Hidden Complexities in the Measurement of Fructosyl-Lysine and Advanced Glycation End Products for Risk Prediction of Vascular Complications of Diabetes

Diabetes is associated with increased glycation of proteins by glucose and methylglyoxal (MG). Early-stage glycation by glucose forms fructosamine adducts, mainly N_ε-fructosyl-lysine (FL), which can be exploited for assessment of glycemic control in the measurement of A1C (FL and N_α-fructosyl-valine). Later- or advanced-stage glycation processes involving degradation of FL and glycation of proteins by MG and related physiological dicarbonyl metabolites form advanced glycation end products (AGEs) (1). AGEs contribute to the pathogenesis of vascular complications of diabetes, and their measurement may provide improved complications-risk prediction.

Stable isotopic dilution analysis liquid chromatography-tandem mass spectrometry has markedly improved the measurement of AGEs, providing robust quantitation of multiple AGEs concurrently in tissues and body fluids (2). Examples of AGEs formed from fructosamine degradation are N_ε-carboxymethyl-lysine (CML) and crosslink glucosepane, and examples of AGEs formed from MG are hydroimidazolone MG-H1 and N_ε-(carboxymethyl)lysine (CEL). Liquid chromatography-tandem mass spectrometry also revealed two major forms of each AGE: AGE moieties of proteins, called AGE residues by conventional biochemical nomenclature (also known by the trivial nomenclature of “protein-bound AGEs”), and AGE free adducts (glycated amino acids) formed mainly by the proteolysis of AGE-modified proteins. Protein AGE residues have low renal clearance and biodistribution limited to that of the protein on which they are formed, whereas free adducts have high renal clearance and membrane permeability via amino acid and other transporters and are the major form in which AGEs are cleared from the body by urinary excretion (2). AGE assays that do not distinguish between protein AGE residues and AGE free adducts are difficult to interpret as they relate to unknown proportions and quantities of both—as applies to an often-used immunoassay for CML (3). Early assessments of AGEs were often limited to AGEs with intense intrinsic fluorescence, such as pentosidine (1). Pentosidine is produced mainly from pentose sugar precursors and is linked to pentose phosphate pathway activity (4). It is a covalent crosslink, but it has very low protein content compared with other crosslinks, such as glucosepane and nonglycation crosslinks of o,o’-dityrosine, N_ε-(γ-glutamyl)lysine and others, suggesting it has limited functional or pathogenic effect in this role (2). Measurement of FL, MG-derived AGE, and pentosidine may provide biomarkers of glycemic control, MG metabolism, and pentose phosphate metabolism, respectively. As all are linked mechanistically to the development of vascular complications (5–7), this may provide a diagnostically powerful biomarker combination in diabetes. In this issue, this is addressed in measurements of plasma protein and skin collagen in the studies by Hanssen et al. (8) and Genuth et al. (9) (Fig. 1A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Measurements of AGEs and association with the development of vascular complications of diabetes. A: Glycation adduct residues of plasma protein and skin collagen measured in studies of risk prediction models for vascular complications of diabetes. B: Biodistribution of FL and AGE residues in tissue and body fluids. Black arrows indicate flows of glycation adducts. Yellow block arrows show movement of plasma protein out of and into the plasma compartment by TER and tissue secretion, respectively, and glomerular filtration of FL- and AGE-modified albumin and FL and AGE free adducts with major recovery of albumin into the renal venous circulation by the albumin retrieval pathway and minor degradation by PTC proteolysis. PTCs, proximal tubular cells; RBCs, red blood cells.

Hanssen et al. (8) measured CML, CEL, and pentosidine residues of plasma protein of patients with type 2 diabetes (T2DM) and also computed an AGE residue content z-score. These were not correlated to A1C. In patients with type 1 diabetes (T1DM), however, we found plasma protein CML and pentosidine residue contents correlated positively with A1C and fasting plasma glucose, respectively (10). The link to glycemic control may be difficult to detect in a less metabolically homogenous study group of patients with T2DM. A higher AGE score was linked to low estimated glomerular filtration rate, low BMI, and low risk of peripheral artery disease (PAD). The low estimated glomerular filtration rate association is likely caused by decreased loss of FL- and AGE-modified albumin through the glomerular filter, and low BMI and PAD associations likely reflect increased transcapillary escape rate (TER) of albumin in obesity and PAD (11,12) (Fig. 1B). ACE inhibitors/angiotensin receptor blockers and lipid-lowering agents may similarly affect plasma protein AGE levels through effects on glomerular tightening and TER, respectively (13,14). These effects on plasma protein FL and AGEs have hitherto been given little consideration, except for two previous studies (13,15). In future work, awareness of these factors would likely improve risk predictor analysis. The positive association of plasma protein CEL with cardiovascular events is interesting, particularly with regard that expression of glyoxalase 1 (Glo1), which metabolizes MG (6), is a major driver for coronary artery disease (16).

Genuth et al. (9) measured furosine (an analytical surrogate of FL) and 6 AGE residues, fluorescence, acid solubility, and pepsin digestibility of skin collagen samples collected at closeout of the Diabetes Control and Complications Trial (DCCT), expanding the analyte panel studied previously. The 10 marker panel improved risk prediction of progression risk of retinopathy and neuropathy but not nephropathy, where glucosepane and MG-H1 emerged as principal new risk predictors. Increased furosine was associated with worsening of retinopathy, nephropathy, and neuropathy.

In the studies discussed, furosine and AGE residues were not analyzed in proteins at sites of complications. This disconnect may be sustained for the desired diagnostic application if there is a proportionate relationship between the change of glycating agent at the sites of complications development and plasma and skin samples analyzed. While associations of skin collagen furosine and AGEs with the development of complications remain after correction for A1C, there is a strong association of furosine with A1C (17). The FL and AGE residue content of proteins is related to the following factors: 1) concentration of glycating agent (glucose or MG, for example), 2) duration of exposure to the glycating agent, 3) intrinsic reactivity of the proteins toward glycation (as reflected in the rate constant k_{Protein, glycating agent}), 4) chemical stability of the glycation adduct, and 5) residence time in the compartment of sampling. Furosine content may have improved risk predictive characteristics over A1C although both report on the same glucose glycation process because it provides a summation of glucose exposure over a greater time period than A1C due to the longer residence time of skin collagen than hemoglobin. Similarly, AGE residue contents of skin collagen, such as glucosepane and CML, may provide improved reporting on longer-term glucose exposure and “memory” of prior glucose exposures than A1C because of their higher chemical stability and residence time—the AGE content remaining even when glycemic control is improved (18).

Given the strengthening associations of increased exposure to MG with risk of microvascular and macrovascular complications (6,16), there is an unmet diagnostic need for the clinical equivalent of A1C for increased exposure to MG. MG-derived AGEs in hemoglobin are candidate biomarkers (10). The disconnect of increased Glo1 activity in red blood cells in diabetes (19) from decreased Glo1 activity at sites of complications (20), however, suggests that hemoglobin may not be the best place to look for MG exposure biomarkers in risk prediction of diabetes complications.

• © 2015 by the American Diabetes Association. Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered.