The role of chemokines in type 1 diabetes‐associated neuropathy

Abstract Introduction To investigate whether circulating chemokines contribute to the development of diabetic peripheral neuropathy (DPN) in patients with type 1 diabetes (T1D). Methods Fifty‐two patients with childhood‐onset T1D (mean age 28 ± 4 yrs.; diabetes duration 19.5 ± 5.5 yrs.) and 19 control subjects (mean age 26.5 ± 4.5 yrs.) were included in a cross‐sectional analysis of this long‐term longitudinal cohort study. A subgroup of 24 patients was followed prospectively for a further 10 yrs. Plasma levels of Th1‐ (CXCL9, CXCL10 and CXCL11), Th2‐ (CCL17 and CCL22) and Th17‐associated (CXCL8 and CCL20) chemokines were assessed in all subjects. Additionally, the TID patients underwent clinical examination and electroneurography. Results The frequency of neuropathy was 21% (11/52). Higher levels of CXCL9 levels were found in patients with DPN compared with control subjects (p = .019); by contrast, no difference between patients without DPN and control subjects was seen after adjustment for multiple comparisons. In patients with DPN, CXCL10 correlated negatively with suralis MCV and suralis SNAP (rho −0.966, p < .001 and rho −0.738, p < .001, respectively) and was positively correlated with the vibration perception threshold (rho 0.639, p = .034), while CXCL8 correlated negatively with the cold perception threshold (rho −0.645, p = .032). The frequency of neuropathy increased to 54% (13/24) in the subgroup of 23 TID patients, followed by an additional 10 yrs. Conclusions Changes in Th1‐ and Th17‐associated chemokines were associated with impaired peripheral sensory nerve function and nerve conduction after long disease duration in childhood‐onset T1D.


| INTRODUC TI ON
Diabetic peripheral neuropathy (DPN) is a common microvascular complication in type 1 diabetes (T1D), and it occurs in nearly 50% after a long disease duration. Currently, much attention has been on the role of immunological factors such as chemokines in the development of diabetic neuropathy and neuropathic pain. 1 Chemokines or chemotactic cytokines are small (8-14 kDa) secreted peptides involved in innate and adaptive immune responses. [2][3][4] Chemokines have a crucial role in establishing and maintaining the type of T helper preponderance in the immune response and may be used as reliably detectable markers for Th1-, Th2-and Th17-associated immunity. 3,5-10 CXCL9, CXCL10 and CXCL11 are ligands for the receptor CXCR3 and are mainly secreted in response to interferon (IFN)-ɣ, which TNF synergistically enhances. The CXCL9, CXCL10, CXCL11/CXCR3 axis regulates migration of activated Th1 cells, as they preferentially express CXCR3. 11 On the contrary, the CCR4 ligands CCL17 (thymus-and activation-regulated chemokine) and CCL22 (monocyte-derived chemokine) are induced by the Th2 cytokines IL-4 and IL-13, and particularly recruit Th2 cells, preferentially expressing CCR4. 12 The chemokines CXCL8 and CCL20 are induced by IL-17 from Th17 cells, preferentially expressing CCR6, binding CCL20. 13 Experimental evidence accumulated in the last years supports the concept that IFN-ɣ inducible Th1-associated chemokines CXCL9, CXCL10 and CXCL11 may play a critical role in developing several autoimmune disorders and in beta cell destruction in T1D. 3,14,15 Elevated chemokines are involved in the development of diabetic complications such as retinopathy and nephropathy. 16,17 Furthermore, some studies indicate that CXCL9 is important in nociceptive transmission contributing to the development of diabetic pain, which is one of the most disabling symptoms in diabetic neuropathy. 18,19 Schwann cells, myelination cells of the peripheral nervous system, play an essential role in the pathogenesis and development of DPN. Studies indicate that the pathophysiological mechanism behind DPN is when the high glucose levels promote the expression of CXCR3 on CD8+ T cells and, on the contrary, induce the expression of CXCL9, CXCL10 and CXCL11 in Schwann cells, leading to the recruitment and infiltration of CD8 + T cells into diabetic peripheral tissues and increased cytotoxicity towards Schwann cells. 20 In addition, Th17 cells may contribute to damage to peripheral neurons through the cytokine IL-17, which may lead to diabetic neuropathy. 13 However, while IL-17 directly modulates neuronal cell behaviours in vitro, it is unclear if these effects also extend to T1D in vivo. 21 Given these considerations, we hypothesized that diabetic neuropathy in individuals with childhood-onset T1D is associated with increased levels of Th1 and Th17 chemokines and conversely with low levels of Th2 chemokines compared with control subjects and patients without neuropathy.

| Subjects and study design
Patients with childhood-onset T1D (N = 102) prospectively followed in a longitudinal cohort study, since childhood and examined on at least one previous occasion with nerve conduction tests, 22,23 were asked to participate in the current cross-sectional analysis at the exact center, the Endocrinology Department of Linköping University Hospital. In all, 52 patients were admitted to participate in the present study from 2007 to 2009. The diagnosis of type 1 diabetes was based on the presence of high blood glucose levels, increased glycated haemoglobin (HbA1c) in addition to the positive titre of at least one of the autoantibodies related to T1D; GAD (glutamic acid decarboxylase antibody) and IA-2 (protein tyrosine phosphatase/insulinoma associated autoantibodies). All patients had been receiving intensive therapy from disease onset, which involved the administration of insulin

| Chemokine analyses
Chemokine concentrations used as markers for Th1-, Th2-and Th17associated immunity ( Table 2) were measured in EDTA plasma by Luminex multiple bead technologies (Milliplex Human Cytokine/ Chemokine Kit, Millipore Corporation), according to the instructions provided by the manufacturer. Samples from patients and controls were analysed in parallel using the same reagent batch. The lower detection limits for chemokines in plasma were CXCL8, 1.6 pg/mL, CXCL9, 3.9 pg/mL, CXCL10 16 pg/mL, CXCL11 7.8 pg/mL, CCL17 3.9 pg/mL, CCL20 4.9 pg/mL and CCL22 80 pg/mL. Values below the detection limit were given half the value of the detection limit.
All samples were analysed in duplicate, and the samples were reanalysed if the coefficient of variance (CV) was >15%.

| Statistical analysis
Categorical variables are expressed as counts and percentages, or  (Table 1 and Table 2) In total, 52 patients (20 women), 20-35 years of age (mean 27.3 SD ± 4.1), and with a disease duration of 19.5 ± 5.5 years, were included. For comparison, a control group was included (11 women: mean age 26.5 ± 4.5 years).

| Demographic and clinical characteristics
All patients were divided into two groups according to the pres- The frequency of neuropathy was 21% (11/52). Nine of the 11 patients with DPN also had signs of clinical diabetic neuropathy. In addition, four patients from the neuropathy group reported symptoms TA B L E 1 Demographics and clinical data for the two groups with type 1 diabetes without neuropathy (TID) and with neuropathy (TID + DPN).  (Figure 1, Table 2)

| Chemokines
In univariate tests, the Th1-associated chemokine CXCL9 was significantly higher in patients with T1D than in healthy controls.
However, only patients with neuropathy had significantly higher CXCL9 levels compared to controls, p = .019, while no difference was found between patients without neuropathy and controls after adjustment for multiple testing. In addition, no significant differences were found between patients with TID with and without neuropathy, or in comparison with healthy controls regarding the other Th1-associated chemokines (CXCL10, CXCL11), the Th2-related chemokines (CCL17, CCL22) or the Th17-associated chemokines CCL20 and CXCL8, after adjustment for multiple testing.

| Electroneurography and chemokines (Table 3)
Bilateral measurements of both peroneal and median motor nerve

| Peripheral nerve tests and chemokines (Table 4)
In order to explore possible associations between chemokines and peripheral sensory nerve function, plasma concentrations of chemokines were associated with three tests of sensory perception ability: heat tolerance test, cold perception threshold and vibration perception in which the nerve function is studied in myelinated C, TA B L E 2 Chemokine levels according to the presence of diabetic polyneuropathy (DPN) in patients with type 1 diabetes (T1D) and in comparison with a control group without diabetes.

F I G U R E 1
In the Kruskal-Wallis H test, significantly higher CXCL9 levels were found between control subjects and patients with DPN (p = .019), while no difference between control subjects and patients without DPN was seen after adjustment for multiple comparisons.
Aδ and Aβ nerve fibres, respectively. For the cold perception, a significant and negative correlation was found for the Th17-associated

| Longitudinal follow-up study
A subgroup of 24 of the cohort of 52 patients was followed up and re-examined a third time with neurophysiological tests after ten years. Of these, ten patients were still free of clinical and subclinical neuropathy; seven patients had persisting neuropathy, while six new patients had developed neuropathy from the previous follow ten years earlier, giving a frequency of neuropathy of 54% (13/24).
No patient had recovered from a prior diagnosis of neuropathy. If the chemokine expression a decade earlier was recalculated in these 24 patients and compared with the former control subjects (n = 19), the same relationship was found that higher levels of CXCL9 levels were expressed in patients with DPN compared with control subjects (p = .021), while no difference between patients without DPN and control subjects were seen after adjustment for multiple comparisons.

| DISCUSS ION
Recently the role of chemokines in the pathogenesis of diabetic neuropathies has been investigated, mostly in animal studies. 19 In the present study, the IFN-ɣ-induced chemokine CXCL9, which reg-

TA B L E 4
Correlations between chemokine levels and quantitative sensory tests (QSTs) were performed bilaterally according to standardized procedures within the groups of patients without diabetes neuropathy (TID) and with diabetes neuropathy (TID + DPN).
serum levels of CXCL10 were not found in patients with T1D compared to healthy controls or between the two groups of T1D individuals. A possible explanation could be the long duration of diabetes disease of 2 decades in our cohort from the onset of diabetes. 2 This is also the case in another cohort study with patients with T1D and similar diabetes duration. 27 On the contrary, several negative correlations were found between suralis nerve conduction data and CXCL10 levels, indicating that Th1-mediated inflammation may be associated with diabetic neuropathy severity. In the same direction, a positive correlation between vibration threshold and CXCL10 levels was found. These . 29 In addition, Th17 cells may contribute to damage to peripheral neurons through the cytokine IL-17, which may lead to diabetic neuropathy. 13 Taken together, therapeutic prospects, which aim to modulate the Th1/Th17 system in relation to disease duration would be exciting to explore in the cases of diabetic complications such as DPN if further deterioration could be inhibited.
There are some limitations to our study. First, although an association was found between CXCL9 and T1D, and several significant correlations between pathological nerve function tests and Th1 and Th17-induced chemokines were noted, the causality between inflammation factors and DPN cannot be established. It would be interesting due to prospectively designed studies to identify changes in biomarker levels before the onset of DPN and metabolic control.
Our patients had a long diabetes duration, and several patients already suffered from DPN. The control subject was well matched due to age, but the proportion of women was more prominent in the control group compared with the patients with type 1 diabetes, which may have affected the results. On the contrary, to our knowledge there is no gender differences in chemokines distribution.
In this novel pilot study that investigated nerve conduction tests and chemokines levels in patients with type 1 diabetes. Another limitation of our study is that the current cohort sample is relatively small but hopefully our results can form the basis of a large prospective study that may give further answers and explanations and facilitate the research on pathogenesis-derived pharmacotherapy. 30,31

| CON CLUS IONS
Higher levels of the Th1-associated chemokine CXCL9 were detected in patients with T1D and DPN after two to three decades of diabetes duration compared to healthy control subjects. Furthermore, Th1 and Th17-associated chemokines were associated with impaired peripheral sensory nerve function and nerve conduction. Our

ACK N O WLE D G E M ENTS
We want to acknowledge coordinating nurses and biochemical medical assistants at the Clinical research center and Neurophysiological Clinic. Special thanks to all patients with type 1 diabetes and control subjects participating in the study.

FU N D I N G I N FO R M ATI O N
Grants from the Swedish state financed this study under the agreement between the Swedish government and the county councils, the ALF agreement, LIO 89939, and by grants from the Region of Östergötland, LIO-790841 and LIO711061.

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data generated or analysed during this study are available from the corresponding author upon reasonable request.

A PPROVA L O F TH E R E S E A RCH PROTO CO L
The study protocol was approved by the Research Ethics Committee,