1. Introduction

Maillard reaction (MR) is one of the most important sources of heat-generated compounds ^[1]. The interactions between IFs components, mainly affect carbohydrates and proteins through of MR, but those involving proteins are especially important in products used in infant feeding because of the high protein requirements of infants ^[2]. The manufacture of IFs includes the component blending, homogenization, pasteurization, and spray-drying ^[2]. The fact that IFs have high lactose and lysine contents, that relatively high temperatures are applied during their manufacturing process and that their storage is quite long, makes them highly sensitive to MR ^[3]. Detrimental effects of thermal processes are also relevant. The loss of thermo labile compounds such as vitamins, essential amino acids (lysine, tryptophan) and/or the formation of undesired tastes and off-flavours cause loss in the nutritional value and sensorial quality of heated foods ^{[4, 5, 6]}. Besides, protein and carbohydrate ingredients are also subjected to thermal processes for their production, so they can also present heat damage ^[7].

The extent of the MR in foods can be monitored with different chemical indexes. The main objective of such indicators are to define the nutritional status, organoleptic characteristics or even the possible toxicity of the foodstuff after the thermal process and storage, then obtaining food products of good quality and high nutritional value ^[8]. Furfural compounds (Hydroxymethylfurfural (HMF) and furfural) are a recognized indicator of the deterioration produced by excessive heating or storage in a wide range of carbohydrate-containing foods ^{[7, 9, 10, 11]}. Furosine determination has also been used to study early stages of the MR during the heat treatment and storage of IFs ^{[2, 12, 13]}. Available lysine content is an indicator of both early and advanced MR phases ^[14], and several studies have been published on lysine loss due to the heat treatment and storage of IFs ^{[2, 13]}, and model systems ^[15]. The lysine, as one of the essential amino acids, is involved in this reaction, and bioavailability is lost ^[16].

Heat-generated compounds may create a risk to human health. For some years, adverse effects from neo-formed contaminants have been the subject to increased attention, particularly acrylamide, nitrosamines, heterocyclic amines, polycyclic aromatic hydrocarbons, furanic compounds (Furan, HMF) and advanced glycation end products ^{[17, 18]}. Furanic compounds arise from heat-treatment processing and contribute to the sensory properties of cooked foods, in general their presence is appreciated; however, due to its potential harmful effects on human health studies related with its formation, content in a variety of foods ^[19]. However, no relevance for humans concerning carcinogenic and genotoxic effects of HMF are available ^[20].

It is clear that the ingredients used in the IFs directly affect their quality; however, there are few studies where it has been established relationships between the presence or absence of ingredients and the different chemical indicators of MR. On the other hand, there are studies about heat damage by MR and nutritional value in powder IFs, but few where it is estimated the potential toxic risk for infants by furanic compounds in different types of commercial IFs. The purpose of this study was to find relationships between ingredients used and heat damage of IFs, in order to evaluate the nutritional value and possible toxic effect of different types of IFs from Spanish market.

2. Materials and Methods

Thirteen commercial powdered IFs (eight adapted and five follow-up (hypoallergenic, and soybean-based)) were purchased in several local markets. The composition protein-sugar and ingredients used according to label are shown in the Table 1.

Table 1. Commercial Powdered IFs Composition According Label

Download as

PowerPoint Slide

Larger image(png format)

• Tables index

Veiw figure

View current table in a new window

View next table

All chemicals used were of analytical grade. Methanol (HPLC grade), hydrochloric acid, acetonitrile and acetic acid glacial were obtained from Panreac (Barcelona, Spain). Sep-Pack cartridges (C₁₈) were purchased from Waters Millipore (Milford, MA, US). Trichloroacetic acid (TCA), N-ε-2, 4- dinitrophenyl-lysine (DNP-L-Lysine) HCl, fluoro-2, 4- dinitrobenzene (FDNB) were purchased from Sigma-Aldrich (Madrid, Spain). 5-(hidroxymethyl)-furfural and 2-furaldehyde were purchased from Merck (Darmstadt, Germany). Furosine was obtained from Neosystem Laboratories (Strasburgo, France).

Furanic compounds were determined following a method described elsewhere ^[7], with some modifications. Approximately, 0.4 g of sample was clarified (with Carrez I and II) and the water fraction was filtered through 0.2-μm disk filter before injection. The HPLC equipment consisted of a Waters 600 controller (Waters Millipore) with a manual injector and Konic model 200 detector (Linear Instrument Corp, Reno Nevada, NV). The integrator program used was a Millennium chromatography manager (Waters Millipore). 50 µL of filtered solution were separated in a reversed-phase C₁₈ column (Nova-Pack 4 μm; 250 mm x 46 mm id., Cartridge, Waters, Milford, MA. US). The mobile phase was water-acetonitrile (95:5) and the flow rate was 1 ml/min. HMF and furfural were quantified using the external standard method. Duplicate analysis of duplicate samples was carried out (n = 4).

Furosine was determined following the method described by Resmini et al. ^[21] with some modifications ^[7]. The HPLC system consisted of a Perkin-Elmer model 250 pump (Norwalk, CT, USA) with a Waters 717 autosampler (Milford, MD, USA) and a Perkin-Elmer model 235 diode array detector (Norwalk, CT, USA). Data were collected with a 1020 software data system Perkin-Elmer. 50 µL of filtered solution were separated using a C₈ furosine-dedicated column (250 mm x 4.6 mm, Alltech, DeerWeld, IL, USA). Calibration of the chromatographic system was made using the external standard method. Duplicate analyses of duplicate samples were carried out (n = 4).

ε-NDP-lysine was determined by HPLC following the method used for IFs by Contreras-Calderón et al. ^[2]. A sample containing approximately 4 mg of protein was derivatized by addition of FDNB. Hydrolysis of FDNB derivative was realized with HCl. The HPLC study was performed in a Perkin-Elmer 250 model with a Waters 717 automatic injector and Perkin-Elmer 235 UV diode array detector. The integrator-computer used here was a 1020 Perkin-Elmer Nelson model. 50 µL of filtered solution were separated in a Nova pack reverse phase C₁₈ HPLC column (150 mm x 3.9 mm Waters) operating at room temperature. ε-DNP-lysine was determined by the external standard method. Duplicate analyses of duplicate samples were carried out (n = 4).

Multivariate statistics (including PC analyses and CA) were employed to quantitatively investigate relationships among the 13 samples of IFs with respect to the 4 indicators; PC analyses variables were standardized before the analysis and principal components (PC) with eigenvalues greater than one were selected. CA was made through Ward method using square Euclidean distances. One-way ANOVA was applied to the results; differences between clusters were established using LSD test with a level of significance of 95 %. Correlations among variables were assessed by means of the Pearson’s correlation tests (level of significance of 95 %). The impact of ingredients used on IFs quality and chemical indicators was evaluated using multifactor ANOVAs, where factors were the ingredients in the IFs with two levels (present, absent). All the statistical analyses were performed in Statgraphics Centurion XVI®. Left censored data for HMF and furfural was assumed to be zero in the statistical analysis.

3. Results and Discussion

HMF is an intermediate in the MR which is formed from the degradation of Amadori products during the intermediate stage of the MR if the heating conditions are appropriate ^[22]. HMF is considered a heat-induced marker for a wide range of carbohydrate containing foods such as milk ^[23]. HMF is also used for monitoring the heating process applied to cereal products such as baby cereals ^[24] and breakfast cereals ^[25]. But, significance of the measurement of HMF is different depending of the type of food being mainly used as marker of either adulteration or over-processing (i.e. milk), and marker of quality of ageing. In other foods, analysis of HMF is taken together with other furanic compounds such as furfural.

The performance of the method was evaluated in commercial follow-up infant formula, including the preparation and RP-HPLC analysis. Mean recovery values were 99.2% for HMF and 71.1% for furfural, and the relative standard deviations (RSD) were 2.42% and 1.23% for HMF and furfural respectively. The detection limit (LOD) (three times the signal-to-noise ratio) and quantification limit (LOQ) (ten times the signal-to-noise ratio) were 4.95 × 10^-3 and 0,017 mg/100 g of protein for both HMF and furfural.

HMF and furfural values ranging between non-detected for IFs 1, 2, 9 and 11.7 and 0.51 mg/100 g of protein, respectively for IF 11 (Table 2). It is clear that the IF 11 has a higher thermal damage than other IFs. Because the ingredients used are similar to the other IFs, these high furanic compounds values may be due to a high heat treatment, long storage time, or that thermally damaged ingredients have been used. In IFs 1, 2 and 9 where no furanic compounds were detected they must have used good quality ingredients (low thermal damage); additionally, storage conditions and thermal treatments during manufacturing of IFs must not have been very drastic.

Table 2. Furosine, Available Lysine, HMF and Furfural Content in Commercial Powdered IFs.

Download as

PowerPoint Slide

Larger image(png format)

• Tables index

Veiw figure

View current table in a new window

View previous table

View next table

Comparing the results with those reported by other authors, it is observed that in general, the IFs of this study showed higher values of furanic compounds; thus, Albalá-Hurtado et al. ^[26] found that free HMF and furfural values between 10.3 and 23.5 µmol/L (0.84 and 1.91 mg/100 g of protein), and 1.10 and 5.30 µmol/L (0.09 and 0.43 mg/100 g of protein), respectively in powdered IFs stored during 9 months at a 20, 30 and 37°C. Ferrer et al. ^[27] found that HMF levels were 0.34 and 1.84 mg /100 g protein, and between not detected and 0.08 mg furfural/100 g of protein, respectively in adapted and follow-up powdered IFs. Chávez-Servín et al. ^[28] found that initial values of 0.85 mg HMF/Kg and 0.012 mg furfural/Kg during storage of powdered IFs not supplemented with long-chain polyunsaturated fatty acids, and 0.58 mg HMF/Kg in supplemented IFs, where no furfural was detected. The HMF values in IFs from this study were also higher to those obtained in a study conducted by Contreras-Calderón et al. ^[7], in protein ingredients used in the manufacture of IFs.

A significant (p< 0.5) positive correlation was observed between HMF and furfural (r = 0.7826), which is expected since it is considered that the formation of both is affected by the same factors. However, Albalá-Hurtado et al. ^[29], did not find correlation between the formation of HMF and furfural from their precursors in liquid IF`s. The differences in these studies may be due to ingredients used in the IFs and the initial thermal damage of the ingredients.

Based on data reported in literature, it is not clear whether human exposure to HMF represents a potential health risk. Janzowski et al. ^[30] concluded that HMF does not pose a serious health risk, even though the highest concentrations in specific foods approach the biologically effective concentration range in cell systems. HMF was previously considered a quality indicator of thermally processed foods until its toxic properties were highlighted ^[31]. The Scientific Panel on food additives, flavorings, processing aids and materials in contact with foods estimated a dietary HMF intake of 1.6 mg/person per day based on an mTAMDI (modified Theoretical Added Maximum Daily Intake) – approach ^[32]. This estimate is far below the threshold of concern of 540 mg/person/day derived from a large database containing data on sub chronic and chronic animal studies ^[33]. Zaitzev et al. ^[34] suggested that a Tolerable Daily Intake (TDI) of 132 mg/person per day by applying a 40-fold margin of safety.

The currently available assessments of dietary intake remain far below this value. Chávez-Servín et al. ^[35] reported that the range of potential HMF consumed by an infant about 6 months old feeding only on formula was estimated between 0.63 mg and 3.25 mg per day. In the present work, the maximum free HMF and furfural concentration was 14.8 mg/Kg for IF 11 (Table 2; HMF plus furfural). For a baby aged 6 months with an average weight of 8 kg ^[36], the consumption of IF is around 115. 2 g/day (6 measures of 4.8 g per bottle for 4 bottles in 24 h, according to label recommendations); considering that 1 Kg of IF 11 has 14.8 mg HMF and furfural, 115.2 g of this IF will provide 1.7 mg/person of these compounds per day. This value is slightly higher than that established by EFSA ^[32] for the use of HMF as a food additive and is within the range set by Chávez-Servín et al. ^[35]. However, these values are far below the threshold of concern established by other authors ^{[33, 34]}. It should be taken into account the HMF’s genotoxicity potential, the importance of lowering HMF content in infants’ milk-based formula and other heated foods, so the dietary exposure can be carefully re-evaluated, and it has also been recommended by Chávez-Servín et al. ^[35].

The content of furosine (ε-N-2-furoylmethyl-L-lysine) present in foods is influenced by the kind of heat treatment and/or the storage time. Levels of furosine tend to decline after prolonged storage or after overheating to give rise to other compounds such as N ε-(carboxymethyl) lysine (CML) ^{[37, 38]}. Furosine is the most specific and important indicator of the initial phase of the Maillard reaction. It is widely used in the analysis of cereal products, since lysine is the limiting amino acid of this product and, thus the presence of furosine is an important marker of protein biological value loss. The precision of the entire procedure, including acid hydrolysis, sample preparation, and RP-HPLC analysis, was evaluated for a commercial follow-up infant formula (n = 8) with a RSD of 3.4%. The detection limit (LOD) (three times the signal-to-noise ratio) was 0.105 mg/100 g of protein. The quantification limit (LOQ) (ten times the signal- to-noise ratio) was 0.35 mg/100 g of protein.

The furosine values had a wide range, between 379 mg/100 g of protein for IF 13 based on soy protein isolates and 1459 mg/100g of protein for IF 9 prepared with hydrolyzed whey proteins (Table 2). The average value of furosine obtained in the eight IFs made with intact dairy proteins (919 mg/100 g of protein) was similar to that obtained in the four IFs made with partially hydrolyzed milk protein (945 mg/100 g of protein). The mean value of furosine in adapted formulas was slightly superior to follow-up formulas (907 vs. 871 mg/100 g of protein), the major proportion of whey protein vs. casein used in the first formulas could justify this results ^[7].

The above values are similar to those found by other authors. Thus, in IFs made with intact proteins, it has been found values between 141 and 1551 mg/100 g of protein, and between 130 and 900 mg/100g of protein in IFs elaborated with hydrolyzed milk protein ^{[12, 13, 39, 40, 41]}. Baptista and Carvalho ^[42] found very low values in IFs, between 21.4 and 81.5 mg/100g of protein for a formulation without lactose and with partially hydrolyzed whey protein respectively. Contreras-Calderón et al. ^[2] found that similar values in IFs made with intact proteins; however, in IFs made with partially hydrolyzed proteins (PHP) the values were lower (299 mg/100 g of protein) to those found in this study. The few data found in the literature on the analysis of IFs made with soy protein ^[39] are lower (approximately, 270 mg/100 g of protein) that those obtained in this study. High values of furosine in IFs with hydrolyzed protein would indicate the use of ingredients with high initial heat damage since furosine content in powdered hydrolyzed IFs is lower than in regular powdered formulas ^[40], as glycation in these IFs cannot be detected when is estimated by the furosine method ^[41]. In addition, hydrolyzed proteins typically have low levels of furosine ^[2].

The different formulations, the thermal damage of different ingredients and processing of IFs make that this indicator, unlike what happens in milk, can only be used to control the process of manufacturing and storage of the same formula.

A significant (p<0.05) negative correlation was observed between HMF and furosine (r = -0.3889), which suggests a progress of the MR, since at high temperatures (spray drying), the Amadori product may be further degraded by different pathways during the advanced stage of the MR ^[43]. However, it is the IF 13 with isolated soy protein which had the lowest furosine and moderate HMF content. Contreras-Calderón et al. ^[7] noted that soy proteins analyzed did not contain furanic compounds and furosine values were low, so it is possible that the moderate values of HMF may come from maltodextrins ^[44] present in the formulation and not to the MR during processing. In the case of the IF 13, furosine would be the thermal damage indicator since the MR is in the early stages.

No significant correlation (p > 0.05) was found between furfural and furosine which would indicate that in these samples Amadori compounds (measured as furosine) are not furfural precursors since furfurals are both the result of Amadori compounds from the MR or of lactose isomerization ^[45], or still furfural already present in sugar ingredients ^[44].

The precision of the method was evaluated in a commercial follow-up infant formula (n = 8), with a RSD of 4.6%. The detection limit (LOD) (three times the signal-to-noise ratio) for adapted infant formula was 7.25 × 10^-4 g/100 g of protein. The quantification limit (LOQ) (ten times the signal-to-noise ratio) was 2.42 × 10^-3 g/ 100 g of protein ^[2].

Available lysine values were between 2.28 and 4.44 g/100g of protein (Table 2), respectively for IF 12 made with hydrolyzed whey protein and IF 7 made with skim milk. The 62% of IFs presented values above 4.00 g/100 g of protein, being mainly the IFs with PHP and soy protein isolate the ones showing lower values compared with IFs elaborated with intact proteins. Contreras-Calderón et al. ^[7] observed that the same trend in this type of proteins. Pereyra Gonzáles et al. ^[46] also found that available lysine contents in soy based IFs were significantly lower than those found in milk protein based formulas.

The nutritional requirements of lysine for babies aged 0 to 6 months are around 107 mg/kg/day ^[47]. For a baby aged 3 months with an average weight of 6.2 kg ^[36], the formula consumption is around 150 g/day (7 measures of 4.3 g per bottle for 5 bottles in 24 h, according to label recommendations), and the need for lysine is approximately 663.4 mg/day. Only the IF 12 does not cover the needs of children since it provides 425 mg lysine/day. The other IFs provide more than 90% of DRI’s, between 590 and 800 mg lysine/day, for IFs 9 and 7, respectively. However, Contreras-Calderón et al. ^[48] found that adapted IFs exceed the minimum estimated need of infants, which can be attributed to the IFs in the present study or the protein ingredients used had higher heat damage as is observed by the low available lysine (Table 2). Given these results, it is important that the preparation of bottles and daily consumption of IFs are made based on the content of available lysine and not on the total lysine of the protein.

The nutritional requirements of lysine for babies aged 7 to 12 months are around 89 mg/kg/day ^[47]. For a baby aged 12 months with an average weight of 10.2 kg ^[36], consumption of formula is around 115.2 g/day (8 measures of 4.8 g per bottle for 3 bottles in 24 h, according to label recommendations), and the need for lysine is approximately 907.8 mg/day. Only IF 8 covers the minimum requirements of lysine since it provides 939 mg lysine/day. The other IFs provide between 678 and 798 mg lysine/day for IFs 5 and 11, respectively, which makes sense because for a child of 12 months IFs are not the only food. These results differ from those found by Contreras-Calderón et al. ^[48], where the follow–up IFs covered the needs of infants. Similar to the observed in adapted IFs, the follow-up IFs of this study or protein ingredients used, could also have high heat damage.

The data found in the literature are highly variable; thus, Ferrer et al. ^[27] in adapted and follow-up IFs found values of 6.67 and 6.61 g/100g of protein, respectively. Pereyra-Gonzáles et al. ^[46], in different commercial powder IFs, made with soy protein isolates, casein, milk, whey fortified milks and cow's milk, found values between 3.3-5.2; 4.13-7.0; 6.2-8.11; 5.4-7.8 and 8.04 g/100 g of protein, respectively. Ferrer et al. ^[12], analyzed an adapted and follow-up IFs elaborated with milk protein, finding initial available lysine values of 8.43 and 7.9 g/100 g of protein, respectively. Chávez-Servín et al. ^[49], during store of powder IFs based on milk proteins and lactose, found initial available lysine values ranged between 6.01 and 6.04 g/100 g of protein. Contreras-Calderón et al. ^[2], in powdered IFs found values between 4.88 and 5.88 g/100g of protein. Finally, Contreras-Calderón et al. ^[48], in similar IFs found values between 4.99 and 6.44 g/100 g of protein. In general, the values reported in literature are much higher than those obtained in the present work, which may be due to a high heat treatment, long storage time, or that ingredients thermally damaged have been used.

Additionally, no significant correlation (p > 0.05) was found between furosine and lysine; this is surprising if we regard that in milk products, the MR occurs between lactose and lysine, leading a significant decrease in lysine bioavailability, through the formation of the Amadori compound ^[50]. Similar results were reported by Contreras-Calderón et al. ^[7], in protein ingredients, although they found a larger amount of available lysine associated with a larger amount of furosine. On the other hand, Contreras-Calderón et al. ^[2] found that lysine losses were inversely proportional to increases in furosine during processing of IFs. These contradictory results can be explained because it is not easy to compare different types of IFs, as there are studies confirming that the specific formulation, processing conditions applied for IFs and the initial damage of the original ingredients are closely related with an increased risk of protein damage by glycation and oxidation reactions ^{[2, 51]}.

In this study, two PC were obtained which explain 77.3% of total variance. Fig. 1 showed that the biplot (PC₂ vs. PC₁) of IFs samples and variables. PC₁ explained 48.6% of the total variance, which has a positive influence of HMF, furfural, and in lesser extent, a negative influence of furosine and a positive influence of lysine. PC₂ explained 28.7 % of the variance; it is mainly characterized by a positive contribution of lysine and furosine (Figure 1). PC₁ could be seen as an indicator of intermediate stages of MR, as it takes into account the positive contribution of furanic compounds and the negative contribution of furosine; lysine also exhibited a positive contribution probably due to the fact it is released during the intermediate stages of MR in the 1-2 and 2-3 enolisation pathways, and lysine is finally incorporated into melanoidins polymers just at final stages of MR ^[52]. On the other hand, PC₂ could be seen as an indicator of initial stages of MR. That is, PC₂ is higher with higher lysine contents which is due to the fact that lysine is the most reactive amino acid concerning the MR; additionally, PC2 is higher with higher furosine contents; this is useful as an estimation of the advance of initial stages of MR.

Download as

• PPT
  PowerPoint Slide
• PNG
  Larger image(png format)

Veiw figureFigures index

•  NEW
  View larger figure in new window

View next figure

Figure 1. Principal component analysis (PCA) of the IFs according to the determinations of heat damage indicators

CA was performed based on the PCA. Figure 2 showed the distribution of samples inside the clusters according to PC₁ and PC₂, in which five clusters were formed. Cluster 3 is formed by samples 4, 5, 6, 7, 8 and 10; this cluster is located close to coordinate origin, which means that these samples show average for furosine (768 mg/100 g protein), HMF (3.46 mg/100 g protein), furfural (0.11 mg/100 g protein) and lysine (4.15 g/100 g protein) nearby to the global average (885, 2.98, 0.12 mg/100 g protein for furosine, HMF and furfural, respectively and 3.91 g/100 g protein for lysine). It implies that IF sin cluster 3 exhibited intermediate heat damage.

Download as

• PPT
  PowerPoint Slide
• PNG
  Larger image(png format)

Veiw figureFigures index

•  NEW
  View larger figure in new window
• PREV
  View previous figure

Figure 2. Clusters formed from IFs samples according to PC₁ and PC₂

Cluster 1, formed by samples 1, 2 and 9 exhibited the lower value in PC₁ and slightly high value in PC₂, which presented average close to the global average for lysine (3.97 g/100 g protein), higher average for furosine (1186 mg/100 g protein) and lower average for HMF and furfural (no detected) than the global average. These results suggest that samples in cluster 1 did not react in greater extent until intermediate stages of MR, but with a high advance in initial stages. This could indicate that samples in cluster 1 were manufacture with good quality raw material (with minimal heat damage) and the thermal processing was not too strong, as MR just advanced until initial stages.

Cluster 2 represents sample 3 which has the higher value in PC₂ and intermediate value in PC₁ with values close to the average for lysine (4.00 g/100 g protein), high average than global average for furfural (0.26 mg/100 g protein) and furosine (1392 mg/100 g protein) and lower average than global for HMF (1.07 mg/100 g protein). The high furfural contents could be due to isomerization of lactose through LA (Lobry de Bruyn- Alberda Van Ekenstein) transformation to lactulose and its posterior degradation to HMF which subsequently could react to form furfural ^[10]. The high furosine content indicates that important heat damage was induced during the manufacturing process or storage, which can promote the lactose isomerization, as was established by Pereyra Gonzáles et al. ^[46] who found significantly higher lactulose concentrations in whey-enriched IFs compared to others milk based IF`s. Therefore, probably this sample has high initial heat damage and it is starting the intermediate stages of MR.

Cluster 4 represents sample 11 which has the higher value for PC₁ and an intermediate value for PC₂ with values close to the global average for furosine (858 mg/100 g protein) and lysine (4.26 g/100 g protein), and higher average for HMF (11.70 mg/100 g protein) and furfural (0.51 mg/100 g protein) than global average. In fact, sample 11 showed the highest values for HMF and furfural. Sample 11 could be manufactured with damaged ingredients or it could be subjected to an intense heat treatment during processing or storage. However, the formation of furfural and HMF could have been carried out not only by MR, but also by other mechanisms already mention, as the lysine content was not further decreased in comparison with other samples. Based on the results, it could be said that this IF exhibited the largest thermal damage.

Cluster 5 is formed by samples 12 and 13 which has the lower value for PC2 and intermediate value for PC₁ with lower average values for lysine (2.83 g/100 g protein), furosine (545 mg/100 g protein) and furfural (0.05 mg/100 g protein) than global average, and values close to global average for HMF (2.64 mg/100 g protein). The low available lysine content could be due to these IFs are elaborate with PHP and soy protein isolate, which showed lower values than IFs elaborated with intact proteins ^{[7, 46]}, and it does not reflect a high thermal damage, as can be confirmed by the low values of furosine and furfural. The intermediate values for HMF might be from maltodextrins ^[44], so these samples exhibit low heat damage.

Ingredients (Table 1) were used to try to explain the belonging of the samples to the clusters. It was found that cluster 5 is formed by one adapted hyper allergenic IF and the only IF with isolated soy protein, both the only samples without lactose; samples in clusters 3, 4 and 5 do not have whey milk and cluster 3 is formed by samples without partially hydrolyzed whey, except for sample 10. However, it was not possible to find a general relation between ingredients used in the IFs and the heat damage indicators that explained the clustering of the samples, maybe because there are other important underlying variables as the possible interactions between ingredients, the impact of manufacturing process of each IF, the quality of the raw material, etc., which were unknown.

Thus, four multifactor ANOVA’s were made in order to find the relation of ingredients in the IFs and the heat damage indicators. Eight ingredients were selected: lactose, dextrinomaltose, starch, whey milk, skimmed milk, milk proteins, partially hydrolyzed whey (PHW) and isolated soy protein (ISP). All factors with two levels: presence or absence of the ingredient in the IF (Table 3).

Table 3. Results from Multifactor ANOVAs Relating Ingredients of IFs and Heat Damage Indicators

Download as

PowerPoint Slide

Larger image(png format)

• Tables index

Veiw figure

View current table in a new window

View previous table

Available lysine showed significant (p<0.05) higher values in IFs with lactose, starch, whey milk or skimmed milk and absence of PHW, milk proteins or ISP. This is expected because the whey milk contained higher lysine levels than PHW and ISP ^[7]. It is difficult to explain how the lactose favored high lysine content since both the high lactose content and the supplementation with whey proteins promote MR in IFs ^[51]. However, in this study all IFs have lactose, except two IFs elaborated with PHW and ISP, proteins with low available lysine content ^[7], so in this case the high lysine content in IFs with lactose is because these IFs contain proteins with higher lysine content. Pereyra Gonzáles et al. ^[46] also found higher available lysine contents in IFs based on skim milk than in IFs based on whole milk and caseins. The starch (a non-reducing polysaccharide) reduces the MR, favoring the conservation of lysine. Explain available lysine content is very difficult because not only depends on the thermal damage but the type and amount of protein used.

Samples with lactose, whey milk, milk proteins, skimmed milk or PHW and absence of starch, showed significant (p<0.05) higher values of furosine. These results were expected, because it is known that the most dominant among the MR products in milk is certainly lactulosyllysine, the Amadori product of lactose and lysine side chains of the milk proteins ^[53]. However, the higher levels of furosine can be expected in IFs with whey protein respect to other IFs ^[54], and also whey protein can already be highly damaged ^[7]. On the other hand, Fenaille et al. ^[40] and Contreras-Calderón et al. ^[2] found lower values of furosine in IFs with PHW, so as already it is mentioned above the PHW used in these IFs may have high initial heat damage. Likewise, the starch does not favor the MR.

HMF showed significant (p<0.05) higher values in IFs with starch or PHW and absence of whey milk. These results could be explained by the HMF present in the starch because during its production process is subjected to various thermal processes that can hydrolyze the starch to reducing sugars which can participate in MR and caramelization ^[55]. On the other hand, the only two IFs with starch are made with PHW, so we believe that this is the reason why statistical analysis relates this protein with a high content of HMF, but this is not necessarily related to the heat damage of the IF or protein. Contreras-Calderón et al. ^[7] did not found HMF in protein hydrolyzed. These same authors found HMF only in whey proteins, so it is not expected these results since whey proteins usually have higher heat damage than other proteins, which confirm that the high HMF values are due to presence of starch, as the IFs with whey proteins have not this carbohydrate.

Furfural was significant (p<0.05) higher in samples with starch in comparison with the samples which did not have this ingredient in their composition. As already mentioned above, the starch may be an important source of furanic compounds, indicating an initial damage of ingredient but not of the IFs during processing or storage.

4. Conclusions

MR indicators are intrinsically linked as they represent complex relations of reactants in the different MR pathways which depend on the nature of the reactants and the conditions of the heat treatment applied to the food matrix. These complex relations are difficult to follow or explain if just one variable is considered. Statistical analysis is a useful tool to help to interpret results of multiple variables. The main advantage of multivariate analysis is to simplify the interpretation of multiple variables results regarding to a common phenomenon. In the present study, it was possible to reduce the number of variables using PC analysis to see more easily the relation between response variables and it was also possible to group samples with similar heat damage and that way made some hypothesis about the quality of the ingredients used and the manufacture process. There are a variety of formulas with different degrees of thermal damage and biological value in the market. Available lysine values were low however all IFs except one covered more than 90% of needs for a baby aged 3 months. Multivariate statistical analysis proved to be a useful tool in the interpretation of the relation of the MR indicators and their connection with the quality of ingredients used and the manufacture process of IFs. PCA explained the 77.3 % of the total variance with two PC, reducing the number of variables from four to two, which facilitated the CA of samples. CA created 5 clusters and efficiently separated samples according to their heat damage. Finally, HMF content does not represent a risk to health of babies.

Acknowledgements

This work was supported by a predoctoral fellowship obtained from the Junta de Andalucía (Spain) and by a research project from Comisión Interministerial de Ciencia y Tecnología (Project AGL 2001 2977). The authors would like to thank Ramiro Restrepo Marin for assisting with the English version.

Statement of Competing Interests

The authors have no competing interests.

References

[1]	Parker, J. K., “The kinetics of thermal generation of flavour,” Journal of the Science of Food and Agriculture, 93 (2), 197-208, January 2013.
	In article		View Article  PubMed
[2]	Contreras-Calderón, J., Guerra-Hernández, E., and García-Villanova, B., “Utility of some indicators related to the Maillard browning reaction during processing of infant formulas,” Food Chemistry, 114 (4), 1265-1270, June 2009.
	In article		View Article
[3]	Palombo, R., Gertler, A., and Saguy, I., “A Simplified Method for Determination of Browning in Dairy Powders,” Journal of Food Science, 49 (6), 1609-1609, November 1984.
	In article		View Article
[4]	Laguerre, J. C., Pascale, G. W., David, M., Evelyne, O., Lamia, A. A., and Inès, B. A., “The impact of microwave heating of infant formula model on neo-formed contaminant formation, nutrient degradation and spore destruction,” Journal of Food Engineering, 107 (2), 208-213, December 2011.
	In article		View Article
[5]	Van Lancker, F., Adams, A., and De Kimpe, N., “Chemical Modifications of Peptides and Their Impact on Food Properties,” Chemical Reviews, 111 (12), 7876-7903, December 2011.
	In article		View Article  PubMed
[6]	Mesías, M., Seiquer, I., and Navarro, M. P., “Consumption of highly processed foods: Effects on bioavailability and status of zinc and copper in adolescents,” Food Research International, 45 (1), 184-190, January 2012.
	In article		View Article
[7]	Contreras-Calderón, J., Guerra-Hernández, E., and García-Villanova, B., “Indicators of non-enzymatic browning in the evaluation of heat damage of ingredient proteins used in manufactured infant formulas,” European Food Research and Technology, 227 (1), 117-124, July 2008.
	In article		View Article
[8]	Delgado-Andrade, C. and Rufián-Henares, J. ., Eds., Assessing the generation and bioactivity of neo-formed compounds in thermally treated foods, Atrio S.L., Granada, 2009.
	In article
[9]	Ferrer, E., Alegría, A., Farré, R., Abellán, P., and Romero, F., “High-performance liquid chromatographic determination of furfural compounds in infant formulas during full shelf-life,” Food Chemistry, 89 (4), 639-645, 2005.
	In article		View Article
[10]	Chávez-Servín, J., Castellote, A., and López-Sabater, C., “Evolution of potential and free furfural compounds in milk-based infant formula during storage,” Food Research International, 39 (5), 536-543, 2006.
	In article		View Article
[11]	Rufían-Henares, J. a and de la Cueva, S. P., “Assessment of hydroxymethylfurfural intake in the Spanish diet.,” Food additives & contaminants. Part A, Chemistry, analysis, control, exposure & risk assessment, 25 (11), 1306-1312, 2008.
	In article		View Article  PubMed
[12]	Ferrer, E., Alegría, A., Farré, R., Abellán, P., Romero, F., and Clemente, G., “Evolution of available lysine and furosine contents in milk-based infant formulas throughout the shelf-life storage period,” Journal of the Science of Food and Agriculture, 83 (5), 465-472, April 2003.
	In article		View Article
[13]	Guerra-Hernandez, E., Leon, C., Corzo, N., Garcia-Villanova, B., and Romera, J., “Chemical changes in powdered infant formulas during storage,” International Journal of Dairy Technology, 55 (4), 171-176, 2002.
	In article		View Article
[14]	Abellán, P., Farré, R., Alegría-Torán, A., Ferrer, E., and Romero, F., “Indicadores de deterioro de la calidad proteica y valor nutritivo de la leche,” Food science and technology international, 5 (6), 447-461, 1999.
	In article		View Article
[15]	Rufián-Henares, J. Á., García-Villanova, B., and Guerra-Hernández, E., “Generation of furosine and color in infant/enteral formula-resembling systems,” Journal of Agricultural and Food Chemistry, 52 (17), 5354-5358, August 2004.
	In article		View Article  PubMed
[16]	Burri, J., Bertoli, C., and Stadler, R. H., “Food Processing and Nutritional Aspects,” in Process-Induced Food Toxicants, R. Stadler and D. Lineback, Eds. John Wiley & Sons, Inc., Hoboken, NJ, USA, 2008, 645-677.
	In article		View Article
[17]	Viegas, O., Novo, P., Pinto, E., Pinho, O., and Ferreira, I. M. P. L. V. O., “Effect of charcoal types and grilling conditions on formation of heterocyclic aromatic amines (HAs) and polycyclic aromatic hydrocarbons (PAHs) in grilled muscle foods,” Food and Chemical Toxicology, 50 (6), 2128-2134, June 2012.
	In article		View Article  PubMed
[18]	Lineback, D. R. and Stadler, R. H., “Introduction to Food Process Toxicants,” in Process-Induced Food Toxicants, D. R. Lineback and R. H. Stadler, Eds. John Wiley & Sons, Inc., Hoboken, NJ, USA, 2008, 3-19.
	In article
[19]	Petisca, C., “Furanic compounds in food products: Assessment and mitigation strategies,” University of Porto, 2013.
	In article
[20]	Abraham, K., Gürtler, R., Berg, K., Heinemeyer, G., Lampen, A., and Appel, K. E., “Toxicology and risk assessment of 5-Hydroxymethylfurfural in food,” Molecular Nutrition & Food Research, 55 (5), 667-678, May 2011.
	In article		View Article  PubMed
[21]	Resmini, P., Pellegrino, L., and Battelli, G., “Accurate quantification of furosine in milk and dairy products by a direct HPLC method,” Italian Journal of Food Science, 3, 173-183, 1990.
	In article
[22]	Gökmen, V. and Morales, F. J., “Hydroxy-methylfurfural,” Encyclopedia of food safety, 2, 404-408.
	In article
[23]	Morales, F. J. and Jiménez-Pérez, S., “Hydroxymethylfurfural determination in infant milk-based formulas by micellar electrokinetic capillary chromatography,” Food Chemistry, 72 (4), 525-531, March 2001.
	In article		View Article
[24]	Fernández-Artigas, P., Guerra-Hernández, E., and García-Villanova, B., “Browning indicators in model systems and baby cereals.,” Journal of agricultural and food chemistry, 47, 2872-2878, 1999.
	In article		View Article  PubMed
[25]	Rufián-Henares, J. A., Delgado-Andrade, C., and Morales, F. J., “Application of a fast high-performance liquid chromatography method for simultaneous determination of furanic compounds and glucosylisomaltol in breakfast cereals.,” Journal of AOAC International, 89 (1), 161-165.
	In article		PubMed
[26]	Albalá-Hurtado, S., Veciana-Nogués, M. T., Mariné-Font, A., and Vidal-Carou, M. C., “Changes in Furfural Compounds during Storage of Infant Milks,” Journal of Agricultural and Food Chemistry, 46, 2998-3003, 1998.
	In article		View Article
[27]	Ferrer, E., Alegría, A., Farré, R., Abellán, P., and Romero, F., “Effects of Thermal Processing and Storage on Available Lysine and Furfural Compounds Contents of Infant Formulas,” Journal of Agricultural and Food Chemistry, 48 (5), 1817-1822, May 2000.
	In article		View Article  PubMed
[28]	Chávez-Servín, J. L., Castellote, A. I., and López-Sabater, M. C., “Evolution of potential and free furfural compounds in milk-based infant formula during storage,” Food Research International, 39, 536-543, 2006.
	In article		View Article
[29]	Albalá-Hurtado, S., Veciana-Nogués, M. T., Mariné-Font, A., and Vidal-Carou, M. C., “Progress of Browning Reactions during Storage of Liquid Infant Milks,” Journal of Agricultural and Food Chemistry, 47 (10), 4033-4037, October 1999.
	In article		View Article  PubMed
[30]	Janzowski, C., Glaab, V., Samimi, E., Schlatter, J., and Eisenbrand, G., “5-Hydroxymethylfurfural: assessment of mutagenicity, DNA-damaging potential and reactivity towards cellular glutathione,” Food and Chemical Toxicology, 38 (9), 801-809, 2000.
	In article		View Article
[31]	Anese, M. and Suman, M., “Mitigation strategies of furan and 5-hydroxymethylfurfural in food,” Food Research International, 51 (1), 257-264, 2013.
	In article		View Article
[32]	EFSA, “Opinion of the scientific panel on food additives, flavourings, processing aids and materials in contact with food (AFC) on a request from the commission related to flavouring group evaluation 13: furfuryl and furan derivatives with and without additional,” EFSA Journal, 215, 1-73, 2005.
	In article
[33]	JECFA, “Toxicological evaluation of certain food additives,” in The forty-fourth meeting of the Joint FAO/WHO ExpertCommittee on Food Additives and Contaminants, 1996.
	In article
[34]	Zaitzev, A. N., Simonyan, T. A., and Pozdnyakov, A. L., “Hygienic standardization of oxymethylfurfurol in food products,” Voprosy pitaniia, 1, 52-55, 1975.
	In article
[35]	Chávez-Servín, J. L., De La Torre Carbot, K., García-Gasca, T., Castellote, A. I., and López-Sabater, M. C., “Content and evolution of potential furfural compounds in commercial milk-based infant formula powder after opening the packet,” Food Chemistry, 166, 486-491, 2015.
	In article		View Article  PubMed
[36]	World Health Organization, “The WHO Child Growth Standards,” 2006. [Online]. Available: https://www.who.int/childgrowth/standards/en/. [Accessed: 15-Sep-2016].
	In article
[37]	Rufián-Henares, J. A. and Delgado-Andrade, C., “Effect of digestive process on Maillard reaction indexes and antioxidant properties of breakfast cereals,” Food Research International, 42 (3), 394-400, 2009.
	In article		View Article
[38]	Delgado-Andrade, C., Rufián-Henares, J. A., and Morales, F. J., “Fast method to determine furosine in breakfast cereals by capillary zone electrophoresis,” European Food Research and Technology, 221 (5), 707-711, October 2005.
	In article		View Article
[39]	Morales, V., Olano, A., and Corzo, N., “Ratio of Maltose to Maltulose and Furosine as Quality Parameters for Infant Formula,” Journal of Agricultural and Food Chemistry, 52 (22), 6732-6736, November 2004.
	In article		View Article  PubMed
[40]	Fenaille, F., Parisod, V., Visani, P., Populaire, S., Tabet, J.-C., and Guy, P. A., “Modifications of milk constituents during processing: A preliminary benchmarking study,” International Dairy Journal, 16 (7), 728-739, 2006.
	In article		View Article
[41]	Penndorf, I., Biedermann, D., Maurer, S. V., and Henle, T., “Studies on N-Terminal Glycation of Peptides in Hypoallergenic Infant Formulas: Quantification of α- N -(2-Furoylmethyl) Amino Acids,” Journal of Agricultural and Food Chemistry, 55 (3), 723-727, February 2007.
	In article		View Article  PubMed
[42]	Baptista, J. A. . and Carvalho, R. C. ., “Indirect determination of Amadori compounds in milk-based products by HPLC/ELSD/UV as an index of protein deterioration,” Food Research International, 37 (8), 739-747, 2004.
	In article		View Article
[43]	Damjanovic Desic, S. and Birlouez-Aragon, I., “The FAST index – A highly sensitive indicator of the heat impact on infant formula model,” Food Chemistry, 124 (3), 1043-1049, 2011.
	In article		View Article
[44]	Rufián Henares, J., Guerra Hernández, E., and García Villanova, B., “Evolución de compuestos furánicos durante el procesado y conservación de fórmulas enterales.,” Revista de tecnología e higiene de los alimentos, 324, 23-28, 2001.
	In article
[45]	Chávez-Servín, J. L., Castellote, A. I., and López-Sabater, M. C., “Analysis of potential and free furfural compounds in milk-based formulae by high-performance liquid chromatography: Evolution during storage,” Journal of Chromatography A, 1076 (1-2), 133-140, 2005.
	In article		View Article  PubMed
[46]	Pereyra-Gonzáles, A. S., Naranjo, G. B., Malec, L. S., and Vigo, M. S., “Available lysine, protein digestibility and lactulose in commercial infant formulas,” International Dairy Journal, 13 (2), 95-99, 2003.
	In article		View Article
[47]	DRIs, Dietary reference intakes for Energy, Carbohydrates, Fiber, Fat, Protein and Amino Acids (macronutrients), USA, 2002.
	In article
[48]	Contreras-Calderón, J., Guerra-Hernández, E., and García-Villanova, B., “Modification of nutritional value during storage of infant formulas elaborated with different intact and partially hydrolyzed proteins and carbohydrates,” VITAE, 22, 87-100, 2015.
	In article		View Article
[49]	Chávez-Servín, J. L., Castellone, A. I., and López-Sabater, M. C., “Evolution of available lysine and lactose contents in supplemented microencapsulated fish oil infant formula powder during storage,” International Journal of Food Science & Technology, 43, 1121-1128, 2008.
	In article		View Article
[50]	Evangelisti, F., Calcagno, C., Nardi, S., and Zunin, P., “Deterioration of protein fraction by Maillard reaction in dietetic milks,” Journal of Dairy Research, 66, 237-243, 1999.
	In article		View Article  PubMed
[51]	Pischetsrieder, M. and Henle, T., “Glycation products in infant formulas: chemical, analytical and physiological aspects,” Amino Acids, 42, 1111-1118, 2012.
	In article		View Article  PubMed
[52]	Eskin, M., Ho, C.-T., and Shahidi, F., “Browning reaction in food,” in Biochemestry of foods, Elsevier, Ed. 2013, 245-290.
	In article
[53]	Erbersdobler, H. F., Dehn, B., Nangpal, A., and Reuter, H., “Determination of furosine in heated milk as a measure of heat intensity during processing,” Journal of Dairy Research, 54, 147-151, 1987.
	In article		View Article
[54]	Birlouez-Aragon, I. et al., “Assessment of protein glycation markers in infant formulas,” Food Chemistry, 87, 253-259, 2004.
	In article		View Article
[55]	Capuano, E., Ferrigno, A., Acampa, I., Ait-Ameur, L., and Fogliano, V., “Characterization of the Maillard reaction in bread crisps,” European Food Research and Technology, 228, 311-319, 2008.
	In article		View Article