Skin autofluorescence predicts cardio-renal outcome in type 1 diabetes: a longitudinal study

Background We aimed to analyze the relationships between skin autofluorescence (SAF) and incident macrovascular events and renal impairment after 4 years of follow-up in patients with type 1 diabetes (T1D). Methods Two hundred and forty-three patients (51.2 ± 16.7 years old) with T1D participated. SAF was measured by AGE-Reader-TM at inclusion. Macrovascular events (MVE), estimated glomerular filtration rate (eGFR) and urinary albumin excretion rate (AER) were recorded then and 4 years later. Multivariate logistic regression was used to analyze the relationships between SAF and incident MVE and renal profile 4 years later. Results Patients with incident MVE had a higher SAF (p = 0.003). SAF predicted incident MVE after adjustment for age, sex, body mass index, tobacco, diabetes duration, hypertension, HbA1c, AER, eGFR (OR 4.84 [95 % CI 1.31–17.89], p = 0.018). However, this relation was no longer significant after adjustment for history of MVE. An inverse relation was found between SAF and incident eGFR (p = 0.0001). Patients with incident eGFR <60 ml/min/1.73 m2 had a SAF higher than patients with normal eGFR. After adjustment for the previous criteria, SAF remained associated with the risk of impaired incident eGFR (OR 7.42 [95 % CI 1.59–34.65], p = 0.018). No relation was found between SAF and increased AER 4 years later. Conclusions SAF predicts MVE in patients with T1D, adjusted for cardiovascular risk factors but the most powerful predictive factor remains history of MVE. SAF also predicts eGFR impairment, adjusted for initial AER and renal function. SAF could be a useful non-invasive tool for estimating risk of cardiovascular or renal impairment in patients with T1D.

promote inflammation and endothelial dysfunction [4]. Their accumulation is increased in CKD, due to reduced clearance of their precursors [5]. High levels of circulating AGEs predict later mortality and cardiovascular events in T1D [6], and high levels of AGEs in skin biopsies predict microangiopathy [7] and sub-clinical cardiovascular disease in the EDIC study [8]. But the complex time consuming analysis of AGEs, and the invasiveness of skin biopsies preclude the use of AGEs for prognosis in a normal clinical routine.
This may change with the development of non-invasive methods for measuring AGEs, based on their fluorescent properties: the skin autofluorescence (SAF) is well correlated with the dermal concentrations of AGEs [9], its measurement is safe, quick and simple. SAF has been shown to predict later microangiopathy [10] and cardiovascular events [11] in patients with type 2 diabetes mellitus (T2D). To date no longitudinal study has been performed in T1D.
We measured SAF in 300 patients with T1D during year 2009. Our baseline results, which show that SAF was related to retinopathy, nephropathy and long term past glucose control, have been published elsewhere [12]. In the present study, we investigated whether SAF was related to their cardio-renal outcome within 4 years of follow-up: cardiovascular events, estimated GFR (eGFR), and AER. Data were available in 243 patients after 4 years of follow-up.

Design and patients
We analyzed the data from two hundred and fortythree patients with T1D included during the year 2009 after written informed consent. The following data were recorded at inclusion, on the day of SAF measurement: age, sex, body mass index (BMI), duration of diabetes, hypertension (defined by a blood pressure ≥140/90 mm Hg and/or treatment with antihypertensive drug) and smoking history. Smokers included patients with past and current smoking history. The biological data recorded included HbA 1c levels, serum creatinine levels, eGFR calculated with the CKD-EPI formula and AER.

Skin autofluorescence (SAF)
The accumulation of advanced glycation end-products was evaluated from SAF measured by illumination of 1 cm 2 of the forearm skin by the AGE reader SU 2.3 (DiagnOptics BV, Groningen, the Netherlands) at inclusion. This technique uses fluorescent properties of some AGEs to estimate AGE accumulation in the skin [9]. The patients with Fitzpatrick phototypes V and VI were not included due to their skin pigmentation, which had ultraviolet reflectance of <10 %. SAF values were calculated by dividing the mean emitted light intensity (excitation light source ranging from 300 to 420 nm) by the mean reflected excitation light intensity from the skin (over 300 to 420 nm). Results were the mean of 3 consecutive measurements of SAF and were expressed in arbitrary units (AU).

Outcomes
The patients were prospectively followed-up and incident macrovascular events (MVE) and renal outcome were recorded after 4 years. MVE were defined as myocardial infarction, stroke, gangrene, or a revascularization procedure for coronary, carotid or lower limb arteries.
The renal outcome analysis included AER (mg/24 h) and renal function (eGFR, ml/min per 1.73 m 2 ), measured 4 years after inclusion.

Statistical methods
The data are presented as mean ± SD for continuous variables and as numbers (percentages) for categorical variables. Student's t test was used to compare continuous variables.
A logistic regression was used to investigate the associations between SAF and MVE, after adjustment for the following co-variables: age, sex, BMI, hypertension, duration of diabetes and tobacco. A multivariate regression analysis was used to investigate the associations between SAF and later AER and eGFR, after adjustment for the following co-variables: age, sex, hypertension, duration of diabetes and tobacco. We also performed complementary stepwise backwards multivariate logistic regression analyses. All analyses were performed with SAS Statistical Package release 9.3 (SAS Institute, Cary, NC) and results were considered significant when p < 0.05.

Skin autofluorescence and renal outcome
Mean SAF was higher in patients with impaired renal function (eGFR <60 ml/min/1.73 m 2 ) at baseline compared to patients with normal eGFR (2.69 ± 0.67 vs 2.05 ± 0.51 AU, p < 0.0001). There was no difference between SAF of patients according to AER at baseline.
We did not find any relation between SAF and increased AER 4 years later after complete adjustment including the history of increased AER (Table 3). By complementary stepwise backwards-multivariate logistic regression analysis, we found that three variables were related to a final AER ≥30 mg/24 h: age, initial eGFR <60 ml/min/1.73 m 2 and initial AER ≥ 30 mg/24H.

Discussion
In our study of 243 patients with T1D, SAF measured during year 2009 relates to the risk of MVE and impaired eGFR 4 years later. Our study is the first to show a relation between SAF and incident MVE in T1D: as shown in Table 1, SAF predicts incident MVE during the four following years. This relation is not significant after Table 1 Relationships between skin autofluorescence and macrovascular events after 4 years of follow-up in subjects with type 1 diabetes SAF skin autofluorescence; CKD chronic kidney disease; MVE macrovascular events a Missing data: BMI (n = 10); HbA 1c (n = 1); Tobacco (n = 13); Albumin Excretion Rate (n = 6); eGFR (n = 7)  [1,2]. Furthermore, we found that SAF predicts later impairment of eGFR in our patients. This relation persists after adjustment to the history of impaired eGFR. By contrast to the relationship between SAF and MVE, incident impaired eGFR is predicted by SAF irrespective of renal function at baseline. We also showed that SAF predicts later eGFR <60 mL/min/1.73 m 2 adjusted for initial AER. Otherwise, we did not find any relationships between SAF and AER after 4 years in contrast to eGFR. This result is not surprising as the studied population was followed in a university hospital with systematic preventive treatment of diabetic nephropathy including renin-angiotensin-system blockers. Moreover, the renal outcome of patients with T1D has been reported as being independent from AER changes that was more related to glycemic, blood pressure and lipid profile control [13]. The early loss of renal function in patients with T1D may begin before onset of proteinuria, although the mechanism Table 2 Relationships between skin autofluorescence and impaired eGFR (<60 mL/min/1.73 m 2 ) after 4 years of followup in subjects with type 1 diabetes a Missing data: BMI (n = 10); HbA 1c (n = 1); tobacco (n = 13); albumin excretion rate (n = 6); eGFR (n = 7); EpiCKD (n = 30)

Table 3 Relationships between skin autofluorescence and increased AER after 4 years of follow-up in subjects with type 1 diabetes
SAF skin autofluorescence; CKD chronic kidney disease; MVE macrovascular events a Missing data: BMI (n = 10); HbA 1c (n = 1); tobacco (n = 13); albumin excretion rate (n = 6); eGFR (n = 7) A prospective study about the role of SAF to predict cardiovascular events in patients with T2D has been reported [11], but such results have yet to be described in patients with T1D. Only cross sectional studies were performed in T1D and showed that SAF was associated with subclinical markers of macroangiopathy as intimamedia thickness [14] and arterial stiffness [15,16]. Furthermore, skin fluorescence could be measured with the SCOUT DM ® (VeraLight, Inc., Albuquerque, NM) another method for assessing intra-patient skin variation irrespective of the Fitzpatrick phototype. Skin intrinsic fluorescence could be related to the severity of coronary calcifications [17] and the intima-media thickness [18]. In 172 patients with long duration T1D, it was associated with a clinical history of coronary heart disease, but the relation lost significance after adjustment for nephropathy [19], similar to our finding at baseline. The EURO-DIAB study has already reported that increased levels of soluble receptor for advanced glycation end products (RAGE) were related to CV disease in T1D [20]. In addition, high circulating levels of AGEs were reported to predict incident CVD in T1D [6], and their concentrations in skin biopsies were related to coronary calcifications, left ventricular masses and progression of the intima-media thickness in the DCCT-EDIC study [8]. Consequently, SAF, a non-invasive tool for estimating accumulation of AGE could be linked to vascular modification and thus assess risk of MVE. However, the value of SAF as a surrogate marker to estimate tissues AGEs accumulation and to predict MVE is a matter of debate [21]. Indeed, SAF has been found to be correlated with skin levels of AGEs such as pentosidine, N-(carboxymethyl) lysine and N-carboxyethyl-lysine, and positively associated with age, diabetes duration, and mean glycated hemoglobin of the previous year in both type 1 and type 2 diabetic patients [22]. Moreover, a relation between SAF and cardiac tissue glycation has been reported [23]. SAF levels could depend on their tissue turnover, so it must be distinguished from the plasma concentrations of AGEs [21]. However, SAF remains a reliable clinical tool for evaluating tissue accumulation levels of AGEs. Indeed, tissue levels of AGEs estimated by skin fluorescence are a potential biomarker to evaluate vascular risks in patients with diabetes [17][18][19]. Finally, further longitudinal studies are needed to clarify whether SAF could be a predictor of the development and progression of long-term MVE. In the same way, the significance of the association between SAF and MVE after adjustment for AER and eGFR also seems important, since diabetic nephropathy is a major predictor of the cardiovascular risk in T1D [1,2].

Univariate analysis OR (CI 95 %)
In this context, SAF has been shown to predict later microalbuminuria in patients with T2D [10], and the doubling of serum creatinine or the need for dialysis in CKD [24]. In the same way, cross-sectional studies performed in T1D have shown associations of SAF with AER and eGFR [12,25]. Moreover, a longitudinal study was especially needed for renal markers, because high SAF may favor, but also may result from impaired GFR [26]. Our current results show that SAF predicts later eGFR <60 mL/min/1.73 m 2 , adjusted for initial AER and eGFR. This is consistent with the recent report that AGEs in skin biopsies related to later increase of serum creatinine in patients with both types of diabetes [27]. Because the relation between SAF and later low eGFR persists after adjustment to AER, we suggest that AGEs may play a role in the yet unexplained rapid decline in GFR observed in some patients with T1D, with [28] or without microalbuminuria [13]. Indeed, high serum levels of AGEs have been related to the progression of glomerular basal membrane in normoalbuminuric T1D patients [29]. Moreover, mean values of skin collagen linked AGEs such as furosine, carboxymethyl-lysine, pentosidine, and fluorescence were significantly higher in the patients with progression of nephropathy [7]. AGEs have been shown to be significant predictors of nephropathy, independently of the mean HbA1c [7].
The prediction of cardiorenal events from SAF may reflect its value as an integrator of long-term glucose control, as a marker of "metabolic memory". In this context, we have recently shown that SAF better relates to plasma glucose levels, HbA 1c and renal function 10 years earlier than to their assessment at the time of SAF measurement in elderly patients from the general population [30]. The risk of microvascular complications like retinopathy has indeed been related more to ancient than recent HbA 1c in T1D [31]. In a previous study, we noted that SAF was better correlated with the previous years than present HbA 1c [12], in line with other authors' findings in T1D [14,25,32]. Measures of skin fluorescence in the DCCT were no longer related to complications after adjustment to the mean HbA 1c [33], which suggested that the relation with complications is supported by metabolic memory, although a major part of the variance of skin fluorescence is not explained by conventional factors such as HbA 1c , age and GFR [34]. However, a more recent work has shown that skin fluorescence could be related to subclinical cardiovascular disease independent of duration of diabetes and HbA 1c [18], possibly due to oxidative stress. Taken together these data indicate new therapeutic strategies for reducing diabetic complications. The clinical review of Goh and Cooper summarized the pharmalogical options that could modulate either AGE ligands accumulation or AGE-RAGE interaction [35].
Our study has some limitations. The inclusion of patients with longstanding T1D (mean duration: 21 years), often treated by medication acting on the Renin-Angiotensin System, was logical to study "hard" complications as MVE, but it may have precluded the finding of earlier events such as microalbuminuria as reported in recent and well-controlled patients with T2D [10]: the incidence of microalbuminuria is known to naturally decline after 20 years of diabetes [36]. The limited number of patients and the percentage of lost subjects represent another limitation. Moreover, the patients experienced few events over the follow-up, with thus large confidence intervals for the association between SAF and events.

Conclusions
SAF that is strongly related to tissues biopsy levels of AGEs, has value for predicting cardiorenal outcome in our patients with T1D. This relation is independent of the usual risk factors, which supports that long-term accumulation of AGEs in tissues contributes to diabetic complications, especially nephropathy. The simple, cheap, and quick measurement of SAF makes it a good candidate for the screening of patients at risk of diabetic nephropathy.

Authors' contributions
FLV and VR analyzed research data and wrote the manuscript. KR researched data. CH helped in the analysis of data and statistical work. SN, MH, LB, EP and BF helped in the recruitment of patients. CA helped for statistical work. HG reviewed the manuscript and helped with editing. All authors read and approved the final manuscript.