Perspective: The Paradox in Dietary Advanced Glycation End Products Research—The Source of the Serum and Urinary Advanced Glycation End Products Is the Intestines, Not the Food

Abstract

Inconsistent research results have impeded our understanding of the degree to which dietary advanced glycation end products (dAGEs) contribute to chronic disease. Early research suggested that Western-style fast foods, including grilled and broiled meats and French fries, contain high levels of proinflammatory advanced glycation end products (AGEs). However, recent studies with state-of-the-art ultraperformance LC-tandem mass spectrometry (UPLC-MS) found that there is no evidence that these foods have elevated levels of dAGEs relative to other foods. Paradoxically, observational research found that the intake of fruits (mainly apples), fruit juices (apple juice), vegetables, nuts, seeds, soy, and nonfat milk, which are foods synonymous with healthy eating, as well as the intake of cold breakfast cereals, whole grains (breads), and sweets, which are sources of high-fructose corn syrup (HFCS), were associated with elevated serum and urinary N-ε-carboxymethyl-lysine (CML). Ironically, these are the same foods found to have lower CML levels, as measured by UPLC-MS. One possible explanation for this paradox is that the source of the elevated CML is the intestines, not the food. When considered collectively, dAGE research results are consistent with the “fructositis” hypothesis, which states that intake of foods and beverages with high fructose-to-glucose ratios (HFCS-sweetened foods and beverages, agave syrup, crystalline fructose, apple juice, and apple juice blends) promotes the intestinal in situ formation of readily absorbed, proinflammatory extracellular, newly identified, fructose-associated AGE, an overlooked source of immunogenic AGEs.

During the last 25 y, in vivo–generated advanced glycation end products (AGEs) and dietary AGEs (dAGEs) garnered all of the attention in AGE research (1–9). However, the extent to which dAGEs contribute to chronic diseases has been debated (10–18). According to early kinetic studies in rats, 10–30% of dAGEs are intestinally absorbed and enter the circulation (12). Research of their metabolic fate postabsorption is limited, and until recently, there was limited consensus regarding the content of dAGEs in foods and beverages because the accuracy of the first dAGEs databases by one research group (19, 20), which showed that high-fat foods have the highest AGE content (19, 20), has been questioned (10–18).

Researchers established these dAGEs databases by measuring food concentrations of N-ε-carboxymethyl-lysine (CML) (19, 20), a well-studied marker of AGEs formation, via ELISA with an anti-CML monoclonal antibody (4G9 Vlassara, Alteon, Inc.) (21, 22). According to their measurements, grilled (cooked by dry heat) meats, cheese, French fries, and fast foods (hamburgers, chicken, and fried fish) had the highest CML concentrations relative to other foods (19, 20). However, independent research groups have been unable to replicate their results, despite the use of the same antibody (12, 23) and state-of-the-art ultraperformance LC tandem-MS (UPLC-MS) (13, 15, 16, 24). The focus of this Perspective is on how paradoxical inconsistencies in dAGE research provide evidence that AGEs form in the gastrointestinal tract (GI) lumen.

Inconsistencies in dAGE Research Provide Support and Plausibility of Enteral Fructose-Associated AGE Formation

A review of outcome differences in dAGE research provides support for the hypothesis that proinflammatory fructose-associated AGEs form intestinally [extracellular, newly identified, fructose-associated advanced glycation end product (enFruAGE)], due to underlying fructose malabsorption (25). In 2011, researchers (12) from Johns Hopkins University, University of California, the Linus Pauling Institute, and the US National Institute on Aging collaborated on a study aimed at describing the adult dietary intake most associated with elevated serum and urinary CML. Smokers and individuals with diabetes mellitus were excluded because smoking is a putative risk factor for AGE exposure, and endogenous AGE production is increased in diabetics. Researchers hypothesized that participants who frequently ate foods prepared with higher cooking temperatures (fast foods, including hamburgers and French fries) would have elevated serum and urinary CML (12). They measured serum and urinary CML in 261 adults, with the same monoclonal antibody (4G9) used to produce the first dAGE database, because one of their goals was to validate earlier findings by Goldberg et al. (19) and Uribarri et al. (20). They compared these results with diet, as assessed by 6 separate 24-h dietary recalls, the Energetics Study (12).

The findings of Semba et al. (12) were unexpected. Higher consumption of foods considered high in CML was not a major determinant of serum or urinary CML. Intake of foods thought to be high in AGEs (fast foods), such as fried chicken, French fries, bacon or sausage, and crispy (high-heat–processed) snacks were not correlated with serum or urinary CML, except for a significant negative correlation between fried chicken and serum CML. Semba et al. (12) noted that their findings were consistent with an earlier study from 2001 by Sebeková et al. (23) who used the same monoclonal antibody (4G9) to test the hypothesis that consumption of high-heat–treated foods (grilled, broiled, and fried meats) would be reflected in higher plasma and urinary CML concentrations of omnivores, relative to vegetarians, because vegetarians consume foods cooked at lower temperatures that require less cooking time than foods consumed by omnivores. Contrary to their working hypothesis, Sebeková et al. (23) found that serum CML concentrations were somewhat higher in vegetarians than omnivores. They questioned whether increased free fructose consumption contributed to increased plasma AGE concentrations in vegetarians, because vegetarians consumed significantly more high-fructose fruits (fresh and dried apples) than omnivores. They concluded that the relevance of higher fructose intake among vegetarians was questionable (23).

Dietary AGE Research Provides Evidence of Enteral Fructose-Associated AGE Formation and a Link to Fructose Malabsorption

In retrospect, the research of Semba et al. (12) and Sebeková et al. (23) provides indirect support for and plausibility of intestinal CML formation between unabsorbed excess free fructose (EFF) and free amino groups from nonmeat proteins (25). EFF is defined as fructose-to-glucose ratios that exceed 1:1 (26). In the Sebeková et al. (23) study, vegetarians, the group with elevated serum CML, consumed a lot of apples and apple juice (23). Apples contain a ≥2:1 ratio of fructose to glucose. One apple contains ∼4.3 g EFF and 237 mL (one 8-oz cup) 100% apple juice contains 8–9 g EFF (27). In fructose malabsorption research, 30% and 10% of participants tested positive for fructose malabsorption after a 25- and 12-g EFF challenge, respectively (28–35); these intake amounts are easily reached by individuals who consume a lot of apples and apple juice, because the EFF content in 237 mL (one 8-oz cup) apple juice and 3 apples comes to ∼22 g (27). Other high-fructose fruits include pears (5.9 g/medium-sized pear), mangoes (4.4 g/mango), and watermelons [2.8/152 g (1 diced 8-oz cup)] (27).

In the Semba et al. (12) study, contradictory to their hypothesis, serum CML was positively correlated with intake of soy, fruit juice, cold breakfast cereal, nonfat milk, whole grains, fruit, nonstarchy vegetables, and legumes, which are foods synonymous with healthy eating, and was negatively correlated with intake of red meat. Urinary CML was positively correlated with intake of starchy vegetables, whole grains, sweets, nuts, seeds, and, to a lesser degree, chicken and was negatively correlated with intake of fast foods (12). It is possible that the correlation with fruits and fruit juice was driven by intake of 100% apple juice, apple juice blends, and high-EFF fruits and that the correlation with breakfast cereals, whole grains (breads), and sweets (cookies, cakes, pies, snack bars, and candy) was driven by the EFF in high-fructose corn syrup (HFCS). During the time period of the Semba et al. (12) study (2010–2011), these foods were significant sources of HFCS (36–39). In fact, HFCS was ubiquitous in the US food supply (36–39). Although fructose amounts in these foods have not been measured, it is possible that they contained more fructose than previously thought, because independent laboratories have measured the fructose content in popular US sodas and found that they contained 65% (40) and 60% (41) fructose, not the 55% that is generally recognized as safe (42). In a per capita average intake of ∼65 g HFCS/d—consumption during the time of Semba et al.’s study (unadjusted for an arbitrary 30% “loss”) (37, 38)—there are 6.4 g of excess free fructose when the fructose percentage is 55%, and 13 and 19.4 g when the fructose percentage is 60% and 65%, respectively.

Notably, nonfat milk, consumed alongside cold cereals, is a rich source of potentially reactive amino acids. In fact, vegetables, legumes, and, in particular, soy are potential sources of free amino groups capable of reacting with unabsorbed fructose, because the reactive amino groups in these foods are far less likely to have already undergone Maillard reaction modifications during cooking, given that they are cooked at lower heat and for less time than meats. Therefore, the formation of a high concentration of readily absorbed CML peptides in the GI lumen is possible (25, 43).

The association between intake of noncaffeinated beverages (types not distinguished) and serum CML approached statistical significance (P = 0.06, correlation coefficient: 0.124) (12). This beverage category, as distinguished from nonfat milk, whole milk, and caffeinated beverages, may have included noncaffeinated soft drinks, fruit drinks, and sports drinks, which are drinks that are sweetened with HFCS (36, 44). Therefore, it is possible that high EFF beverages, included in this beverage category, drove the association with serum CML, particularly because soft drinks have been found to contain more fructose than is generally recognized as safe (40–42). The inclusion of sucrose-sweetened beverages may have attenuated this relation. Notably, fructose malabsorption is not associated with intakes of sucrose (i.e., table sugar) (28–35). There was no association with caffeinated beverages (types not distinguished) (12). However, the inclusion of sucrose-sweetened beverages (coffees and teas) may have prevented a true assessment of the association with caffeinated beverages sweetened with HFCS.

Use of UPLC-MS in dAGE Research Resulted in Consensus, but Did Not Explain Associations between Dietary Intake and Elevated Serum and Urinary CML

Dietary AGEs research with UPLC-MS did not explain the associations found between dietary intakes and elevated serum and urinary CML. One of the first UPLC-MS–based studies found that evaporated milk had high concentrations of CML, but not cooked beef (24). Hull et al. (15) found that, other than Doner kebab (rotisserie-style, skewered, slowly grilled pork, lamb, or chicken), red meat did not contain high CML concentrations, irrespective of the cooking method. Rather, canned grilled salmon, grilled mackerel, and shortbread had the highest CML (milligrams per 100 g) concentrations; intermediate concentrations were found in roasted peanuts, peanut butter, chocolates, varieties of cold cereals, toasted breads, and biscuits; and the lowest CML concentrations were found in fruits and vegetables. In contrast with earlier ELISA-based studies by Goldberg et al. (19) and Uribarri et al. (20), Hull et al. (15) found that the CML content in hamburgers and French fries was relatively low.

The findings of Hull et al. (15) were consistent with another UPLC-MS study by Scheijen et al. (16), who reported that foods highest in CML (milligrams per 100 g) were black pudding made from fried crushed bacon, frankfurters, dark chocolate, chocolate sprinkles, and (≥90%) peanut butter; intermediate concentrations were found in toast, biscuits, evaporated milk, peanuts, peanut butter (<90%), cocktail nuts, varieties of cold cereal, and cake; and the lowest CML concentrations were found in fruits and vegetables. Notably, Scheijen et al. (16) also found that the CML content in hamburgers and French fries was relatively low (16). Importantly, the results of Scheijen et al. (16) largely confirmed and validated the results of Hull et al. (15). Recent research by Trevisan et al. (13) also provided further evidence of low CML in cooked beef (hamburgers) (13). This repeatability is an important tenet of biomedical research. It is a fair assumption that results with UPLC-MS are accurate. Therefore, it is reasonable to hypothesize that frequent consumers of bacon, frankfurters, dark chocolate, grilled salmon, grilled mackerel, and shortbread would have elevated serum and urinary CML. However, this is not what Sebeková et al. (23) and Semba et al. (12) found.

Additionally, dAGE research has shown that diets previously thought to be high in dAGEs were not associated with elevated biomarkers of inflammation (11, 45–47), whereas diets low in dAGEs (high in fruits, vegetables, and whole grains) were consistently associated with low inflammation and serum biomarkers (10, 11, 17, 45–47), synonymous with good health. It is possible that these associations have less to do with dAGE concentrations and more to do with the fact that fruits, vegetables, and whole grains are rich in antioxidants (10, 11, 17, 45–47). It has been suggested that bottom-up proteomics would be a useful tool to disentangle the paradoxical results of dAGE research (1), because it will provide valuable information regarding the protein source of postprandial and fasting serum and urinary CML peptides and of other proinflammatory AGEs [pentosidine, glucosepane, N-ε-carboxyethyl-lysine (CEL), and hydroimidazolones] (8, 9, 11) that may be harbingers of chronic disease.

When considered collectively, these studies provide indirect support and plausibility of enteral fructose-associated advanced glycation end product (FruAGE) formation, an overlooked source of proinflammatory AGEs (25). Conditions in the digestive tract may accelerate CML formation, because bicarbonate from pancreatic juice and phosphates from the phosphoric acid in soda are potent anion catalysts of the Maillard reaction (1, 48, 49). Recent in vitro research provides evidence of AGE formation between fructose and reactive amino acids and peptides (but not with glucose) at a pH consistent with the intestines within 1 h of exposure, a timeframe well within the time window of digestion (50). As this paper was being written, another study was published that showed that FruAGE adducts form between fructose and egg protein (ovalbumin) at times (30 min) and concentration ranges (10 mM) plausibly found in the intestines, but not with glucose (51).

NMR-based research has shown that peptides as small as 7 amino acids long, with CML and CEL, were capable of receptor (i.e., receptor of AGE) binding and proinflammatory signaling via oligomerization in a manner that was independent of any specific amino acid sequence that followed CML or CEL (7, 52). Hence, a myriad of partially digested small CML and CEL peptides, as may form in the GI lumen, can function as receptor activating immunogens, irrespective of the amino acid sequence that follows CML and CEL (43). Indeed, size differences between dAGEs (larger, not readily absorbed) and enFruAGEs (smaller, readily absorbed) may explain the differences in the fates of these AGEs. Notably, recent epidemiologic research with nationally representative survey data (NHANES) has provided evidence that intake of beverages with high fructose-to-glucose ratios, including HFCS-sweetened soda, fruit drinks, and apple juice, are associated with higher odds and prevalence of asthma (43, 53), chronic bronchitis (54), non–wear-and-tear, non–age-associated arthritis (55), and coronary heart disease (56), possibly due to underlying fructose malabsorption and enteral FruAGE formation (53–56). There were no associations with orange juice, a juice with a 1:1 ratio of fructose to glucose, or diet drinks (53–56).

In conclusion, the paradox in dAGE research—that elevated serum and urinary CML was associated with intake of foods low in CML—as measured by UPLC-MS (13, 15, 16, 24), supports the hypothesis that the source of the elevated serum and urinary CML is the intestines, not the food. It is possible that readily absorbed CML could have formed in the GI lumen between unabsorbed EFF (from HFCS-sweetened foods and beverages, apple juice, and high-fructose fruits) and the free amino groups in nonmeat proteins, because they are less likely to have undergone Maillard reaction modifications given that they are cooked for less time and at lower temperatures than meats. Recent epidemiologic (43, 53–56) and in vitro (50, 51) studies support the hypothesis that intake of foods and beverages with high fructose-to-glucose ratios promotes the intestinal in situ formation of readily absorbed, proinflammatory enFruAGEs that are harbingers of chronic disease (25). These results suggest that the ubiquitous presence of high-fructose sweeteners [HFCS-sweetened foods and beverages, agave syrup (57, 58), crystalline fructose, and apple juice (27)] in the US food supply is a public health hazard. Additional research into the greater consequences of fructose malabsorption is long overdue.