Insulin resistance (IR) is a pathophysiological condition in which organs—mostly skeletal muscle, adipose tissue, and liver—do not respond at an adequate rate to insulin, and it is considered to be a consequence of the disruption of different molecular pathways affected by insulin in these tissues [1]. In the general population, sensitivity to insulin-mediated glucose disposal in several tissues varies greatly [2]. The major consequence of IR, type 2 diabetes, arises when people who are insulin-resistant are unable to maintain the level of hyperinsulinemia required to correct the insulin action deficiency. Clinically, it appears as a defect in insulin-mediated glucose control in tissues, prominently in the above named muscle, fat and liver. Primary characteristics of IR are inhibited lipolysis in adipose tissue, impaired glucose uptake by muscle and inhibited gluconeogenesis in liver [3]. Therefore, IR also encompasses defects in lipid metabolism, in line with the multifaceted roles of insulin in metabolism regulation [4]. IR is one of the earliest manifestations of a constellation of human pathologic conditions that include metabolic syndrome, type 2 diabetes, cardiovascular diseases and aging [5]. Lifestyle modifications, including reduced intake of ultraprocessed foods containing advanced glycation (AGEs) and lipo-oxidation end products (ALEs), body weight loss and increased physical activity, have been shown to increase insulin sensitivity, thereby preventing IR ^[6][7][6,7].

Whether total caloric intake with body fat accretion or the presence of specific nutrients or diet-derived insulin-signaling disruptors is mostly responsible for the IR of aging is unclear [8]. Of the many dietary factors that may play a role in the aging-related lack of or decreased insulin sensitivity, AGEs appear potentially important. Recent clinical data suggest that food-derived AGEs may contribute to IR [9].

2. Evidence of an Association between Dietary AGEs and IR

2.1. Animal Data

A role of dietary AGEs as a causative agent in IR has been well documented in several studies in different mouse strains by independent teams. RWesearchers review some of these studies here and in Table 1. Reduced AGE intake leads to lower levels of circulating AGEs and to improved insulin sensitivity in the db/db mouse IR model ^[10][50]. To demonstrate this, db/db mice were randomly placed for 20 weeks (more than 50% of their usual life span) on a diet with either low AGE content (LAGE) or a 3.4-fold higher content of AGEs (HAGE). LAGE mice showed lower fasting plasma insulin levels and body weight compared with HAGE mice, despite equal caloric intake. LAGE mice had improved responses to both glucose (at 40 min, p = 0.003) and insulin (at 60 min, p = 0.007) tolerance tests, which correlated with a doubling of glucose uptake by adipose tissue. LAGE mice had twofold lower serum carboxymethyllysine (CML) and methylglyoxal (MG) concentrations and a better-preserved structure of pancreatic islets compared with HAGE mice ^[10][50]. Thus, the effect of dietary AGEs affects multiple tissues (liver, adipose tissue, pancreas), leading overall to impaired metabolism. To overcome a potential effect of the db/db genotype itself, another group focused on euglycemic mice. Nontransgenic C57/BL6 mice were randomly assigned to high-fat diets (35% g fat) to induce IR, with either high (HAGE-HF group; 995.4 units/mg AGE) or low (by 2.4-fold LAGE-HF group; 329.6 units/mg AGE) dietary AGE content for 6 months (approximately 20% of the usual life span) ^[11][51]. At the end of 6 months, 75% of the HAGE-HF mice had become diabetic, while none of the LAGE-HF mice had, despite a similar rise in body weight and plasma lipids. Moreover, the HAGE-HF group showed markedly impaired glucose and insulin responses during glucose tolerance tests and euglycemic and hyperglycemic clamps and abnormal pancreatic islet structure and function compared with those of LAGE-HF mice. These findings demonstrate that the development of IR and type 2 diabetes during prolonged high-fat feeding in mice are linked to the excess AGEs/ALEs in fatty diets ^[11][51]. In addition to type 2 diabetes, IR has been associated with cognitive dysfunctions, such as Alzheimer’s disease ^[15][54]. To determine whether dietary AGEs promote aging-related cognitive decline, mice were subjected to different levels of AGEs in their diets for 18 months (i.e., 60% of their life span) ^[12][52]. Those mice in the high-AGE diet developed metabolic syndrome (with IR), increased brain amyloid-β42, intracerebral deposits of AGEs, gliosis, and cognitive deficits, accompanied by suppressed expressions of SIRT1, nicotinamide phosphoribosyltransferase, AGE receptor 1, and PPARγ. These changes were not due to aging or caloric intake, as none of them were present in age-matched, pair-fed low-AGE mice. The animal data were strengthened by the demonstration of significant temporal correlations between high circulating AGEs, impaired cognition, and insulin sensitivity in elderly subjects ^[16][55]. Clinically, it has been postulated that IR could explain part of the age-related decrease in cognitive function ^[17][18][56,57]. Considering the importance of the microbiota in IR ^[19][20][58,59] (see above), yet another study showed that a high-AGE diet induced IR and altered the gut microbiota composition and structure, reducing its diversity in mice ^[13][28]. The authors postulated that the loss of butyrate-producing bacteria in the AGE-loaded animals might have impaired the colonic epithelial barrier, thereby triggering chronic low-grade inflammation and possibly IR ^[13][28]. An involvement of AGEs in general, not necessarily dietary AGEs, in the development of IR mediated by alteration of sphingolipid metabolism was demonstrated in two different models of IR in mice, one genetically diabetic and the other diet-induced IR (fed a 60% trans-fat diet). Supplementation of a group of mice with pyridoxamine that lowered AGE levels reduced the development of IR ^[14][53]. All in all, these independent studies highlight the potential role of dietary AGEs as modifiable agents in the development of IR in mice, acting on several factors (peripheral insulin function, microbiota, cellular stress responses).

2.2. Epidemiological Evidence Linking Dietary AGEs and IR in Humans

In a large cross-sectional study conducted in young healthy Slovakian individuals of both sexes (n = 2769) IR, assessed through the Quantitative Insulin Sensitivity Check Index (QUICKI), was associated with serum and urinary levels of some α-dicarbonyls (AGE precursors, such as methylglyoxal) and AGEs, independently of cardiometabolic risk markers and sex ^[21][60]. In an American cohort, an increased association of very high dietary AGE intake (defined as the top quartile) and metabolic syndrome was described in adolescents aged 12–19 years from NHANES (years 2003–2004 and 2005–2006) ^[22][61]. The latter study also demonstrated that very high dietary AGE intake was significantly associated with three of five criteria for metabolic syndrome: waist circumference, serum triglyceride, and HDL cholesterol levels ^[22][61]. A meta-analysis of 17 randomized controlled trials comprising 560 participants also demonstrated that IR, measured by HOMA-IR, was significantly reduced in a low-AGE compared to a high-AGE diet, although there was no significant difference in fasting insulin, 2 h insulin and insulin area under the curve results between both diets after an oral glucose tolerance test ^[23][62]. This apparent contradiction stresses the need for adequate standardization of the methods to define IR, as is discussed in detail later in this chapter ^[24][63].

2.3. Randomized Controlled Interventional Studies Testing the Association between Dietary AGEs and IR in Humans

Table 2 describes eight independent clinical trials ^{[25][26][27][28][29][30][31][32]}[64,65,66,67,68,69,70,71] that have looked at the effects of an AGE-restricted diet on IR markers, comprising a total of more than 440 participants. Six of the studies demonstrated an association between decreased dietary AGE intake and improved insulin sensitivity. One study, in healthy subjects, indicated that a low-AGE diet led to decreased serum levels of AGEs in parallel with HOMA-IR ^[25][64]. Five other studies looked at the effect of a low-AGE diet on IR in overweight and/or metabolic syndrome subjects. In four of these studies, the low-AGE diet was associated with lower serum or urinary levels of AGEs, as well as parallel decreases in IR as assessed by HOMA-IR in three of them ^[26][28][29][65,67,68] or improved glucose uptake assessed by euglycemic clamp in the fourth study ^[27][66]. One of the studies was a randomized 6-week prospective intervention in type 2 diabetes subjects with a standard diet (n = 13) versus low-AGE diet (n = 13), which showed a significant decrease in TNF-α and malondialdehyde levels in the low-AGE diet group, but without a significant change in HOMA-IR or serum AGEs ^[32][71]. Since circulating AGE levels appear to be useful surrogates of dietary AGE exposure, their lack of change during theis study raises the possibility that the dietary intervention might not have been effective. A second study also failed to show a relationship between dietary AGE restriction and changes in insulin sensitivity ^[30][69]. However, several key facts, mainly related to differences in the composition and methods to measure AGEs of the experimental diets, could help clarify the inconclusive outcome of the study in comparison with other studies. For example, mass-spectrometry-based AGE measurement may differ profoundly in the content of several bio-accessible AGE products assessed by immunological measurements such as ELISA. Moreover, in those studies showing a clear effect of dietary AGEs in IR, dietary AGEs were delivered mostly by ingestion of meats and animal products, highly processed and cooked at high temperatures, while in this negative study ^[30][69], most dietary AGEs came predominantly from cereals and were based on dietary frequency in a specific Dutch population ^[33][72]. Processing of samples before AGE measurement, particularly the inclusion or not of delipidation, may also be relevant, as AGE content in dietary lipids is important ^[34][73]. All these points emphasize that standardization is a key factor in designing dietary interventions. Of interest, individual factors, such as ethnicity and age, may also be relevant in interpreting insulin sensitivity tests ^[35][36][74,75]. Genotypic characteristics have been shown to influence interaction with dietary AGEs (for example, FADS2 (definer) polymorphisms) ^[37][76]. Lastly, intervention length is key in interpretating data employing methods with a high interindividual variation, such as the hyperglycemic–euglycemic clamp, which in the case of the cited negative study showed a 50% standard deviation over the mean values ^[30][69]. In fact, a prior randomized controlled trial employing isotope-based euglycemic clamps showed that insulin sensitivity changed only after a six-month intensive weight-loss and exercise program [6].

2.4. Clinical Trials with Mediterranean and Vegan Dietary Patterns Support the Association between Dietary AGEs and IR in Humans

The above studies (Table 2) show that—generally, although not universally—reducing dietary AGE intake can diminish IR markers. One of the limitations of these studies is that they are of short duration. A longer study, CORDIOPREV, indirectly evaluated the potential impact of dietary AGEs in IR ^[38][77]. Reduction in serum AGE levels in subjects following a Mediterranean diet as part of the CORDIOPREV study, lasting for 5 years, was shown to increase significantly the probability of type 2 diabetes remission, a hard measure of IR. All participants had previous cardiovascular events and type 2 diabetes when recruited ^[38][77]. In a cohort of overweight subjects (n = 244) randomly assigned to an intervention with a low-fat plant-based diet (n = 122) or a control diet (n = 122) for 16 weeks, dietary AGE consumption decreased on a low-fat plant-based diet, and this was associated with changes in body weight, body composition, and insulin sensitivity, independently of energy intake ^[39][78]. Therefore, the epidemiological data and most of the interventional studies in humans support a long-term reduction in dietary AGEs being associated with an improvement in clinically relevant outcomes (insulin sensitivity, weight loss, probability of type 2 diabetes remission) requiring minimal time for revealing its beneficial effects.

Author, Year, Reference	Study Design	Intervention	Number of Participants	Randomized	Participant Characteristics	Duration and Allocation	Specified Outcomes	Findings
Birlouez, 2010 [64]	Crossover	High- and low-AGE diets	62	Yes	Healthy individuals	1 month, France	Changes in serum AGEs and HOMA	Decreased serum AGEs and HOMA
Uribarri, 2011 [70]	Parallel	High- and low-AGE diets	18	Yes	Patients with diabetes	3 months, USA	Changes in serum AGEs, markers of OS and inflammation and HOMA	Decreased levels of serum AGEs, markers of OS *, inflammation and HOMA
Luevano-Contreras 2013 [71]	Parallel	High- and low-AGE diets	26	Yes	Patients with diabetes	1.5 months, Mexico	Changes in serum AGEs, markers of OS and inflammation and HOMA	Decreased markers of OS and inflammation, but no changes in serum AGEs or HOMA
Mark, 2014 [65]	Parallel	High- and low-AGE diets	74	Yes	Overweight women	1 month, Denmark	Changes in urinary AGEs and HOMA	Decreased urinary AGEs and HOMA
De courten 2016 [66]	Crossover	High- and low-AGE diets	20	Yes	Overweight individuals	0.5 months, Denmark	Changes in serum AGEs and insulin resistance (hyperinsulinemic–euglycemic clamp and intravenous glucose tolerance test)	Decreased serum AGEs and insulin resistance
Vlassara, 2016 [67]	Parallel	High- and low-AGE diets	138	Yes	Patients with metabolic syndrome	12 months, USA	Changes in serum AGEs, markers of OS and inflammation and HOMA	Decreased levels of serum AGEs, markers of OS, inflammation and HOMA
Goudarzi 2020 [68]	Parallel	High- and low-AGE diets	40	Yes	Patients with metabolic syndrome	2 months, Iran	Changes in serum AGEs, markers of OS and inflammation and HOMA	Decreased serum AGEs, markers of OS, inflammation, HOMA and weight (were also calorie restricted)
Linkens 2022 [69]	Parallel	High- and low-AGE diets	82	Yes	Patients with obesity	1.5 months, Netherlands	Changes in serum AGEs, markers of OS and inflammation and insulin resistance (hyperinsulinemic–euglycemic and hyperglycemic clamp)	Decreased circulating AGEs but not markers of OS/inflammation or insulin sensitivity