Non-invasive evaluation of advanced glycation end products in hair as early markers of diabetes and aging

Abstract

Continuous metabolic monitoring is essential for assessing lifestyle-related disease risks. Hair, an easily accessible tissue, allows for long-term metabolic evaluation, with glycated proteins linked to diabetic complications found in hair. We established a mass spectrometry system to detect advanced glycation end products (AGEs) in hair samples from humans and rats, assessing their variations with aging and disease. Hair samples were hydrolyzed and processed using a cation-exchange column for mass spectrometric analysis. Regardless of temperature variations, the levels of AGEs [N^ε-(carboxymethyl)lysine (CML), and methylglyoxal-derived hydroimidazolone-1 (MG-H1)] in human hair remained stable for one week. Age and CML levels, or AGEs z-scores combined with CML and CEL levels in human hair samples, were positively correlated. In streptozotocin-induced insulin-deficient diabetic model (DM) rats, hair CEL and MG-H1 levels were higher than in non-diabetic rats. Receiver operating characteristic curve analysis showed an area under the curve of 1 for hair CEL and MG-H1 levels. Serum and hair CML levels were positively correlated. Hair AGE levels vary more between DM and non-DM rats than serum AGE levels. They remain stable under heat treatment and correlate with age, indicating that hair analysis is an effective non-invasive method for assessing metabolic fluctuations.

Introduction

The International Diabetes Federation (IDF) reported that in 2013, approximately 380 million people worldwide were living with diabetes, and this number is expected to increase to 540 million by 2021, with projections indicating a continued rise¹. Diabetes is associated with complications such as nephropathy, retinopathy, and neuropathy, which are difficult to treat. Therefore, there is an urgent need for biomarkers that can predict disease progression.

Hemoglobin A1c (HbA1c) is used to evaluate glycemic control in diabetes and reflects blood glucose levels over the preceding 1–2 months. High HbA1c levels are a risk factor for diabetic complications; however, HbA1c alone is inadequate for precisely predicting long-term complications such as nephropathy and cerebrovascular disease. Moreover, high HbA1c levels do not always lead to the development of diabetic complications². Given the difficulty in reversing these complications after their onset, early detection and prevention are essential. Some studies have demonstrated that advanced glycation end products (AGEs) from biological samples could serve as indicators of disease onset and progression^{3, 4–5}. However, the necessity for clinical visits for blood sampling makes frequent self-monitoring challenging. In Japan, 35% of individuals suspected of having diabetes do not receive any treatment⁶, with the major reason being the lack of time for hospital visits. A more accessible and less invasive sampling method could facilitate regular metabolic assessments. Hair, which can be self-collected, is a promising option for first-line metabolic screening before more precise blood-based diagnostics. For example, an increase in adipic acid levels in the hair of pregnant women has been reported as a potential biomarker for gestational diabetes mellitus (GDM)⁷.

AGEs have been proposed as potential metabolic markers for diabetes and its complications⁴. AGEs are formed by non-enzymatic glycation reactions between reducing sugars and proteins or nucleic acids via the Maillard reaction⁸. Initially, the carbonyl groups of the reducing sugars react with the amino groups of proteins or nucleic acids, leading to the formation of Schiff bases, 1,2-enamines, and Amadori rearrangement products. In subsequent phases, oxidation, dehydration, and condensation reactions result in the formation of AGEs. Additionally, AGEs can be formed through the oxidative degradation of reducing sugars or Amadori products, which generate dicarbonyl intermediates, such as glyoxal⁹, leading to a variety of structurally diverse AGEs.

Among the various AGEs, N^ε-(carboxymethyllysine) (CML) has been linked to oxidative stress¹⁰, N^ε-(carboxyethyllysine) (CEL) to dyslipidemia¹¹, and methylglyoxal-derived hydroimidazolone-1 (MG-H1) to glucose metabolism disorders¹². Furthermore, elevated levels of serum protein-bound¹³ and free amino acid-derived AGEs¹⁴, including CML, CEL, and MG-H1, have been observed in diabetes and its complications³.

Although serum AGE measurements provide valuable clinical insights, blood sampling requires trained medical personnel and proper sample handling, making frequent assessments challenging. Skin AGEs can be measured noninvasively using AGE sensors¹⁵; however, fluorescence-based techniques do not enable the precise quantification of specific AGE structures. Hair undergoes glycation¹⁶, and immunological methods¹⁷ have been used to detect glycated proteins in hair samples. Human hair consists of a hair root embedded in the stratum corneum of the skin and a hair shaft that extends externally. The dermal papilla is connected to capillaries at the base of the hair root and is surrounded by matrix cells and melanocytes¹⁸. Hair matrix cells incorporate melanin produced by melanocytes, along with components from the bloodstream, undergo keratinization, and contribute to hair growth at a rate of approximately 1 cm/month in humans¹⁷. In rats, the growth rate ranges from approximately 0.15 cm/day¹⁹. Cortisol, a component of blood, has been identified as a factor influencing hair, and elevated hair cortisol levels have been associated with a potential risk for cardiovascular disease²⁰. Hair metabolites have been used for drug testing²¹ and disease assessment²².

Therefore, hair AGE levels may indicate metabolic abnormalities and disease status in patients. Based on this hypothesis, we developed a method to more accurately quantify hair AGEs using mass spectrometry and evaluated their stability and correlation with aging. In this study, we focused on a 1 cm hair length from the root in both humans and rats, as the levels of AGEs from different parts of the hair length may not differ significantly during the short term²³.

Results

Measurement of ages in human hair

Unlike serum, measuring AGEs in hair requires extraction. Attempts to pulverize hair have been hindered by static electricity, complicating precise sample management. Some hair fragments were detected after melting. Hair samples were dissolved using hydrochloric acid. Using mass spectrometry, AGEs were successfully detected in the hair of healthy individuals (Supplementary Figure S1). Moreover, Fig. 1 shows the structure of each AGE and the internal standard used for AGEs.

Fig. 1 [Images not available. See PDF.]

Structures of AGEs and the position of the deuterium atom. The asterisk (*) indicates the position of the deuterium atom. (a) N^ε-(carboxymethyllysine) (CML), (b) N^ε-(carboxyethyllysine (CEL), (c) methylglyoxal-derived hydroimidazolone-1 (MG-H1).

Because the stability of each AGE in hair, unlike in serum, is not well understood (particularly under varying conditions such as temperature), we evaluated the effect of storage temperature on AGE levels in one male subject. The results showed that CML, MG-H1, Arg, and Lys levels remained unchanged after one week of storage at all tested temperatures, including room temperature (Fig. 2(a), (c)-(e)). CEL levels did not show a significant change at − 80 and 4 °C, whereas CEL levels decreased at 40 °C after one week (Fig. 2(b)).

Fig. 2 [Images not available. See PDF.]

Advanced glycation end products (AGEs) in human hair stored at various temperatures. (a) Levels of N^ε-(carboxymethyl)lysine (CML), (b) levels of N^ε-(carboxyethyl)lysine (CEL), (c) levels of methylglyoxal-derived hydroimidazolone-1 (MG-H1), (d) lysine (Lys) levels, (e) arginine (Arg) levels. Pink: 40 °C, green: 4 °C, blue: −80 °C. n = 3. Statistical analysis: Wilcoxon rank-sum test; * P < 0.05.

For practical applications, we conducted measurements using five strands of hair per sample. The coefficient of variation (CV) for all AGEs was less than 20% (Supplementary Figure S2). Previous studies have reported an association between intrinsic AGE levels in serum and aging^3,24. To investigate whether a similar correlation exists in hair, we analyzed hair samples from 30 Japanese women aged 30–60 years in this study. In the correlation analysis between hair AGE levels and age, we focused exclusively on female hair samples to eliminate the influence of hormones related to androgenetic alopecia and sex differences.

In our previous study, we evaluated pathological conditions using AGE levels in biological samples, with AGE content assessed based on a z-score derived from multiple AGE values^3,4. Consequently, only the CML levels and z-scores for AGEs (CML + CEL) showed a tendency to increase with aging (Fig. 3(a)–(d)). Therefore, we conducted a correlation analysis between hair AGE levels, z-scores, and age. The results demonstrated a positive correlation between hair CML levels and age (CML-age: R = 0.33, P < 0.018), as well as between AGEs z-scores (CML + CEL) and age (AGEs z-scores-age: R = 0.36, P < 0.0017) (Fig. 3(e)).

Fig. 3 [Images not available. See PDF.]

Correlation between AGEs levels in hair and age in healthy participants. Hair AGEs levels of 30 healthy females and their ages. (a) Hair N^ε-(carboxymethyl)lysine (CML), (b) Hair N^ε-(carboxyethyl)lysine (CEL), (c) Hair methylglyoxal-derived hydroimidazolone-1 (MG-H1), (d) Z-score with Hair CML + CEL levels and ages. Statistical analysis: Spearman’s analysis was performed with a significance level set at P < 0.05. Crosses inside the circles indicate non-significant correlations.Green and pink text indicate positive and negative correlations, respectively.

Measurement of ages in hair and disease assessment in a diabetic rat model

To explore the potential application of hair AGE measurements in patients with disease, hair AGE levels in streptozotocin (STZ)-induced insulin-deficient diabetic model (DM) rats were analyzed (Supplementary Figure S3) and compared with those in non-diabetic control rats. Serum CML and CEL levels showed an increasing trend. In contrast, hair CEL and MG-H1 levels were significantly elevated in STZ-induced insulin-deficient DM rats (Fig. 4). Additionally, hair CML and CEL levels tended to be higher than those in the serum.

Fig. 4 [Images not available. See PDF.]

Comparison of AGEs levels in serum and hair in diabetes (DM)-induced and non-diabetic (control) rats at 12 weeks. (a) Serum N^ε-(carboxymethyl)lysine (CML), (b) Hair CML, (c) Serum N^ε-(carboxyethyl)lysine (CEL), (d) Hair CEL, (e) Serum methylglyoxal-derived hydroimidazolone-1 (MG-H1), (f) Hair MG-H1 levels. Control (n = 7) and DM (n = 6). Data are shown as box plots, with bold horizontal bars indicating median values. Statistical analysis: Wilcoxon rank-sum test; * P < 0.05; ** P < 0.01; *** P < 0.001.

Blood samples are commonly used to assess diabetes and its associated complications. However, to evaluate whether easily obtainable hair samples could effectively distinguish diabetes status, we conducted a receiver operating characteristic (ROC) analysis. AGEs that showed significant differences between the control and diabetic rats (Fig. 4) were selected for further analysis. The area under the curve (AUC) values for the candidate AGE biomarkers were as follows: CEL (AUC = 1) and MG-H1 (AUC = 1) (Fig. 5).

Fig. 5 [Images not available. See PDF.]

Receiver operating characteristic (ROC) analysis of hair AGE levels in rats. ROC analysis was performed to assess the discriminatory ability of hair AGE levels and to determine the optimum cutoff point. The box plot shows individual sample values from the control (green, non-DM) and diabetes-induced (red, DM) rat groups. (a) Hair N^ε-(carboxyethyl)lysine (CEL) and (b) Hair methylglyoxal-derived hydroimidazolone-1 (MG-H1) levels.

Finally, we examined whether changes in systemic AGE levels were reflected in hair by performing a correlation analysis between serum and hair AGE levels. A positive correlation (R = 0.42) was observed between serum CML and hair CML levels, whereas a negative correlation (R = − 0.45) was observed between serum MG-H1 and hair MG-H1 levels (Fig. 6).

Fig. 6 [Images not available. See PDF.]

Correlation between hair and serum AGE levels in rats. Statistical analysis: Spearman’s analysis; significance level was set at P < 0.05. Green and pink text indicate positive and negative correlations, respectively.

Discussion

Previous studies have investigated the potential of hair glycation as an indicator of pathological conditions, such as diabetes. For example, keratin, a hair protein, undergoes glycation²⁵, and the AGE precursor fructosamine²⁶ is subjected to acid hydrolysis to produce N^ε-(2-furoylmethyl)lysine (furosine)²⁷. Furthermore, AGEs in the hair of both diabetic and non-diabetic individuals have been measured using immunological techniques¹⁷. However, due to the unclear epitope specificity of anti-AGE antibodies²⁸ and the technical challenges associated with the immunological quantification of AGEs in solid samples, this method has not been widely adopted.

Previous studies have reported approximately 1.7-fold increased serum CML levels in diabetic rats²⁹. In contrast, our findings showed only a slight increase in serum CML levels. This variation may be attributed to the relatively stable nature of circulating CML, which does not significantly change with the severity of diabetes. Notably, our study found a greater increase in CEL and MG-H1 levels in the hair than in the serum of diabetic rats. This may be due to enhanced glucose uptake by actively proliferating hair matrix cells after diabetes onset, leading to increased production of methylglyoxal³⁰, a precursor of CEL and MG-H1 via glycolysis¹². Furthermore, it is possible that CEL and MG-H1 in the bloodstream of diabetic rats are partially incorporated into growing hair, similar to the integration of blood caffeine into hair³¹.

While tissues such as blood, lymph, and the brain contain 80–90% water, hair has a lower water content of approximately 15%³². Due to this low moisture level, the frequent formation of AGEs within hair is unlikely. Instead, AGE levels in hair may reflect AGE formation in hair matrix cells and the partial uptake of circulating AGEs. ROC analysis, which focused on distinguishing diabetes using AGE levels, identified significant differences between control and diabetic rats. This analysis revealed that CEL and MG-H1 levels in hair exhibited high diagnostic performance, with an AUC of 1.0. Thus, CEL and MG-H1 levels in hair demonstrated greater diagnostic power than serum AGE levels. Serum AGEs such as CEL and MG-H1 did not show any differences between control and DM. This may be due to the metabolic imbalances caused by diabetes being more prominently reflected in hair AGE levels rather than in serum. The observed positive correlation between serum CML and hair CML, as well as the negative correlation between serum MG-H1 and hair MG-H1, suggest that changes in circulating AGE levels after disease onset can be assessed using hair AGE measurements.

The extended section of hair consists of keratinized, non-living cells that retain blood-derived components for short- to long-term durations⁷. Unlike serum sampling, hair sampling can be performed by non-medical personnel, and hair processing in clinical laboratories and hospitals is relatively straightforward. Additionally, while biological samples such as serum must be frozen and disposed of carefully, hair can be stored at room temperature and easily discarded. Serum AGEs are influenced not only by endogenous production but also by dietary intake³³, necessitating a minimum fasting period of 16 h for accurate measurements. In contrast, hair is less likely to reflect short-term dietary influences immediately before sampling, making it a more convenient and stable medium for measurement. However, dietary-derived AGEs, including both free and protein-bound forms, may affect hair AGE levels over the long term, and future studies will be necessary to investigate this.

To establish a method for measuring hair AGE levels, we initially confirmed the required sample amount and the stability of hair AGEs. Previous studies have used several milligrams of hair for AGE measurements²⁷. To assess the latest metabolic condition, we focused on the proximal portion of the hair shaft of the participants. Collecting over 10 hair samples from the first centimeter near the root was challenging; therefore, we determined the minimum feasible sample size to be five strands of hair, from which we calculated the AGE content per strand. The chromatograms of each AGE obtained from the five strands were acceptable when analyzed by MS. However, the CV was above 15%. To enhance precision, it may be necessary to increase the number of hair strands used for measurement. Since most AGEs are stable under acidic conditions^34,35, our study confirmed that the levels of two specific AGEs without CEL remained unchanged after one week of storage at various temperatures, suggesting that other AGE species may also be present in hair. In the quantification of CEL, where the internal standard was added before acid hydrolysis, hair CEL levels decreased after incubation at 40 °C for 7 days. This suggests that the decrease was likely caused by the 7 day incubation at 40 °C rather than the acid hydrolysis itself, although the exact cause of the decrease remains unknown. Furthermore, measuring hair AGEs in Japanese women demonstrated an age-dependent increase, similar to serum AGEs, implying that systemic age-related changes are reflected in the hair. However, how these levels change under metabolic disorders remains unknown and is a subject for future investigation. Consequently, hair sampling and AGE measurement in clinical settings may offer a simpler approach for assessing diseases.

This study focused on samples of black hair. Hair can be classified as pigmented (black) or non-pigmented (gray) hair. Additionally, some individuals dye their hair according to their lifestyle preferences. The potential for assessing diseases using AGEs in hair samples that are not black or have been dyed remains unclear. As hair growth is influenced by hormonal factors³⁶, further investigations incorporating sex differences and other physiological parameters, along with larger sample sizes, are warranted. Despite the numerous challenges in using hair AGE levels for disease assessment, this study provides a foundation for future research.

Although health awareness is increasing, continuous self-monitoring of physiological conditions remains challenging. Hair AGE levels, which can be measured using a minimally invasive and easily accessible sample, may serve as an initial screening tool for the early identification of metabolic disorders such as diabetes, complementing traditional blood-based diagnostic methods. In addition, they may facilitate disease risk assessment and promote timely lifestyle modifications. If multiple hair samples, including some AGE levels, are processed simultaneously using a fully automated solid-phase extraction system (FSPES)³⁷, it quickly provides an indication of the physiological status through hair AGE levels.

This study successfully established a system for quantifying AGE levels in hair, a minimally invasive and readily obtainable biological sample. Furthermore, we confirmed that hair AGE levels vary with diabetes and correlate with aging in humans. These findings suggest the potential use of hair AGE levels in assessing disease risk, including that of diabetes.

Methods

Chemical reagents

We purchased the following chemicals: CML and CEL (Poly Peptide Laboratories, Strasbourg, France); Arg and Lys (Wako, Osaka, Japan); and isotope-labelled internal standards (ISTDs) of [²H₂] CML, [²H₄] CEL, and [²H₃] MG-H1 (PolyPeptide Laboratories France S.A.S., Strasbourg, France), as well as [¹³C₆] lysine and [¹³C₆] arginine (Cambridge Isotope Laboratories, Inc., Tewksbury, MA, USA). Meanwhile, MG-H1 was synthesized as previously described³⁸.

Human hair

In collaboration with the Living Appliances and Solutions Company, Panasonic Corporation, hair samples were obtained from one healthy male in his 50 s and 30 healthy females aged between their 30 s and 60 s (approval number: 20220617-A01). All participants provided their informed consent in writing. The hair samples were stored at − 24 °C. These hair samples had not undergone any coloring or bleaching treatment. The measurement of AGEs in human hair was approved by Tokai University (Approval No.: 22140).

Variation of ages in hair due to storage temperature

Five individual hairs per strand were collected 1 cm from the root from the same male participant and stored in 2 mL tubes at − 80 °C, 4 °C, and 40 °C for 0, 3, and 7 days. The hair samples were then subjected to the same pretreatment as described below for AGEs analysis.

Experiments on animals

All animal experiments were approved by Tokai University (Approval No.: 2020002) and complied with the “Guidelines for the Care and Use of Animals for Scientific Purposes at Tokai University” (established April 1, 2007). Additionally, all animal experiments were in accordance with the ARRIVE guidelines. Wistar rats were purchased from Kyudo Corporation (Kumamoto, Japan). The rats were kept in a pathogen-free, barrier-free facility (12-hour light/dark cycle) and fed a normal rodent diet (Clair, Tokyo, Japan). The rats were randomly divided into two groups (n = 2–3/cage). Diabetes was induced in 8-week-old male rats, each weighing approximately 200 g, through a single intravenous injection of STZ (50 mg/kg body weight) administered via the tail vein in 0.2 mL of 0.05 M saline-citrate buffer (pH 5.0)³⁵. Diabetic (n = 6) and non-diabetic (n = 7) rats (12 weeks after STZ injection) were sacrificed by decapitation under isoflurane anesthesia. Serum samples were collected at 1, 3, 5, 7 and12 weeks, and hair samples were collected for measurement over weeks 1, 3, 7, and 12. The serum samples were analyzed for serum glucose concentration using the Glucose CII-Test Wako (Fujifilm Wako Pure Chemicals, Osaka, Japan) and AGE concentrations using electrospray-ionized quadrupole time-of-flight mass spectometry (ESI-QTOF-MS). Tissue samples were immediately frozen and stored at − 80 °C until analysis.

Preparation of rat hair samples

Rat body hair samples were collected using an electric hair clipper for humans. Rat body hair samples were 1 cm long, and tissue samples were washed with 70% ethanol, immediately frozen, and stored at − 80 °C until analysis.

Hair pretreatment for measurement of AGE levels

Pretreatment of hair for measuring AGEs was based on a previous paper^3,29. A 1 cm segment of human hair, cut as close as possible to the scalp using dissecting scissors, was collected and used as a sample. This is because hair near the scalp is considered to strongly reflect the internal body condition³⁹. Five hair samples were placed in a 2 mL tube. The rat hairs were cut 1 cm from the root and placed in 2 mL tubes, each containing five hairs. Head hair was immersed in a mixture of 50 µL ultrapure water, 50 µL 0.2 M boric acid, and 2 mM 1,1-Diphenyl-2-picrylhydrazyl solution. The samples were then reduced with 5 µL 2 M NaBH₄ (in 0.1 M NaOH) for 4 h.

After the reduction process, five hairs were dissolved by incubation with 1 mL of 6 M HCl at 100 °C for 1 h, followed by a 5-fold dilution with 6 M HCl. Then, 0.01 nmol of [²H₂] CML, [²H₄] CEL, [²H₃] MG-H1, and 5 nmol of [¹³C₆] lysine and [¹³C₆] arginine were added, and hydrolysis was performed at 100 °C for 17 h. After the hydrolysis, the samples were concentrated to dryness using a centrifuge concentrator. Most AGEs are stable under strong acidic conditions; however, some compounds, such as MG-H1, have lower acid tolerance and are somewhat unstable. Despite this, they can still be measured without complete degradation. Therefore, to maintain precise quantification, we performed calibration using internal standards³⁴. Ultrapure water was added to the dried samples, and the mixture was resuspended in the water. The Strata-X-C column (Phenomenex, Torrance, CA, USA) was attached to the vacuum manifold, washed with 1 mL of MeOH, and equilibrated with 1 mL of ultrapure water and the sample solvent. The column was then washed with 2 mL of 2% formic acid (FA) and eluted with 2 mL of 7% ammonia; then, elution samples were collected. The collected elution samples were concentrated to dryness at 40 °C using a blow-tube concentrator. The dried samples were suspended in 0.2 mL 20% acetonitrile (MeCN) containing 0.1% FA. All samples were then filtered using a 0.45 μm filter and completely transferred into a glass insert for analysis using ESI-QTOF-MS⁴⁰.

ESI-QTOF-MS

A Bruker Daltonics spectrometer (Bruker, Bremen, Germany) was used for the measurements. The instrument was externally calibrated for each evaluation using 50% 2-propanol containing 5 mM sodium formate. The dry temperature was set to 200 °C, and the capillary voltage was 4.5 kV. The nebulizer pressure was set at 1.6 bar. The mass range was from 50 to 1000 m/z. LC was performed on a ZIC^®-HILIC column (2.1 × 150 mm, 5 μm; Merck Millipore, Billerica, MA, USA) maintained at 40 °C. The mobile phase consisted of 0.1% FA with a two-step gradient of MeCN (0–2 min, 90% MeCN; 2–16 min, 90–10% MeCN; 16–19 min, 10% MeCN). The flow rate was set at 0.2 mL/min, and the injection volume was 5 µL. The AGEs (CML, CEL, and MG-H1), lysine, arginine, and their internal standards were detected by ESI-positive MS. The retention times for these AGEs and amino acids were approximately 13–15 min. Those ISTDs were analyzed using amino acid analysis. We detected lysine m/z 147.1128 ± 0.01, [¹³C₆] lysine m/z 153.1329 ± 0.01, arginine m/z 175.1190 ± 0.01, [¹³C₆] arginine m/z 181.1391 ± 0.01, CML m/z 205.1183 ± 0.01, [²H₂] CML m/z 207.1308 ± 0.01, CEL m/z 219.1339 ± 0.01, [²H₄] CEL m/z 223.1590 ± 0.01, MG-H1 m/z 229.1295 ± 0.01, and [²H₃] MG-H1 m/z 232.1483 ± 0.01 in ESI-QTOF-MS.

Statistical analysis

All data were analyzed using R-4.2.1 for the Windows operating system. A significance threshold of P < 0.05 was applied to all data. Biological parameters and AGE levels were compared for each variable using the Wilcoxon rank-sum test. Spearman’s correlation was used to analyze the correlations between human or rat hair samples, rat serum samples, and AGE levels. AGE z-scores were calculated by averaging and standardizing the z-scores of CML and CEL³. ROC analyses were performed using the MetaboAnalyst software⁴¹.

Acknowledgements

The authors thank Shimadzu Corporation for their technical assistance. The authors thank Research Institute of Agriculture from Tokai University for assistance of analysis.

Author contributions

S.K. and R.N. designed the experiments. A.O. contributed to the clinical sample collection. R.N. acquired funding. Y.S. contributed to the study methodology and investigation. S.K. contributed to the investigation, development of the methodology, and data analysis. S.K. wrote the manuscript. R.N. reviewed the manuscript. All authors have read and approved the final version of the manuscript.

Data availability

The datasets used in this study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethical standards

This study adhered to the Declaration of Helsinki (amended at the 2013 WMA Fortaleza Meeting) and followed the Ethical Guidelines for Medical Research Involving Human Subjects (notified by the Ministry of Education, Culture, Sports, Science, and Technology and the Ministry of Health, Labor, and Welfare). This study was conducted with the deliberation and approval of the Ethics Review Committee of Living Appliances and Solutions Company, Panasonic Corporation (Approval No.: 20220617-A01) for the project titled “AGEs research for the development of a new value-adding facial device.”

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. The international diabetes federation (IDF). IDF diabetes atlas 2021, (2021). Available from: https://diabetesatlas.org/atlas/tenth-edition/.

2. Rigon, FA et al. Flash glucose monitoring system in special situations. Arch. Endocrinol. Metab.; 2022; 66, pp. 883-894. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35657123][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10118756](DOI 10.20945/2359-3997000000479) (PubMed: 35657123)

3. Kato, S; Matsumura, T; Sugawa, H; Nagai, R. Correlation between serum advanced glycation end-products and vascular complications in patient with type 2 diabetes. Sci. Rep.; 2024; 14, 18722.1:CAS:528:DC%2BB2cXhvVens7jO [DOI: https://dx.doi.org/10.1038/s41598-024-69822-5)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39134632][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11319737]

4. Lee, J; Yun, JS; Ko, SH. Advanced glycation end products and their effect on vascular complications in type 2 diabetes mellitus. Nutrients; 2022; 14, 3086.1:CAS:528:DC%2BB38XitFGntLnL [DOI: https://dx.doi.org/10.3390/nu14153086)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35956261][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9370094]

5. Beisswenger, PJ et al. Early progression of diabetic nephropathy correlates with methylglyoxal-derived advanced glycation end products. Diabetes Care; 2013; 36, pp. 3234-3239.1:CAS:528:DC%2BC3sXhs1yltL7I [DOI: https://dx.doi.org/10.2337/dc12-2689)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23780945][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3781566]

6. Goto, M et al. Factors associated with untreated diabetes: analysis of data from 20,496 participants in the Japanese National health and nutrition survey. PLOS One; 2015; 10, e0118749. [DOI: https://dx.doi.org/10.1371/journal.pone.0118749)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/25756183][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4355906]

7. Jang, WJ et al. Hair metabolomics in animal studies and clinical settings. Molecules; 2019; 24, 2195.1:CAS:528:DC%2BC1MXhvVWqsrzE [DOI: https://dx.doi.org/10.3390/molecules24122195)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31212725][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6630908]

8. Bronowicka-Szydełko, A et al. Effect of advanced glycation end-products in a wide range of medical problems including COVID-19. Adv. Med. Sci.; 2024; 69, pp. 36-50. [DOI: https://dx.doi.org/10.1016/j.advms.2024.01.003)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38335908]

9. Nagai, R et al. Peroxynitrite induces formation of N-carboxymethyl lysine by the cleavage of Amadori product and generation of glucosone and Glyoxal from glucose. Diabetes; 2002; 51, pp. 2833-2839.1:CAS:528:DC%2BD38Xmsleru7s%3D [DOI: https://dx.doi.org/10.2337/diabetes.51.9.2833)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/12196478]

10. Knels, L et al. Effects of advanced glycation end products-inductor Glyoxal and hydrogen peroxide as oxidative stress factors on rat retinal organ cultures and neuroprotection by UK-14,304. J. Neurochem; 2008; 106, pp. 1876-1887.1:CAS:528:DC%2BD1cXhtVegtLjL [DOI: https://dx.doi.org/10.1111/j.1471-4159.2008.05540.x)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/18624919]

11. Nagai, R; Nagai, M; Shimasaki, S; Baynes, JW; Fujiwara, Y. Citric acid inhibits development of cataracts, proteinuria and ketosis in streptozotocin (type 1) diabetic rats. Biochem. Biophys. Res. Commun.; 2010; 393, pp. 118-122.1:CAS:528:DC%2BC3cXis1aktbc%3D [DOI: https://dx.doi.org/10.1016/j.bbrc.2010.01.095)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20117096][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2917331]

12. Gonçalves, CA et al. Glycolysis-derived compounds from astrocytes that modulate synaptic communication. Front. Neurosci.; 2018; 12, 1035. [DOI: https://dx.doi.org/10.3389/fnins.2018.01035)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30728759]

13. Diallo, AM et al. Association between the tissue and Circulating advanced glycation end-products and the micro- and macrovascular complications in type 1 diabetes: the DIABAGE study. Diabetes Ther.; 2022; 13, pp. 1531-1546.1:CAS:528:DC%2BB38XitVGgsr%2FN [DOI: https://dx.doi.org/10.1007/s13300-022-01285-1)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35779209][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9309113]

14. Rabbani, N et al. Analysis of serum advanced glycation endproducts reveals methylglyoxal-derived advanced glycation MG-H1 free adduct is a risk marker in non-diabetic and diabetic chronic kidney disease. Int. J. Mol. Sci.; 2022; 24, 152. [DOI: https://dx.doi.org/10.3390/ijms24010152)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36613596][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9820473]

15. Papachristou, S et al. Skin advanced glycation end products as a screening tool of neuropathy in type 2 diabetes mellitus. J. Diabetes Complications; 2022; 36, 108356.1:CAS:528:DC%2BB38XivFGgsLzL [DOI: https://dx.doi.org/10.1016/j.jdiacomp.2022.108356)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36395605]

16. Oimomi, M et al. Glycation of hair protein in the assessment of long-term control of blood glucose. Jpn J. Med.; 1988; 27, pp. 277-280.1:STN:280:DyaL1M%2Fls1WjtQ%3D%3D [DOI: https://dx.doi.org/10.2169/internalmedicine1962.27.277)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/3143025]

17. Nissimov, J et al. Method for chronological recording of antigen appearance in human head-hair shafts and its use for monitoring glycation products in diabetes. J. Immunol. Methods; 2007; 320, pp. 1-17.1:CAS:528:DC%2BD2sXislKhtL4%3D [DOI: https://dx.doi.org/10.1016/j.jim.2006.10.014)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17210162]

18. Paus, R; Cotsarelis, G. The biology of hair follicles. The biology of hair follicles; 1999; 341, pp. 491-497.1:STN:280:DyaK1MzltlGjtA%3D%3D [DOI: https://dx.doi.org/10.1056/NEJM199908123410706]

19. Ibrahim, L; Wright, EA. The growth of rats and mice vibrissae under normal and some abnormal conditions. J. Embryol. Exp. Morphol.; 1975; 33, pp. 831-844.1:STN:280:DyaE28%2Fis1emsw%3D%3D [DOI: https://dx.doi.org/10.1242/dev.33.4.831)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/1176877]

20. Iob, E; Steptoe, A. Cardiovascular disease and hair cortisol: a novel biomarker of chronic stress. Curr. Cardiol. Rep.; 2019; 21, 116. [DOI: https://dx.doi.org/10.1007/s11886-019-1208-7)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31471749][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6717172]

21. Cuypers, E; Flanagan, RJ. The interpretation of hair analysis for drugs and drug metabolites. Clin. Toxicol. (Phila); 2018; 56, pp. 90-100.1:CAS:528:DC%2BC2sXhsFCqtLrF [DOI: https://dx.doi.org/10.1080/15563650.2017.1379603)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28938866]

22. Toyo’oka, T., LC-ESI- & MS/MS. Biol. Pharm. Bull. 39, 1397–1411. Diagnostic approach to disease using non-invasive samples based on derivatization and (2016). https://doi.org/10.1248/bpb.b16-00453) (PubMed: 27582321).

23. Hodes, A et al. Hair cortisol in the evaluation of Cushing syndrome. Endocrine; 2017; 56, pp. 164-174.1:CAS:528:DC%2BC2sXislWisbc%3D [DOI: https://dx.doi.org/10.1007/s12020-017-1231-7)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28194652][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5437744]

24. Verzijl, N et al. Age-related accumulation of Maillard reaction products in human articular cartilage collagen. Biochem. J.; 2000; 350, pp. 381-387.1:CAS:528:DC%2BD3cXmvVGksLY%3D [DOI: https://dx.doi.org/10.1042/bj3500381)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10947951][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1221264]

25. Muir, R., Forbes, S., Birch, D. J. S., Vyshemirsky, V. & Rolinski, O. J. Keratin intrinsic fluorescence as a mechanism for non-invasive monitoring of its glycation. Methods Appl. Fluoresc. 11 10.1088/2050–6120/aca507 (2022). (PubMed: 36541527).

26. Paisey, RB et al. Glycosylation of hair: possible measure of chronic hyperglycaemia. Br. Med. J. (Clin Res. Ed); 1984; 288, pp. 669-671.1:STN:280:DyaL2c7ivFegsw%3D%3D [DOI: https://dx.doi.org/10.1136/bmj.288.6418.669)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/6421426]

27. Takemura, A. et al. Determination of furosine in hair by liquid chromatography-tandem mass spectrometry. Biomed. Chromatogr. 11, 61–62 (DOI 10.1002/(SICI)1099 – 0801(199701)11:1 < 61::AID-BMC649 > 3.0.CO;2-R) (PubMed: 9051221) (1997).

28. Nagai, R et al. Usefulness of antibodies for evaluating the biological significance of ages. Ann. N Y Acad. Sci.; 2008; 1126, pp. 38-41.2008NYASA1126..38N1:CAS:528:DC%2BD1cXmtFygtbw%3D [DOI: https://dx.doi.org/10.1196/annals.1433.001)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/18079488]

29. Yamanaka, M et al. Soft-shelled turtle eggs inhibit the formation of ages in the serum and skin of diabetic rats. J. Clin. Biochem. Nutr.; 2016; 58, pp. 130-134.1:CAS:528:DC%2BC2sXpvFemug%3D%3D [DOI: https://dx.doi.org/10.3164/jcbn.15] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/27013779][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4788400]

30. Berends, E., van Oostenbrugge, R. J., Foulquier, S. & Schalkwijk, C. G. Methylglyoxal, a highly reactive Dicarbonyl compound, as a threat for blood brain barrier integrity. Fluids Barriers CNS. 20https://doi.org/10.1186/s12987-023-00477-6) (2023). (PubMed: 37875994).

31. Meier, U; Briellmann, T; Scheurer, E; Dussy, F. Distribution pattern of Ethyl glucuronide and caffeine concentrations over the scalp of a single person in a forensic context. Drug Test. Anal.; 2017; 9, pp. 1594-1603.1:CAS:528:DC%2BC2sXnsV2qurs%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28296341](DOI 10.1002/dta.2186) (PubMed: 28296341)

32. Wortmann, FJ; Stapels, M; Elliott, R; Chandra, L. The effect of water on the glass transition of human hair. Biopolymers; 2006; 81, pp. 371-375.1:CAS:528:DC%2BD28XjtVant7c%3D [DOI: https://dx.doi.org/10.1002/bip.20429)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16358248]

33. Koschinsky, T. et al. Orally absorbed reactive glycation products (glycotoxins): an environmental risk factor in diabetic nephropathy. Proc. Natl Acad. Sci. U. S. A. 94, 6474–6479 (1997). https://doi.org/10.1073/pnas.94.12.6474) (PubMed: 9177242).

34. Scheijen, J. J. L. & J. M. The analysis of advanced glycation endproducts: a mass spectrometry–based approach & its applications, Doctoral thesis, Maastricht University. Datawyse/Universitaire Pers Maastricht (https://doi.org/10.26481/dis.20170621js)

35. Ohno, RI et al. Glucoselysine is derived from Fructose and accumulates in the eye lens of diabetic rats. J. Biol. Chem.; 2019; 294, pp. 17326-17338.1:CAS:528:DC%2BC1MXitlKlurfM [DOI: https://dx.doi.org/10.1074/jbc.RA119.010744)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31594865][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6873178]

36. Grymowicz, M. et al. Hormonal effects on hair follicles. Int. J. Mol. Sci.21, 5342. https://doi.org/10.3390/ijms21155342 (2020). (PubMed: 32731328).

37. Kato, S. et al. Rapid pretreatment for multi-sample analysis of advanced glycation end products and their role in nephropathy. J. Clin. Biochem. Nutr.70, 256–261. 10.3164/jcbn.21–175 (2022). (PubMed: 35692677).

38. Yamaguchi, H; Nagai, M; Sugawa, H; Yasuda, H; Nagai, R. Development of a conventional immunochemical detection system for determination of N^δ-(5-hydro-5-methyl-4-imidazolone-2-yl)-ornithine in methylglyoxal-modified proteins. Glycoconj. J.; 2021; 38, pp. 293-301.1:CAS:528:DC%2BB3cXisVeqtrrE [DOI: https://dx.doi.org/10.1007/s10719-020-09957-5)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33241449]

39. Pragst, F; Balikova, MA. State of the Art in hair analysis for detection of drug and alcohol abuse. Clin. Chim. Acta; 2006; 370, pp. 17-49.1:CAS:528:DC%2BD28XntlWns7w%3D [DOI: https://dx.doi.org/10.1016/j.cca.2006.02.019)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16624267]

40. Arakawa, S. et al. Mass spectrometric quantitation of ages and enzymatic crosslinks in human cancellous bone. Sci. Rep.10https://doi.org/10.1038/s41598-020-75923-8) (2020). (PubMed: 33139851).

41. Kusama, K et al. Characterization of serum metabolome and proteome profiles identifies SNX5 specific for pregnancy failure in Holstein heifers. Life (Basel); 2022; 12, 309.1:CAS:528:DC%2BB38XosVKqtLs%3D [DOI: https://dx.doi.org/10.3390/life12020309)] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35207596]