for the effects of age and criteria for types of neuropathy, selection bias, and reproducibility of
MAU, logTGFβ1 above 31.6 has 4.5 times the odds of neuropathic tests. Neurology, 41, 799−807.
predicting DPN in our series (Table 3). These results have Fukumoto, H., Naito, Z., Asano, G., & Aramaki, T. (1998). Immunohis-
not, to the best of our knowledge, been previously reported tochemical and morphometric evaluations of coronary atherosclerotic
in the literature. plaques associated with myocardial infarction and diabetes mellitus.
Journal of Atherosclerosis and Thrombosis, 5, 29−35.
Finally, the statistical accuracy (sensitivity, specificity, Gerstein, H. C., Mann, J. F., & Yi, Q. (2001). Albuminuria and risk of
positive predictive value, negative predictive value) of these cardiovascular events, death, and heart failure in diabetic and non
three independent variables (logTGFβ1, age, and logMAU) diabetic individuals. Journal of the American Medical Association, 286,
in predicting DPN falls within acceptable levels in the 421−426.
Gruden, G., Bruno, G., Chaturvedi, N., Burt, D., Schalkwijk, C., Pinach, S.,
literature (Dyck et al., 1991; Partanen et al., 1995; Tesfaye
Stehouwer, C. D., Witte, D. R., Fuller, J. H., & Perin, P. C. (2008).
et al., 2005). Serum heat shock protein 27 and diabetes complications in the
In conclusion, simple efficient methods are needed to EURODIAB prospective complications study: A novel circulating
detect the diabetic population at high risk of DPN and marker for diabetic neuropathy. Diabetes, 57(7), 1966−1970.
progression of the disease should be prevented by controlling Hadi, H. A., & Suwaidi, J. A. (2007). Endothelial dysfunction in diabetes
the modifiable risk factors. In our series, DPN's categorical mellitus. Vascular Health Risk Management, 3, 853−876.
Haffner, S. M., Lehto, S., Ronnemaa, T., Pyorala, K., & Laakso, M. (1998).
diagnosis can be significantly and independently predicted Mortality for coronary heart disease in subjects with type 2 diabetes and
by TGFβ1, DFN, and age. Accordingly, TGFβ1 not only in nondiabetic subjects with and without prior myocardial infarction.
appears as a key effector cytokine in DFN but also seems to New England Journal of Medicine, 339, 229−234.
display significant independent correlations with diabetic Hong, S. Y., & Yang, D. H. (1994). Insulin levels and fibrinolytic activity in
neuropathy and, hence, could be used as a novel biomarker patients with end-stage renal disease. Nephron, 68, 329−333.
Iwasaki, T., Togashi, Y., & Terauchi, Y. (2008). Significant association of
molecule for DPN diagnosis. Further prospective studies serum albumin with severity of retinopathy and neuropathy, in addition
are warranted. to that of nephropathy, in Japanese type 2 diabetic patients. Endocrine
Journal, 55(2), 311−316.
Jiang, X., Kanai, H., Shigehara, T., Maezawa, A., Yano, S., & Naruse, T.
References (1988). Metabolism of transforming growth factor-beta in patients
receiving haemodialysis specially those with renal osteodystrophy. Re-
Agrawal, R. P., Sharma, P., Pal, M., Kochar, A., & Kochar, D. K. (2006). nal Failure, 20, 135−145.
Magnitude of dyslipedemia and its association with micro and macro Jurado, J., Caula, J., & Pou, J. M. (2006). Selection of risk and diagnosis in
vascular complications in type 2 diabetes: A hospital based study from DPN. Validation of method of new systems. Atencion Primaria, 38,
Bikaner (Northwest India). Diabetes Research and Clinical Practice, 116−121.
73, 211−214. Jurado, J., Ybarra, J., Ferrandiz, M., Comerma, L., & Pou, J. M. (2007). Is
Anjaneyulu, M., Berent-Spillson, A., Inoue, T., Choi, J., Cherian, K., & the aminoterminal pro-brain natriuretic peptide (NT-proBNP) a marker
Russell, J. W. (2008). Transforming growth factor-beta induces cellular for DPN? Diabetes Care, e86, 30.
injury in experimental diabetic neuropathy. Experimental Neurology, Jurado, J., Ybarra, J., Romeo, J. H., & Pou, J. (2009). Clinical screening and
211, 469−479. diagnosis of diabetic polyneuropathy: The North Catalonia Diabetes
Benbow, S. J., Wallymahmed, M. E., & Macfarlane, I. A. (1998). Diabetic Study. European Journal of Clinical Investigation, 39(3), 183−189.
peripheral neuropathy and quality of life. Quarterly Journal of Medicine, Krag, S., Osterby, R., Chai, Q., Nielsen, C. B., Hermans, C., & Wogensen,
91, 733−737. L. (2000). TGFβ1.induced glomerular disorders is associated with
Benjafield, A. V., Glenn, C. L., Wang, X. L., Colagiuri, S., & Morris, B. J. impaired concentrating ability mimicking primary glomerular disease
(2001). TNFRSF1B in genetic predisposition to clinical neuropathy and with renal failure in men. Laboratory Investigation, 80, 1855−1958.
effect on HDL cholesterol and glycosylated hemoglobin in type 2 Mi, Q., Rivière, B., Clermont, G., Steed, D. L., & Vodovotz, Y. (2007).
diabetes. Diabetes Care, 24, 753−757. Agent-based model of inflammation and wound healing: Insights into
Blobe, G. C., Schiemann, W. P., & Lodish, H. F. (2000). Role of diabetic foot ulcer pathology and the role of transforming growth factor-
transforming growth factor β in human disease. New England Journal of beta1. Wound Repair and Regeneration, 15, 671−682.
Medicine, 342, 1350−1358. Navarro, J. F., & Mora, C. (2006). Diabetes, inflammation, proinflammatory
Boulton, A. J., Malik, R. A., Arezzo, J. C., & Sosenko, J. M. (2004). cytokines, and diabetic nephropathy. Scientific World Journal, 6,
Diabetic somatic neuropathies. Diabetes Care, 27, 1458−1486. 908−917.
Boulton, A. J. M., Vileikyte, L., Ragnarson-Tennvall, G., & Apelqvist, J. Nikol, S., Isner, J. M., Pickering, J. G., Kearney, M., Leclerc, G., & Weir, L.
(2005). The global burden of diabetic foot disease. Review. Lancet, 366, (1992). Expression of transforming growth factor-beta 1 is increased
1719−1724. in human vascular restenosis lesions. Journal of Clinical Investigation,
Chantry, D., Turner, M., Abney, E., & Feldmann, M. (1989). Modulation of 90, 1582−1592.
cytokine production by transforming growth factor-beta. Journal of Partanen, J., Niskanen, L., Lehtinen, J., Mervaala, E., Siitonen, O., &
Immunology, 142, 429−435. Uusitupa, M. (1995). Natural history of peripheral neuropathy in patients
Cameron, N. E., & Cotter, M. A. (2008). Pro-inflammatory mechanisms in with non-insulin-dependent diabetes mellitus. New England Journal of
diabetic neuropathy: Focus on the nuclear factor kappa B pathway. Medicine, 333, 89−94.
Current Drug Targets, 9, 60−67. Pedersen, O., & Gaede, P. (2003). Intensified multifactorial intervention and
Delarue, J., Maingourd, C., Couet, C., Vidal, S., Bagros, P., & Lamisse, F. cardiovascular outcome in type 2 diabetes: The Steno-2 study. Meta-
(1998). Effects of oral glucose on intermediary metabolism in bolism, 52, 19−23.
continuous ambulatory peritoneal dialysis patients versus healthy Quattrini, C., Tavakoli, M., Jeziorska, M., Kallinikos, P., Tesfaye, S.,
subjects. Peritoneal Dialysis International, 18, 505−511. Finnigan, J., Marshall, A., Boulton, A. J., Efron, N., & Malik, R. A.
312 J. Ybarra et al. / Journal of Diabetes and Its Complications 24 (2010) 306–312

(2007). Surrogate markers of small fiber damage in human diabetic Smith, A. G., Russell, J., Feldman, E. L., Goldstein, J., Peltier, A., Smith, S.,
neuropathy. Diabetes, 56(8), 2148−2154. Hamwi, J., Pollari, D., Bixby, B., Howard, J., & Singleton, J. R. (2006).
Reeves, W. P., & Androly, T. E. (2000). Transforming growth factor a Lifestyle intervention for pre-diabetic neuropathy. Diabetes Care, 29,
contributes to progressive diabetic nephropathy. Proceedings of the 1294−1299.
National Academy of Sciences of the United States of America, 97, Stamler, J., Vaccaro, O., Neaton, J. D., & Wentworth, D. (1993). Diabetes,
7667−7669. other risk factors, and 12-yr cardiovascular mortality for men screened in
Roberts, R. A., Michel, C., Coyle, B., Freathy, C., Cain, K., & Boitier, E. the Multiple Risk Factor Intervention Trial. Diabetes Care, 16,
(2004). Regulation of apoptosis by peroxisome proliferators. Toxicology 434−444.
Letters, 149, 37−41. Tesfaye, S., Chaturvedi, N., Eaton, S. E., Ward, J. D., Manes, C., Ionescu-
Rubino, A., Rousculp, M. D., Davis, K., Wang, J., Bastyr, E. J., & Tesfaye, Tirgoviste, C., Witte, D. R., & Fuller, J. H., for the EURODIAB
S. (2007). Diagnosis of diabetic peripheral neuropathy among patients Prospective Complications Study Group. (2005). Vascular risk factors and
with type 1 and type 2 diabetes in France, Italy, Spain, and the United diabetic neuropathy. New England Journal of Medicine, 352, 341−350.
Kingdom. Primary Care Diabetes, 1(3), 129−134. Tsuzura, S., Ikeda, Y., Suehiro, T., Ota, K., Osaki, F., Arii, K., Kumon, Y.,
Sandhu, H. S., Butt, A. N., Powrie, J., & Swaminathan, R. (2008). & Hashimoto, K. (2004). Correlation of plasma oxidized low-density
Measurement of circulating neuron-specific enolase mRNA in diabetes lipoprotein levels to vascular complications and human serum
mellitus. Annals of the New York Academy of Sciences, 1137, 258−263. paraoxonase in patients with type 2 diabetes. Metabolism, 53, 297−302.
Schneider, C. A. (2006). Improving macrovascular outcomes in type 2 Turner, R. C., Millns, H., Neil, H. A., Stratton, I. M., Manley, S. E.,
diabetes: Outcome studies in cardiovascular risk and metabolic Matthews, D. R., & Holman, R. R. (1998). Risk factors for coronary
control. Current Medical Research and Opinion, 22(Suppl 2), artery disease in non-insulin dependent diabetes mellitus: United
S15−S26. Kingdom Prospective Diabetes Study (UKPDS:23). British Medical
Shlipak, M. G., Ix, J. H., Bibbins-Domingo, K., Lin, F., & Whooley, M. A. Journal, 316, 823−828.
(2008). Biomarkers to predict recurrent cardiovascular disease: The Vinik, A. I., Maser, R. E., Mitchell, B. D., & Freeman, R. (2003). Diabetic
Heart and Soul Study. American Journal of Medicine, 121, 50−57. autonomic neuropathy. Diabetes Care, 26, 1553−1579.