Concentrations of Maillard reaction products and advanced glycation end-products in commercially available extruded dry, retorted wet, and mildly cooked (fresh) cat foods from the United Kingdom and United States

Abstract

Maillard reaction products (MRP), including advanced glycation end-products (AGE), form during food processing, particularly under high temperatures, and vary across feline diet formats. Increased AGE consumption has been associated with adverse health effects in humans and rodent models. The health implications of AGE intake in cats remain unclear, and limited data on MRP levels in feline-specific diets exist, particularly in mildly cooked (fresh) formats. The objective of this study was to quantify MRP (mg/kg DM), including the early-stage MRP fructoselysine (FL) and three advanced MRP (AGE): Nε-carboxymethyllysine (CML), Nε-carboxyethyllysine (CEL), and methylglyoxal-derived hydroimidazolone-1 (MG-H1), in commercially available cat foods in the United Kingdom and United States representing mildly cooked (n = 6), retorted wet (n = 10), and extruded dry (n = 10) formats to better understand dietary exposure and potential biological relevance. This study also estimated daily AGE intake and evaluated associations between dietary macronutrient composition and MRP concentrations. Lastly, total lysine (TL) and reactive lysine (RL), were measured to assess potential processing-related changes in lysine availability. Selected MRP, including AGE were quantified by ultra-performance liquid chromatography–tandem mass spectrometry and compared by one-way ANOVA. FL concentrations differed among diet formats (P < 0.05) and were greater in mildly cooked and retorted wet than in extruded dry. CML concentrations differed among diet formats (P < 0.05), with retorted wet greater than extruded dry, while mildly cooked diets were intermediate. Concentrations of CEL (P < 0.05), MG-H1 (P < 0.05), and total AGE (P < 0.05) were greatest in retorted wet diets, whereas mildly cooked and extruded dry diets did not statistically differ. Estimated daily AGE intake (mg/kg BW^0.75) also differed among formats (P < 0.05), with retorted wet diets resulting in greater intake than mildly cooked and extruded dry. Levels of TL (P < 0.05) and RL (P < 0.05) differed among diet formats, with extruded dry containing the lowest levels, although the RL: TL ratio did not significantly differ among formats. Within-format correlations between nutrients and MRP were limited and differed by format. These findings demonstrate that commercially available feline diets differ in concentrations of MRP, including AGE, and estimated AGE exposure, with retorted wet diets exhibiting the greatest values for all AGE among evaluated formats.

Advanced glycation end-products vary substantially across commercial cat food processing formats. Retorted wet diets exhibited the greatest concentrations and estimated dietary exposure, highlighting the influence of processing methods on feline AGE intake.

Introduction

Feeding of commercially available retorted wet and extruded dry diets continues to predominate in cat feeding practices (Dodd et al. 2020; Vanderhoydonck 2022; O’Halloran et al. 2024). In a 2019 global owner survey, 80.7% of respondents reported feeding their cats exclusively retorted wet food (4.5%), extruded dry food (16.1%), or a mixture of both (60.1%) (O’Halloran et al. 2024). Similarly, a 2017 international observational study reported that 90% of cats were offered conventional dry (kibble) and wet (canned, pouches, rolls) diets, with 86% fed these formats daily (Dodd et al. 2020). However, purchasing of non-commercial and unconventional pet foods was more common than previously reported (Dodd et al. 2020). This trend aligns with growing consumer interest in alternative formats, with mildly cooked (often marketed as fresh; refrigerated or frozen) cat food sales increasing by 14.5% from 2019 to 2020 in the United States (Semple 2021). Parallel to this market growth, research evaluating this emerging category of fresh, mildly cooked, and minimally processed diets has also expanded (Algya et al. 2018; Bridglalsingh et al. 2024; Geary et al. 2024; Kocadağlı et al. 2026; Campbell et al. 2026).

Commercial feline diets differ substantially in processing intensity, moisture content, and ingredient composition (Tanprasertsuk et al. 2026), which may influence heat-induced nutrient modifications and the formation of Maillard reaction products (MRP) (Zhang et al. 2020). Extruded dry foods are produced by grinding and mixing ingredients, preconditioning with steam (70–90 °C), and rapidly (10–270 s) subjecting the mixture to high temperature (125–150 °C) and pressure within an extruder, followed by heated drying (90–180 °C) to create small, crunchy pieces with a final moisture content <12% (Tran 2008; Rokey et al. 2010; Bigeard et al. 2014; Tanprasertsuk et al. 2026). Retorted wet foods are formulated from ground or emulsified ingredients that may be pre-processed under heat (110–240 °C) as loaves or restructured chunks combined with gravy or jelly, then packaged and sterilized through retort processing (115–130 °C for 20–120 min) to produce shelf-stable products with approximately 80% moisture (Dierking et al. 2006; Niamnuy and Devahastin 2011; Mathieu et al. 2016, 2017; Tanprasertsuk et al. 2026). Mildly cooked (fresh) diets, an emerging category, are typically prepared by grinding and mixing raw ingredients that are packaged and cooked at lower temperatures (70–90 °C), often by steaming or kettle cooking, resulting in products containing >65% moisture and requiring refrigeration or freezing to maintain freshness (Tanprasertsuk et al. 2021, 2026).

Advanced glycation end-products (AGE) arise during food processing primarily through the Maillard reaction, the principal chemical pathway responsible for heat-induced modification of proteins (Hellwig and Henle 2014; Murata 2021). It is initiated by the reaction between carbonyl groups of reducing sugars and amino groups of amino acids, particularly the ε-amino group of lysine and the guanidino side chain of arginine (Hellwig and Henle 2014; Zhao et al. 2019; Murata 2021; Twarda-Clapa et al. 2022). Both lysine and arginine are essential amino acids for cats and may undergo modifications during the Maillard reaction; however, lysine is of particular nutritional relevance because glycation reduces reactive lysine (RL) availability, which serves as a well-established marker of heat-induced nutritional damage, whereas arginine modification is less well characterized nutritionally (Rutherfurd et al. 2007; van Rooijen et al. 2013, 2014).

While the Maillard reaction contributes to desirable sensory attributes, it is also responsible for the generation of a range of MRP, including fructoselysine (FL), Nε-carboxymethyllysine (CML), Nε-carboxyethyllysine (CEL), 5-hydroxymethyl-2-furfural (HMF), lysinoalanine (LAL), and methylglyoxal-derived hydroimidazolone-1 (MG-H1) (Rutherfurd et al. 2007; Uribarri et al. 2010; Hellwig and Henle 2014; van Rooijen et al. 2014; Scheijen et al. 2016; Twarda-Clapa et al. 2022). Formation of early intermediates such as FL, an Amadori product, blocks the reactive site of lysine, thereby reducing its bioavailability, and may further react through a series of pathways to form advanced MRP, collectively referred to as AGE (Goldberg et al. 2004; Erbersdobler and Somoza 2007; van Rooijen et al. 2014; Rowan et al. 2018; Twarda-Clapa et al. 2022). Common AGE markers formed by sugar-derived carbonyls include CEL and MG-H1, whereas CML may also arise from lipid peroxidation and oxidative stress (Hull et al. 2012; Scheijen et al. 2016; Rowan et al. 2018). The extent of MRP formation, including AGE formation, is influenced by temperature, moisture, processing duration, pH, substrate availability, and storage conditions (Uribarri et al. 2010; Lund and Ray 2017). Elevated temperature, reduced moisture, extended processing time, higher pH, and increased availability of protein and reducing sugars generally promote greater accumulation (Uribarri et al. 2010; van Rooijen et al. 2014; Zhang et al. 2020).

Growing evidence from human and rodent studies suggests that dietary AGE may exert biological effects beyond their contribution to sensory and nutritional changes in foods. In humans, greater dietary AGE intake has been associated with increased circulating AGE concentrations and markers of oxidative stress and inflammation (Uribarri et al. 2010; Nowotny et al. 2018). Similarly, numerous observational studies have found positive associations between greater AGE intake (particularly CML, CEL, MG-H1) and increased risk of chronic diseases, including cardiovascular disease (Di Pino et al. 2017; Si et al. 2024), cancer (Jiao et al. 2015; Peterson et al. 2020; Foussard et al. 2021; Jahromi et al. 2023a), metabolic syndrome (Uribarri et al. 2015; Angoorani et al. 2016; Mendoza-Herrera et al. 2018), overweightness (Cordova et al. 2020), dementia and cognitive decline (Schnaider Beeri et al. 2022; Zhang et al. 2023; Mooldijk et al. 2024), liver disease (Jahromi et al. 2023b), and chronic kidney disease (Ejtahed et al. 2016). Experimental rodent models further demonstrate that dietary CML can accumulate in tissues and contribute to endothelial dysfunction, insulin resistance, inflammatory responses, arterial stiffness, and aging (Roncero-Ramos et al. 2014; Grossin et al. 2015; Li et al. 2015; Tessier et al. 2016). Although AGE are also formed endogenously (Nowotny et al. 2018; Twarda-Clapa et al. 2022), exogenous intake represents a modifiable exposure source and has been identified as a major contributor to the circulating AGE pool (Scheijen et al. 2018; Zgutka et al. 2023). Accordingly, numerous studies have investigated the absorption, metabolism, and biological mechanisms underlying AGE-associated effects (Rowan et al. 2018; Snelson and Coughlan 2019; Chen and Guo 2021; Sergi et al. 2021).

In companion animals, data remains limited. A controlled feeding study in eight healthy dogs comparing dietary and plasma AGE concentrations across canned wet, dry, air-dried, and mildly cooked foods reported that dietary AGE intake correlated with total plasma AGE concentrations, with the canned wet diet containing the greatest AGE content corresponding to the greatest circulating concentrations, whereas the mildly cooked diet contained the lowest (Bridglalsingh et al. 2024). In a feeding study of 12 adult female cats fed six commercially processed dry and six moist diets, van Rooijen et al. (2016) demonstrated that daily MRP intake correlated with urinary MRP excretion, indicating systemic absorption of dietary MRP. When reported as-fed concentrations were standardized to a dry matter basis, MRP concentrations were greater in moist than dry diets (van Rooijen et al. 2016). Additional studies quantifying levels of MRP, including AGE, in pet foods have similarly reported greater concentrations in retorted wet products compared with extruded dry or pelleted formats (van Rooijen et al. 2014; Oba et al. 2022). Recent analyses of commercial dog foods likewise reported greater concentrations of several AGE compounds in wet diets compared with mildly cooked and extruded dry foods (Kocadağlı et al. 2026). More recently, a controlled feeding study in dogs reported lower plasma concentrations of two AGE (CML and pyrraline) as well as improved markers of mitochondrial and fatty acid metabolism in dogs fed a fresh, human-grade diet compared with an extruded dry diet (Yamka et al. 2025), providing preliminary evidence that dietary AGE exposure may influence metabolic outcomes in companion animals. However, the typically greater protein content of feline diets compared with canine diets may influence MRP formation, including AGE formation, through increased substrate availability for the Maillard reaction and should be considered when extrapolating findings across species. Despite these findings, quantification of MRP, particularly AGE, in commercial pet foods remains limited and has primarily focused on extruded dry and retorted wet formats. Comparative data across contemporary diet formats are sparse, and information on feline-specific mildly cooked diets is currently nonexistent.

A recent market survey revealed that health and ingredient quality are primary drivers of purchasing decisions for many cat owners, yet a majority report difficulty determining whether ingredients and processing methods support their cat’s nutritional needs (Vanderhoydonck 2022). While the long-term health implications of exogenous AGE intake in cats remain unclear, processing conditions are known to influence formation of MRP, including advanced MRP such as AGE. Given the growing consumer interest in alternative diet formats, comparative data on MRP concentrations, particularly AGE, across contemporary feline diet types are warranted to better understand dietary exposure and potential biological relevance. Assessment of lysine, particularly RL (nutritionally available), provides a complementary indicator of heat-induced nutritional changes alongside MRP formation. Therefore, the objectives of the present study were to quantify four common dietary MRP, including the Amadori compound FL and the AGE (CML, CEL, and MG-H1) in commercially available feline diets representing mildly cooked, retorted wet, and extruded dry formats. Additionally, we aimed to estimate daily AGE intake and evaluate associations between dietary macronutrient composition and AGE concentrations within each format. Secondly, total lysine (TL) and RL were quantified to assess lysine availability and to compare the RL:TL ratio as an assessment of lysine damage across diet formats. We hypothesized that mildly cooked diets subjected to lower-temperature processing would contain lower concentrations of AGE than conventional retorted and extruded formats. Additionally, we hypothesized that greater dietary protein and carbohydrate concentrations would be positively associated with overall MRP formation, including AGE formation.

Materials and methods

Cat food diets and sample preparation

This study did not involve live animals, animal handling, or animal-derived biological samples. Therefore, Institutional Animal Care and Use Committee (IACUC) approval was not required. All procedures were limited to the analysis of commercially available pet food products. A total of 26 commercially available complete cat food diets were purchased in either the United States or the United Kingdom. Diets were categorized according to diet format into extruded dry (n = 10), retorted wet (n = 10), or mildly cooked (fresh; n = 6), reflecting typical processing approaches. All diets were labeled as formulated to meet nutritional requirements for adult maintenance or all life stages and listed chicken as the first named meat ingredient. Mildly cooked diets were selected to include all commercially available brands meeting these criteria in the U.S. and U.K. at the time of sampling, reflecting the limited number of products available within this category. For extruded dry and retorted wet formats, sample selection aimed to capture a range of commercial products available in the U.S. and U.K. markets by including products from brands with high market share (Pet Care 2023: Cat Food (United States and United Kingdom) 2023), as well as those marketed as “premium” (through claims emphasizing ingredient quality and high whole meat inclusion), thereby capturing variation within these product categories under our selection criteria.

All diets were purchased and shipped in original packaging to Wageningen University (Wageningen, The Netherlands) during spring of 2025 using a global GDP-compliant, cold chain shipping specialist (Biocair Inc.). Mildly cooked diets were maintained frozen (−20 °C) during shipment, whereas retorted wet and extruded dry diets were shipped under ambient conditions consistent with commercial storage.

All laboratory analyses and sample preparation procedures were performed at Wageningen University in late 2025. Moist (retorted wet and mildly cooked) foods were freeze-dried before grinding and analysis to facilitate sample homogenization and storage. Dry foods were analyzed directly. All samples were ground to pass a 1-mm sieve in a centrifugal mill (ZM100, Retsch BV, Retsch Benelux, Aartselaar, Belgium) and stored in airtight plastic containers at 4 °C until analysis. Ground food samples were defatted and ball milled before analysis (van Rooijen et al. 2014).

Determination of MRP and AGE

MRP, including the AGE (Nε-carboxymethyllysine [CML], Nε-carboxyethyllysine [CEL], methylglyoxal-derived hydroimidazolone-1 [MG-H1]), and the early intermediate fructoselysine (FL, measured as furosine) were quantified using ultra-performance liquid chromatography–tandem mass spectrometry (UPLC–MS/MS), following established materials and methods (Scheijen et al. 2016), with minor adjustments.

Briefly, prepared food samples (10 mg) were weighed into borosilicate glass tubes and reduced with sodium borohydride, except for FL analysis, for which the reduction step was omitted. Samples were hydrolyzed with 6 M hydrochloric acid for 23 h at 110 °C after flame-sealing tubes under vacuum. After hydrolysis, an internal standard mixture of deuterated AGE was added to obtain a final concentration of 1 mg/L, and hydrolysates were evaporated to dryness using a vacuum concentrator (SpeedVac Plus, SC210A, Thermo Fisher Scientific Inc., Waltham, MA). Residues were reconstituted in a mixture of 50% mobile phase A and mobile phase B and filtered over a 0.2 µm syringe filter before UPLC-MS/MS analysis. The butylation step was omitted, as hydrophilic interaction liquid chromatography (HILIC) was chosen over reversed phase chromatography for separation.

UPLC–MS/MS analysis

Analysis was performed on a triple quadrupole UPLC-MS/MS system (Acquity UPLC + Quattro Premier XE, Waters, Milford, MA). Optimal conditions for CEL, CML, furosine and MG-H1 were acquired by infusion of a 1 mg/L standard solution. The multiple reaction monitoring (MRM) transitions for CEL, CML, furosine, and MG-H1 were, respectively, 219.2 > 130.1, 205.3 > 130.1, 255.3 > 130.1, and 229.3 > 166.2 m/z. The corresponding MRM transitions for CEL-d4, CML-d4, furosine-d4, and MG-H1-d4 were 233.2 > 134.1, 209.3 > 134.1, 259.3 > 134.1, and 233.3 > 170.2 m/z.

Separation was performed by injecting 2 µL on an Accucore HILIC column (2.6 µm, 2.1 × 100 mm) (Thermo Fisher Scientific Inc., Waltham, MA) using a gradient of 18 mM ammonium formate, adjusted to pH 3.0 (mobile phase A) with formic acid and 0.1% formic acid in acetonitrile (mobile phase B) at a flow of 0.2 mL/min. Mobile phase A was kept at 30% for 1 min, then ramped to 60% over 6 min, kept at 60% for 2 min, then returned to 30% and equilibrated for 4 min before the next injection.

Furosine was used as an indirect marker of FL, and FL concentrations were calculated assuming a 32% conversion of peptide-bound FL to furosine during 6 M acid hydrolysis (Krause et al. 2003).

The limits of detection and quantification were 1 and 5 mg/kg, respectively, for each MRP. MRP concentrations measured in defatted samples were corrected for removed fat during sample preparation before expression on a whole-diet dry matter basis (mg/kg DM). Equivalent caloric-basis data (mg/1,000 kcal ME) are additionally provided in the Supplementary Material.

Determination of TL and RL

TL was determined in ground, defatted food samples after hydrolysis with 6 M HCl for 23 h. Hydrolysates were evaporated to dryness using a vacuum concentrator (SpeedVac Plus, SC210A, Thermo Fisher Scientific Inc., Waltham, MA), after which the residue was dissolved in sample loading buffer (pH 2.2). Amino acids were separated by ion exchange chromatography and determined by post-column reaction with ninhydrin, using photometric detection at 570 nm according to ISO 13903 (International Organization for Standardization 2005).

For RL, samples were incubated for 7 d with O-methylisourea (OMIU) to convert RL into homoarginine. The solvent was evaporated to dryness using the vacuum concentrator. Homoarginine was then determined after the procedure for TL. The amount of RL was calculated from the amount of homoarginine multiplied by 0.77. Unreactive lysine (URL) was determined by subtracting RL from TL. Lysine fractions were expressed on a DM basis (% DM). Equivalent caloric-basis data (g/1,000 kcal ME) are additionally provided in the Supplementary Material.

For comparison, RL was also independently estimated with the furosine method using FL concentrations, assuming a 56% conversion of peptide-bound FL to regenerated (unreactive) lysine during 6 M acid hydrolysis (Krause et al. 2003). The corresponding URL fraction was calculated and subtracted from measured TL to estimate RL. Associations between the two methods for determining RL were assessed using Spearman’s rank correlation coefficients. This approach was used solely to evaluate agreement with the OMIU-derived RL values, and FL-derived RL values were reported only for method comparison.

Proximate analysis and calculations

Proximate composition, including DM, crude ash, nitrogen (N), crude fat (CFat), and crude fiber (CFiber) was determined as described previously (van Rooijen et al. 2014). All analyses were performed in duplicate. Crude protein (CP) was calculated as N × 6.25. Starch and reducing sugar concentrations were also determined to further characterize carbohydrate composition.

Determination of reducing sugars was based on previously described methods, with the segmented flow analyzer omitted (van Vuuren et al. 1993). Briefly, samples were extracted with 40% ethanol and hydrolyzed with hydrochloric acid. The extracted sugars reduced copper II to copper I, which was colored with neocuproine (2,9-dimethyl-1,10-phenanthroline hydrochloride). The resulting complex was measured photometrically at 460 nm using glucose solutions as a calibration curve.

Starch content was determined enzymatically according to ISO 15914 (International Organization for Standardization 2004). Metabolizable energy (ME, kcal/100g, as-is) of each diet was calculated using standardized predictive equations (National Research Council 2006; European Pet Food Industry Federation (FEDIAF) 2025), as outlined below.

Nitrogen-free extract (NFE):

$$\mathit{NFE}\mathrm{ }\left(\mathrm{\%}\right)=100-(\mathrm{\%}\mathrm{ }\mathit{moisture}+\mathrm{\%}\mathrm{ }\mathit{CFat}+\mathrm{\%}\mathrm{ }CP+\mathrm{\%}\mathrm{ }\mathit{crude}\mathrm{ }\mathit{ash}+\mathrm{\%}\mathrm{ }\mathit{CFiber})$$

Calculate gross energy (GE):

$$GE\mathrm{ }(\mathit{kcal}/100g)=(5.7\mathrm{ }\times \mathrm{ }\mathrm{\%}\mathrm{ }CP)+(9.4\times \mathrm{\%}\mathrm{ }\mathit{CFat})+[4.1\times (\mathrm{\%}\mathrm{ }\mathit{NFE}+\mathrm{ }\mathrm{\%}\mathrm{ }\mathit{CFiber}\left)\right]$$

Calculate energy digestibility (ED):

$$ED\mathrm{ }\left(\mathrm{\%}\right)\mathrm{ }=87.9-\mathrm{ }(0.88\times \mathrm{\%}\mathrm{ }\mathit{CFiber}\mathrm{ }in\mathrm{ }DM)$$

Calculate digestible energy (DE):

$$DE\mathrm{ }(\mathit{kcal}/100g)\mathrm{ }=GE\times \frac{ED}{100}$$

Calculate metabolizable energy (ME):

$$ME\mathrm{ }(\mathit{kcal}/100g)\mathrm{ }=\mathrm{ }DE\mathrm{ }-\mathrm{ }(0.77\mathrm{ }\times \mathrm{ }\mathrm{\%}\mathrm{ }CP)$$

Proximate composition and ME variables are reported on a DM basis unless otherwise specified. Equivalent caloric-basis data (g/1,000 kcal ME) are additionally provided in the Supplementary Material.

Estimated daily AGE intake

The maintenance energy requirement (MER; kcal/day) was calculated as 100 x BW^0.67 for a 4 kg adult cat in ideal body condition (National Research Council 2006; European Pet Food Industry Federation (FEDIAF) 2025). Daily intake of diet (DID; g/day) was then estimated by dividing the MER for a 4 kg cat (253.15 kcal/day) by the ME of each diet (kcal/g). Finally, the daily intake of AGE (DIAGE; mg/day) for each diet was calculated by multiplying the DID by the dietary AGE concentrations (mg/g, as-fed).

To facilitate comparison with prior studies and to allow comparison across diet formats on a body size-adjusted basis, estimated daily AGE intake was also expressed per unit of metabolic body weight (mg/kg BW^0.75 per day), calculated as DIAGE divided by BW^0.75, consistent with previous studies assessing dietary AGE exposure in companion animals (van Rooijen et al. 2014; Oba et al. 2022).

Statistical methods

Statistical analyses were performed in R (version 4.5.0; R Foundation for Statistical Computing, Vienna, Austria). Dietary MRP concentrations, estimated AGE intake, and nutrient composition were compared across diet categories (extruded dry, retorted wet, mildly cooked) using one-way ANOVA. Significance level was set at α = 0.05, and P-values from pairwise comparisons were adjusted to control the false discovery rate using the Benjamini–Hochberg method. Associations among individual MRP, and between MRP concentrations and dietary macronutrient composition, were assessed using Spearman’s rank correlation coefficients separately within each diet category without P-value adjustment for multiple comparisons, given the exploratory nature of this analysis.

Results

Proximate composition of commercially available cat foods

The proximate composition and ME content of the evaluated cat foods differed markedly across diet formats, including mildly cooked, retorted wet, and extruded dry formats (Table 1). The proximate composition is also reported on a g/1,000 kcal ME basis in Table S1.

Table 1.

Proximate composition and metabolizable energy (ME) content of 26 commercially available nutritionally complete cat foods.

Component		Mildly cooked (n = 6)	Retorted wet (n = 10)	Extruded dry (n = 10)	P-value
Moisture, % “as is”	Mean ± SD	71.72 ± 3.82a	79.99 ± 3.50b	6.73 ± 1.22c	<2.00E-16
	(Min–Max)	(64.40–74.80)	(73.40–84.50)	(4.80–9.00)
CP, % DM	Mean ± SD	58.98 ± 9.28a	63.97 ± 11.69a	38.93 ± 5.45b	8.41E-06
	(Min–Max)	(45.30–70.10)	(45.10–84.00)	(28.80–46.40)
CFat, % DM	Mean ± SD	21.45 ± 6.80a	17.40 ± 6.31a	10.81 ± 3.76b	3.17E-03
	(Min–Max)	(16.10–34.70)	(8.40–30.00)	(6.10–15.60)
CFiber, % DM	Mean ± SD	1.57 ± 1.14	2.60 ± 2.21	4.28 ± 2.47	5.47E-02
	(Min–Max)	(0.60–3.50)	(0.70–7.00)	(2.10–8.50)
Crude ash, % DM	Mean ± SD	8.98 ± 1.28	8.08 ± 2.00	8.13 ± 0.91	4.64E-01
	(Min–Max)	(7.40–11.10)	(5.40–12.20)	(6.80–9.50)
NFE 1 , % DM	Mean ± SD	9.00 ± 8.03a	7.95 ± 7.34a	37.88 ± 7.60b	8.15E-09
	(Min–Max)	(1.90–23.90)	(-2.00–21.00)	(27.90–49.50)
Sugar, % DM	Mean ± SD	1.60 ± 0.97	1.80 ± 1.36	1.89 ± 1.08	8.92E-01
	(Min–Max)	(0.70–3.40)	(0.20–4.40)	(0.20–3.60)
Starch, % DM	Mean ± SD	4.40 ± 7.59a	4.08 ± 3.64a	24.98 ± 6.12b	2.57E-08
	(Min–Max)	(0.00–18.80)	(0.00–11.40)	(16.60–34.80)
ME, kcal/100g DM	Mean ± SD	457.8 ± 35.0a	440.3 ± 31.6a	387.9 ± 24.5b	1.98E-04
	(Min–Max)	(424.0–520.0)	(372.0–493.0)	(354.0–419.0)

¹Calculated as % DM: 100 − [CFat(%) + CP(%) + crude ash(%) + CFiber(%)]

^a,b,cMeans with unlike superscripts differ (P < 0.05).

Moisture content (% as-is) differed significantly among all three diet formats (P < 2.00E-16). Retorted wet diets contained the greatest moisture content (79.99 ± 3.50), followed by mildly cooked diets (71.72 ± 3.82), whereas extruded dry diets contained the lowest moisture content (6.73 ± 1.22). On a DM basis, CP levels differed significantly among diet formats (P = 8.41E-06). While CP content (% DM) was comparable between mildly cooked (58.98 ± 9.28) and retorted wet diets (63.97 ± 11.69), levels were significantly lower in extruded dry diets (38.93 ± 5.45). Crude fat content (% DM) followed a similar pattern, with greater levels in mildly cooked (21.45 ± 6.80) and retorted wet diets (17.40 ± 6.31) compared with extruded dry diets (10.81 ± 3.76; P = 0.003). Starch levels (% DM) also differed significantly among diet formats (P = 2.57E-08), with extruded dry diets containing substantially greater starch levels (24.98 ± 6.12) than mildly cooked (4.40 ± 7.59) and retorted wet diets (4.08 ± 3.64).

ME content (kcal/100g DM) differed significantly among diet formats (P = 1.98E-04). Mildly cooked (457.8 ± 35.0) and retorted wet diets (440.3 ± 31.6) had greater ME value than extruded dry diets (387.9 ± 24.5), consistent with the lower CFat and CP and greater NFE content of extruded dry diets.

TL and RL content of commercially available cat foods

The lysine content of the evaluated cat food varied widely (Table 2; Figure 1) and is also reported on a g/1,000 kcal ME basis in Table S2. The RL content of all diets exceeded the minimum lysine requirement (0.85% DM; 2.13 g/1,000 kcal ME) for growth and reproduction (European Pet Food Industry Federation (FEDIAF) 2025).

Table 2.

Lysine fractions and reactive lysine ratio in 26 commercially available cat foods categorized by diet format.

Component		Mildly cooked (n = 6)	Retorted wet (n = 10)	Extruded dry (n = 10)	P-value
TL, % DM	Mean ± SD	3.79 ± 0.59a	3.03 ± 1.66a	1.88 ± 0.54b	9.13E-03
	(Min–Max)	(2.92–4.52)	(1.72–6.66)	(1.17–3.15)
RL 1 , % DM	Mean ± SD	3.42 ± 0.60a	2.76 ± 1.59ab	1.71 ± 0.51b	1.42E-02
	(Min–Max)	(2.69–4.17)	(1.55–6.22)	(0.99–2.89)
URL 2 , % DM	Mean ± SD	0.37 ± 0.17a	0.27 ± 0.10ab	0.18 ± 0.08b	1.04E-02
	(Min–Max	(0.19–0.61)	(0.16–0.44)	(0.04–0.32)
RL:TL	Mean ± SD	0.90 ± 0.05	0.90 ± 0.04	0.90 ± 0.04	9.91E-01
	(Min–Max)	(0.83–0.95)	(0.81–0.94)	(0.83–0.98)

¹Reactive lysine calculated using the O-methylisourea (OMIU) method.

²Unreactive lysine calculated by subtracting RL from TL.

^a,bMeans with unlike superscripts differ (P < 0.05).

Figure 1.

Total lysine concentration of 26 commercially available cat foods categorized by diet format. Each bar represents a sample. The dashed line indicates the minimum lysine requirement (0.85% DM) for growth and reproduction according to FEDIAF (European Pet Food Industry Federation) nutritional guidelines.

Total lysine (TL; % DM) differed significantly among diet formats (P = 9.13E-03). TL content was comparable between mildly cooked (3.79 ± 0.59) and retorted wet diets (3.03 ± 1.66), whereas levels were significantly lower in extruded dry diets (1.88 ± 0.54). Among diet formats, RL concentrations (% DM), also varied significantly (P = 1.42E-02). Concentrations in retorted wet (2.76 ± 1.59) did not differ markedly from mildly cooked (3.42 ± 0.60) or extruded dry (1.71 ± 0.51) diets, whereas RL levels in extruded dry diets were significantly lower than mildly cooked diets. URL concentrations also differed significantly among diet formats (P = 1.04E-02), following a similar pattern. Levels in extruded dry diets (0.18 ± 0.08) were significantly lower than mildly cooked diets (0.37 ± 0.17), while levels in retorted wet diets (0.27 ± 0.10) were intermediate. Despite differences in lysine fractions, the RL: TL ratio (lower values are indicative of greater lysine modification) did not differ among diet formats and was consistent around 0.90 across formats.

RL determined using the OMIU method, as reported herein, was compared with values estimated from the furosine method using FL concentrations. The two methods showed a strong correlation (Figure S1 [see online supplementary material for a color version of this figure]; ρ  =  0.98; P < 2.00E-16), and this relationship remained significant after controlling for diet type (partial ρ  =  0.97, P = 9.59E-15). However, RL values calculated via the furosine method were, on average, approximately 7.30% greater.

MRP concentrations, including AGE, in commercially available cat foods

Concentrations of individual and combined MRP differed significantly among cat food formats (Table 3; Figure 2). The MRP concentrations are also reported on a mg/1,000 kcal ME basis in Table S3. Considerable variability was also observed within format categories.

Table 3.

Average amount (mg/kg DM) of individual and combined Maillard reaction products (MRP) of 26 commercially available cat foods categorized by diet format.

Maillard reaction product		Mildly cooked (n = 6)	Retorted wet (n = 10)	Extruded dry (n = 10)	P-value
MG-H1	Mean ± SD	24.17 ± 9.33a	202.7 ± 92.72b	46.80 ± 12.45a	2.05E-06
	(Min–Max)	(17.00–40.00)	(56.00–396.0)	(27.00–63.00)
CEL	Mean ± SD	37.83 ± 19.50a	88.10 ± 40.97b	45.90 ± 19.17a	3.59E-03
	(Min–Max)	(21.00–68.00)	(45.00–187.0)	(18.00–85.00)
CML	Mean ± SD	51.67 ± 24.70ab	59.50 ± 19.61a	31.30 ± 11.57b	7.01E-03
	(Min–Max)	(30.00–92.00)	(25.00–94.00)	(17.00–50.00)
Total AGE (CML + CEL + MG-H1)	Mean ± SD	113.7 ± 38.30a	350.3 ± 99.90b	124.0 ± 38.91a	1.39E-07
	(Min–Max)	(68.00–165.0)	(162.0–511.0)	(62.00–198.0)
FL	Mean ± SD	2,240 ± 1,234a	1,691 ± 893a	752 ± 407b	5.80E-03
	(Min–Max)	(967–3,826)	(305–3,328)	(226–1,312)

^a,bMeans with unlike superscripts differ (P < 0.05).

Figure 2.

Individual Maillard reaction product (MRP) concentrations (mg/kg DM) of 26 commercially available cat foods categorized by diet format. Each bar represents a sample. Bars corresponding to the advanced glycation end-products (AGE) Nε-carboxymethyllysine (CML), Nε-carboxyethyllysine (CEL), and methylglyoxal-derived hydroimidazolone-1 (MG-H1) are shown with full opacity, whereas bars corresponding to fructoselysine (FL), an early MRP are shown with reduced opacity.

Retorted wet diets on average had the greatest concentrations (mg/kg DM) of all three AGE: MG-H1 (202.7 ± 92.72), CEL (88.10 ± 40.97), and CML (59.50 ± 19.61). Concentrations of both MG-H1 (P = 2.05E-06) and CEL (P = 3.59E-03) differed significantly among diet formats. Retorted wet diets contained significantly greater concentrations of both AGE than mildly cooked (MG-H1, 24.17 ± 9.33; CEL, 37.83 ± 19.50) and extruded dry diets (MG-H1, 46.80 ± 12.45; CEL, 45.90 ± 19.17). Although mean MG-H1 and CEL concentrations were lower in mildly cooked than extruded dry diets, no significant differences were detected. Among diet formats, CML concentrations also varied significantly (P = 7.01E-03). Concentrations in mildly cooked (51.67 ± 24.70) did not differ markedly from retorted wet or extruded dry diets, whereas extruded dry diets contained the lowest average CML concentrations (31.30 ± 11.57), which were significantly lower than those measured in retorted wet diets.

Concentrations of FL (mg/kg DM), an early MRP, differed significantly among diet formats (P = 5.80E-03). Mildly cooked (2,240 ± 1,234) and retorted wet diets (1,691 ± 893) were comparable, and both contained significantly greater FL concentrations than extruded dry diets (752 ± 407).

The concentrations (mg/kg DM) of the three AGE (CML, CEL, and MG-H1) were combined to estimate total AGE concentration, and significant differences among diet formats were observed (P = 1.39E-07; Figure 3). Retorted wet diets contained the greatest AGE concentrations (350.3 ± 99.90), which were approximately 3-fold greater than those measured in mildly cooked (113.7 ± 38.30) and extruded dry diets (124.0 ± 38.91). No difference in AGE concentrations was detected between mildly cooked and extruded dry diets.

Figure 3.

Total advanced glycation end-product (AGE) concentration of 26 commercially available cat foods categorized by diet format. Each bar represents a sample.

Estimated daily intake of AGE

Estimated intake of individual and total AGE, expressed per unit of metabolic body weight (mg/kg BW^0.75) for a 4-kg adult cat, differed significantly among cat food formats, and substantial variability was observed within formats, as reflected by the reported ranges (Table 4; Figure 4).

Table 4.

Estimated daily intake (mg/kg BW^0.75) of individual and total advanced glycation end-products (AGE) of 26 commercially available cat foods categorized by diet format.

Advanced glycation end-product		Mildly cooked (n = 6)	Retorted wet (n = 10)	Extruded dry (n = 10)	P-value
MG-H1	Mean ± SD	0.47 ± 0.20a	4.13 ± 1.93b	1.07 ± 0.30a	3.70E-07
	(Min–Max)	(0.30–0.80)	(1.10–8.10)	(0.60–1.50)
CEL	Mean ± SD	0.73 ± 0.40a	1.78 ± 0.76b	1.05 ± 0.41a	3.30E-03
	(Min–Max)	(0.40–1.40)	(1.00–3.60)	(0.40–1.90)
CML	Mean ± SD	1.00 ± 0.45ab	1.23 ± 0.41a	0.72 ± 0.24b	1.61E-02
	(Min–Max	(0.60–1.70)	(0.50–1.90)	(0.40–1.10)
Total AGE (CML + CEL + MG-H1)	Mean ± SD	2.20 ± 0.78a	7.14 ± 2.03b	2.86 ± 0.85a	2.65E-07
	(Min–Max)	(1.30–3.10)	(3.30–10.50)	(1.40–4.40)

^a,bMeans with unlike superscripts differ (P < 0.05).

Figure 4.

Estimated daily intake (mg/kg BW^0.75) of advanced glycation end-products (AGE) from mildly cooked (n = 6), retorted wet (n = 10), and extruded dry (n = 10) cat foods. ^a,bMeans with unlike letters differ (P < 0.05).

Estimated daily intake of MG-H1 (P = 3.70E-07) and CEL (P = 3.30E-03) differed among diet formats. Retorted wet diets resulted in significantly greater intake of both MG-H1 (4.13 ± 1.93) and CEL (1.78 ± 0.76) compared with mildly cooked diets (MG-H1, 0.47 ± 0.20; CEL, 0.73 ± 0.40) and extruded dry diets (MG-H1, 1.07 ± 0.30; CEL, 1.05 ± 0.41). Although mean intake of MG-H1 and CEL was lower for mildly cooked than extruded dry diets, these differences were not significant. Estimated intake of CML also differed among diet formats (P = 1.61E-02). Cats consuming retorted wet diets had the greatest CML intake (1.23 ± 0.41), whereas those consuming extruded dry diets had the lowest intake (0.72 ± 0.24), which was significantly lower than retorted wet diets. Intake from mildly cooked diets (1.00 ± 0.45) did not differ significantly from either retorted wet or extruded dry diets.

When MG-H1, CEL, and CML were combined to estimate total AGE intake, significant differences among diet formats were observed (P = 2.65E-07). Retorted wet diets resulted in the greatest total AGE intake (7.14 ± 2.03), which was more than 3-fold greater than mildly cooked diets (2.20 ± 0.78) and approximately 2.5-fold greater than extruded dry diets (2.86 ± 0.85). No significant difference in estimated total AGE intake was detected between mildly cooked and extruded dry diets.

Dietary factors associated with MRP levels

Correlation analyses were conducted to evaluate associations between nutrient composition (% DM), and concentrations of individual MRP and total AGE (mg/kg DM) within each diet format (Figure 5).

Figure 5.

Heatmap of Spearman correlation coefficients (ρ) between nutrient concentrations (% DM) and Maillard reaction products (MRP; mg/kg DM) in commercially available cat foods categorized by diet format. Asterisks indicate significant correlations (P < 0.05). Significant associations were observed as follows: in mildly cooked fresh samples, CML–CP (P = 0.033); in retorted wet samples, CML–CP (P = 0.028), CML–CFat (P = 0.028), and FL–Sugar (P = 0.035); and in extruded dry samples, total AGE–crude ash (P = 0.034), MG-H1–crude ash (P = 0.032), and FL–crude ash (P = 0.046). Color intensity reflects the strength and direction of correlations (blue = negative; red = positive). Total AGE, advanced glycation end products (CML + CEL + MG-H1).

In mildly cooked diets, CML was positively correlated with CP (ρ  =  0.89; P = 3.33E-02). No other significant associations between nutrients and individual MRP or total AGE were detected in this diet format. In retorted wet diets, CML was negatively correlated with CP (ρ = −0.69; P = 2.82E-02) and positively correlated with CFat (ρ  =  0.69; P = 2.82E-02). Additionally, FL was positively correlated with sugar (ρ  =  0.68; P = 3.51E-02). No other significant nutrient–MRP associations were identified in the retorted wet format. In extruded dry diets, crude ash was positively correlated with total AGE (ρ  =  0.67; P = 3.38E-02) and MG-H1 (ρ  =  0.67; P = 3.23E-02) and negatively correlated with FL (ρ = −0.64; P = 4.61E-02). No significant associations were observed between CP, CFat, starch, sugar, or CFiber and MRP in this format.

Correlations among MRP are presented in Figure S2 (see online supplementary material for a color version of this figure). In retorted wet diets, CML was positively correlated with MG-H1 (ρ  =  0.71; P = 2.11E-02) and total AGE (ρ  =  0.70; P = 2.45E-02), and MG-H1 was strongly correlated with total AGE (ρ  =  0.88; P = 1.98E-03). In extruded dry diets, CML (ρ  =  0.83; P = 5.56E-03), CEL (ρ  =  0.87; P = 2.68E-03), and MG-H1 (ρ  =  0.77; P = 9.43E-03) were each positively correlated with total AGE. Additionally, CEL was positively correlated with MG-H1 (ρ  =  0.68; P = 2.95E-02). No significant inter-MRP correlations were detected in mildly cooked diets.

Discussion

This study quantified four common MRP (FL, CML, CEL, and MG-H1), three of which are AGE, in 26 cat foods commercially available in the U.K. and U.S. All diets were nutritionally complete for adult cats or all life stages, and represented mildly cooked, retorted wet, and extruded dry formats. We also calculated the estimated daily intake of CML, CEL, and MG-H1 as an index of dietary AGE exposure. Retorted wet diets contained the greatest CEL, MG-H1, and total AGE concentrations, whereas mildly cooked and extruded dry diets were broadly comparable. CML followed a slightly different pattern, with retorted wet diets being greatest and extruded dry diets lowest, while mildly cooked diets were intermediate. FL, an early Maillard product, was greater in mildly cooked and retorted wet diets than in extruded dry diets. Estimated intake patterns paralleled concentration differences, with retorted wet diets yielding the greatest total AGE intake, driven primarily by greater CEL and MG-H1 exposure.

In companion animals, published information on dietary MRP and AGE exposure remains limited, particularly for feline diets and newer product categories. While AGE are formed endogenously in vivo, thermal processing can contribute substantially to dietary AGE exposure in heat treated foods (Uribarri et al. 2010; Twarda-Clapa et al. 2022). In humans, greater dietary AGE exposure has been associated with adverse health conditions (Nowotny et al. 2018; Si et al. 2024), motivating interest in dietary strategies to reduce exogenous AGE intake (Uribarri et al. 2010). Comparable causal relationships have not been established in dogs or cats, but available evidence indicates that dietary AGE are systemically absorbed. Urinary excretion of CML, CEL, and LAL was lowest in dogs and cats fed raw food diets compared with extruded dry and retorted wet formats (Palaseweenun et al. 2021), and in cats, urinary MRP excretion increased with dietary intake, supporting systemic absorption and contribution to circulating pools (van Rooijen et al. 2016). In dogs, greater dietary AGE intake has also been linked to increased circulating AGE concentrations (Bridglalsingh et al. 2024). A recent canine study reported reduced serum AGE and improved metabolic markers in dogs fed a mildly cooked (fresh) diet compared with an extruded dry diet (Yamka et al. 2025); these results offer preliminary evidence that AGE intake may negatively affect pet health, although long-term health effects has not yet been reported. As such, while the biological context for why high MRP intake, particularly of AGE, could negatively impact pet health exists, the present results should be interpreted as a description of dietary MRP concentrations and as a foundation for future studies assessing long-term health relevance in cats.

To our knowledge, no other study has estimated AGE intake in mildly cooked diets, and among feline studies quantifying AGE in commercial cat foods, CML is the only overlapping AGE for which estimated daily intake has been reported or could be calculated across studies. Previous studies estimated that adult cats (4 kg) consume 0.8, 1.5, and 17.1 mg/kg BW^0.75 CML per day from retorted wet diets and 0.3, 1.2, and 2.41 mg/kg BW^0.75 CML per day from extruded dry diets (van Rooijen et al. 2014, 2016; Oba et al. 2022). The present study mirrored increased CML intake from retorted wet (1.23 mg/kg BW^0.75) versus extruded dry diets (0.72 mg/kg BW^0.75). However, estimated intakes varied greatly across studies. The present study also estimated the intake of CML for a 4 kg adult cat fed mildly cooked diets (1.00 mg/kg BW^0.75), which was intermediate between retorted wet and extruded dry diets. These intake estimates are based on predicted energy requirements and normalization to BW^0.75, consistent with prior studies, but do not reflect measured food intake and rely on generalized scaling assumptions, which may introduce uncertainty in absolute exposure estimates. Additionally, differences in analytical methodology used to quantify AGE across studies may further limit direct comparison of reported dietary AGE concentrations and intake estimates. This study also found that cats fed retorted wet diets on average consume the greatest levels of CML, CEL, and MG-H1, with estimated daily AGE intake approximately a 3.2-fold increase from mildly cooked diets, and a 2.5-fold increase from extruded dry diets. Retorted wet diets therefore represented the greatest exposure to the advanced markers quantified; however, the number and composition of products within each format reflect the sampled market set and may not represent all products in each category.

The diet format differences observed here support the interpretation that processing may influence dietary AGE concentrations in commercial feline diets, although the broad range observed within diet formats indicates that factors beyond processing method alone, such as ingredients, additives, and formulation, contribute to variability. Pet food palatability enhancers and animal digests produced using hydrolysis and the Maillard reaction may additionally contribute to dietary MRP exposure, particularly in coated extruded products (van Rooijen et al. 2013; Samant et al. 2021). The influence of processing is most evident when comparing mildly cooked and retorted wet diets in the present dataset, as proximate composition was broadly similar across these two formats except for moisture, yet advanced reaction products (particularly CEL and MG-H1) and the combined total AGE metric were greater in retorted wet products. Because these were commercial diets, the specific time–temperature histories and pre-processing steps of ingredients and additives are unknown; however, the greater AGE burden in retorted wet diets aligns with broader evidence that higher heat processing can increase AGE formation in foods (Uribarri et al. 2010; Zhang et al. 2020). Consistent with this interpretation, preliminary internal observations indicated that when a mildly cooked diet was subjected to sterilization conditions comparable to retort processing, CEL and MG-H1 concentrations increased markedly (unpublished data), further supporting processing intensity as a driver of AGE formation. In contrast, extruded dry diets did not exhibit the same AGE burden as retorted wet diets in this dataset, despite being manufactured using significant thermal conditions. This pattern is likely reflective of shorter processing times for extruded dry foods at peak temperature (10–270 s) compared with retorted wet foods (20–120 min) (Tanprasertsuk et al. 2026). Additionally, formulation differences, including lower protein and fat in extruded dry diets, may constrain accumulation of certain AGE, and the low moisture and dry texture of extruded foods may further restrict reactant mobility and AGE formation (Uribarri et al. 2010; Zhang et al. 2020; El Hosry et al. 2025). The observation that mildly cooked diets, presumed to experience lower processing temperatures, contained a comparable amount of AGE to the extruded dry format underscores that AGE formation likely reflects the interplay of processing, matrix, and formulation rather than temperature alone.

The observed separation between FL and advanced markers (CML, CEL, and MG-H1) is consistent with their different positions within the Maillard pathway. FL, an Amadori compound, is formed early from lysine glycation and may undergo further rearrangements and oxidation, but its abundance does not necessarily predict downstream advanced MRP and AGE formation (van Rooijen et al. 2014; Twarda-Clapa et al. 2022). As one of the first identified MRP, FL is often quantified as a marker of heat treatment; however, because it forms under much milder conditions than advanced MRP, the latter may better represent high heat treatment (Erbersdobler and Somoza 2007). Across the limited literature quantifying FL in commercial pet foods, retorted wet products have generally shown greater FL concentrations than extruded dry products, although absolute values vary substantially between studies (van Rooijen et al. 2014, 2016; Oba et al. 2022), likely reflecting differences in analytical methods and the products sampled (Nowotny et al. 2018). The elevated FL observed in retorted wet diets relative to extruded dry diets is consistent with the findings in this study; however, the current dataset also evaluated FL levels in mildly cooked diets, which had not been done previously. Mildly cooked diets had the greatest mean FL concentration, which may reflect differences in moisture content, as early glycation products do not require high heat treatment for formation (Erbersdobler and Somoza 2007; Uribarri et al. 2010; El Hosry et al. 2025). The retorted wet diets had significantly greater moisture (73.40–84.50%) than the mildly cooked diets (64.40–74.80%), which may have diluted the reactants, while other substrates known to influence MRP formation (protein, fat, sugar) were comparable between these two diet formats.

From a nutritional perspective, quantification of lysine fractions provides additional context for interpreting FL formation, as lysine bound in FL is not nutritionally available (Erbersdobler and Somoza 2007). In the present study, RL levels, representing nutritionally available lysine, exceeded the minimum lysine requirement for growth and reproduction in all diets regardless of format. The RL: TL ratio, with lower values commonly used as an indicator of greater heat-induced lysine damage (van Rooijen et al. 2014; Oba et al. 2022), also did not significantly differ among formats, indicating that the proportion of lysine affected by processing was relatively consistent across formats. The RL: TL ratios observed in the present study (0.90 across formats) were generally comparable to previous reports in commercial feline diets (van Rooijen et al. 2014; Oba et al. 2022), although published data for mildly cooked feline diets are currently lacking. Lower RL: TL ratios have also been reported (Rutherfurd et al. 2007), likely reflecting differences in formulation, ingredients, and analytical methodology. Inclusion of lysine-containing palatability enhancers in commercial pet foods, particularly in extruded dry diets, may have influenced measured lysine fractions in finished products (van Rooijen et al. 2013; Samant et al. 2021). Despite having the greatest FL concentrations, mildly cooked diets also had the greatest mean TL and RL concentrations, likely reflecting their ingredient profile and high inclusion of whole meat ingredients (Tanprasertsuk et al. 2026). Retorted wet diets had the broadest range of lysine values, which likely reflects the heterogeneity in ingredient composition, with premium brands (high whole meat content) exhibiting lysine concentrations more comparable to mildly cooked diets. Collectively, these findings indicate that although FL reflects early lysine modification, it did not translate to differences in lysine availability among diet formats. Furthermore, consistent with previous work, the RL: TL ratio reflects overall lysine reactivity but does not indicate the formation of advanced MRP (van Rooijen et al. 2014; Oba et al. 2022), reinforcing that multiple markers are required to fully characterize Maillard reaction progression.

The measurement of more advanced MRP is important to better understand the impact of heat treatment, as well as because advanced MRP, particularly AGE, are associated with chronic disease risk in humans and experimental models (Li et al. 2015; Nowotny et al. 2018; Si et al. 2024). Both CEL and MG-H1 are commonly linked to glycation reactions involving reactive dicarbonyls such as methylglyoxal (Rowan et al. 2018; Twarda-Clapa et al. 2022). Their marked elevation in retorted wet diets in this dataset suggests that retorting conditions may preferentially promote dicarbonyl-driven pathways and/or provide sufficient thermal exposure for advanced products to accumulate relative to other formats. To our knowledge, the levels of CEL and MG-H1 in pet foods have only been quantified in a limited number of studies. One prior study quantified both these AGE, but assessed only a single sample per category (canned wet food, dry food, air-dried food, and mildly cooked food), and all samples were canine diets (Bridglalsingh et al. 2024). The CEL pattern was similar to that observed here, with CEL greatest in the retorted wet sample and comparable between the extruded dry and mildly cooked samples (Bridglalsingh et al. 2024). This CEL pattern was also observed in a more recent survey of 41 commercial dog foods representing fresh (mildly cooked), kibble, wet, freeze-dried, and cold-pressed formats, with wet diets again exhibiting the greatest CEL concentrations, whereas mildly cooked and dry (kibble) diets contained significantly lower and comparable concentrations (Kocadağlı et al. 2026). In contrast, the previous study quantifying MG-H1 found a different pattern than that reported herein, with MG-H1 greatest in the mildly cooked sample (Bridglalsingh et al. 2024). However, differences in species, formulation, analytical methodology, and product representation across studies may also contribute to variation between datasets.

The pattern for CML differed from CEL and MG-H1, possibly because CML can form through multiple routes, including lipid-derived oxidation pathways, in addition to glycation chemistry (Hull et al. 2012; Scheijen et al. 2016). In this dataset, retorted wet diets had greater CML than extruded dry diets, while mildly cooked diets were intermediate. Prior studies have similarly reported greater CML in retorted wet cat foods than extruded dry cat foods (van Rooijen et al. 2014, 2016; Oba et al. 2022), although absolute values differed, potentially reflecting differences in analytical methodology, product selection, and ingredient profiles (Nowotny et al. 2018). No feline studies have reported CML levels in mildly cooked diets. A canine study reported lower CML concentrations in a single mildly cooked sample (7.41 mg/kg DM), potentially reflecting its lower protein and fat content (Bridglalsingh et al. 2024), whereas a more recent study of 15 fresh (mildly cooked) dog foods reported concentrations (69 ± 19 mg/kg DM) more comparable to those observed here (Kocadağlı et al. 2026). Consistent with prior work reporting positive correlations between dietary CML and crude fat (van Rooijen et al. 2014), we also observed the same correlation within retorted wet diets. This observation supports the interpretation that lipid oxidation may contribute to CML variability and explain the divergence from CEL and MG-H1 patterns across diet formats. A similar fat-CML correlation was not detected in the mildly cooked format, possibly due to lower cooking temperature. However, significantly greater crude fat levels in both the retorted wet and mildly cooked formats than the extruded dry format support the notion that CML formation may be greater in fat rich foods.

Additional format-specific correlative patterns among MRP and between MRP and proximate nutrients were detected, further underscoring that formulation and processing, among other factors, may influence how individual markers co-vary, although the relatively small number of samples within each diet format should be considered when interpreting these correlation analyses. Earlier studies often evaluated correlations across pooled species (dog and cat) and diet formats, which may have obscured format-specific relationships (van Rooijen et al. 2014; Oba et al. 2022). Because nutrient–MRP associations were limited and differed by format, proximate composition alone provides an incomplete explanation for advanced MRP formation, particularly given the likely contribution of uncharacterized processing variables and palatability enhancers that may themselves contain MRP. The format-specific interrelationships among MRP further suggest the complexity of the Maillard reaction. Within some diet formats, advanced MRP may accumulate in tandem, whereas in others, variability in ingredient sourcing or pre-processing may create distinct precursor pools. We did not detect a consistent FL–CML association, as was seen previously (van Rooijen et al. 2014). Given FL was the most abundant MRP across all samples, our results may indicate that FL abundance does not necessarily translate into greater advanced MRP concentrations, which is plausible given that FL sits earlier in the Maillard pathway and can be diverted into multiple downstream or competing pathways (Goldberg et al. 2004; Erbersdobler and Somoza 2007). Together, these observations support that no single MRP is sufficient to represent heat damage and downstream AGE formation across heterogeneous products and that multi-analyte profiling is more informative for comparing commercial diets. As research on the health implications of AGE in companion animals’ advances, understanding the factors associated with MRP variability within specific diet formats will be important for informing strategies to mitigate dietary exposure.

From an applied perspective, these results indicate that diet format can meaningfully affect dietary exposure to AGE in nutritionally complete commercial cat foods. However, within-format variability remains large, pointing to ingredient sourcing, additive use, modifiable formulation, and processing levers. Future work should therefore prioritize controlled processing experiments using matched formulations, with staged sampling before and after key manufacturing steps, although the limited transparency of commercial processing conditions may constrain such approaches (Tanprasertsuk et al. 2026). Notably, the present study extends the available dataset on feline diets to include mildly cooked diets and provides concurrent quantification of multiple advanced MRP markers commonly used in dietary AGE exposure assessment (Uribarri et al. 2010; Nowotny et al. 2018; Twarda-Clapa et al. 2022).

Glossary

Abbreviations

AGE

advanced glycation end-products

BW

body weight

CEL

Nε-carboxyethyllysine

CFat

crude fat

CFiber

crude fiber

CML

Nε-carboxymethyllysine

CP

crude protein

DE

digestible energy

DIAGE

daily intake of AGE

DID

daily intake of diet

DM

dry matter

ED

energy digestibility

FL

fructoselysine

GE

gross energy

HILIC

hydrophilic interaction liquid chromatography

HMF

5-hydroxymethyl-2-furfural

LAL

lysinoalanine

ME

metabolizable energy

MER

maintenance energy requirement

MG-H1

methylglyoxal-derived hydroimidazolone-1

MRM

multiple reaction monitoring

MRP

Maillard reaction products

N

nitrogen

NFE

nitrogen-free extract

OMIU

O-methylisourea

RL

reactive lysine

TL

total lysine

UPLC–MS/MS

ultra-performance liquid chromatography–tandem mass spectrometry

URL

unreactive lysine

Funding

Funding support for this article was provided by KatKin (Anikin Ltd), London, United Kingdom.

Author contributions

Devon E. Tate (Conceptualization, Formal analysis, Investigation, Methodology, Writing—original draft, Writing—review & editing), Jirayu Tanprasertsuk (Conceptualization, Formal analysis, Investigation, Methodology, Writing—review & editing), Dominique S. Tarr (Conceptualization, Writing—review & editing), Justin Shmalberg (Conceptualization, Writing—review & editing), and Ryan W. Honaker (Conceptualization, Formal analysis, Investigation, Methodology, Supervision, Writing—review & editing)

Supplementary data

Supplementary data are available at Journal of Animal Science online.

Conflict of interest statement. D.E.T., J.T., D.S.T., and R.W.H. are employees of KatKin (Anikin Ltd), a cat food company based in the United Kingdom. J.S. serves as a paid advisor to Katkin. KatKin did not influence the analyses, interpretation, or writing of the manuscript. The paper reflects the independent views of the authors.