Toxic element levels in ingredients and commercial pet foods

Nowadays, there is a growing concern about contamination of toxic metals (TM) in pet food due to the great potential for health risks of these elements. TM concentrations in commercial pet foods (n = 100) as well as in ingredients used in their composition (n = 100) were analyzed and compared to the Food and Drug Administration (FDA) maximum tolerable level (MTL), and the TM concentrations found in the different sources of carbohydrate, protein, and fat were compared. The TM concentrations were determined by inductively coupled plasma with optical emission spectrometry (ICP-OES). Concentrations above the MTL for aluminum, mercury, lead, uranium, and vanadium were observed in both dog and cat foods, and the percentage of dog foods that exceeded the MTL of these TM were: 31.9%; 100%; 80.55%; 95.83%; and 75%, respectively, and in cat foods: 10.71%; 100%; 32.14%; 85.71%; 28.57%, respectively. The MTL values of these TMs and the mean values in dog foods (mg/kg dry matter basis) (MTL [mean ± standard deviation]) were: aluminum: 200 (269.17 ± 393.74); mercury: 0.27 (2.51 ± 1.31); lead: 10 (12.55 ± 4.30); uranium: 10 (76.82 ± 28.09); vanadium: 1 (1.35 ± 0.69), while in cat foods were: aluminum: 200 (135.51 ± 143.95); mercury: 0.27 (3.47 ± 4.31); lead: 10 (9.13 ± 5.42); uranium: 10 (49.83 ± 29.18); vanadium: 1 (0.81 ± 0.77). Dry foods presented higher concentrations of most TM (P < 0.05) than wet foods (P < 0.05). Among the carbohydrate sources, there were the highest levels of all TM except cobalt, mercury, and nickel in wheat bran (P < 0.05), while among the protein sources, in general, animal by-products had higher TM concentrations than plant-based ingredients. Pork fat had higher concentrations of arsenic, mercury, and antimony than fish oil and poultry fat. It was concluded that the pet foods evaluated in this study presented high concentrations of the following TM: aluminum, mercury, lead, uranium, and vanadium.

www.nature.com/scientificreports/ of Sao Paulo, Pirassununga-Brazil. Repetition was performed when the variation coefficient between samples was greater than 5.0%.
Sample preparation for the toxic metal's determination. The preparation of all samples (except samples of fat sources) was carried out by the wet method according to Pedrinelli 57 . For all samples, 0.5 g were weighed and placed in polypropylene tubes, and afterward, 1.5 mL of HNO 3 P.A. (65% m/v) (brand: Synth ® ) (Diadema, Brazil) and 2.0 mL of H 2 O 2 (30% m/v) (brand: Dinâmica ® ) (Indaiatuba, Brazil) were added to each tube. After 30 min, the volume was completed with 4.5 mL of ultrapure water type I (18.2 MΩ cm resistivity; conductivity: 0.054 µS/cm; TOC: < 5 PPB [< 5 µg/L]), obtained from a Milli-Q purification system (Millipore, USA). Then, the tubes were placed in a microwave oven (Multiwave GO, Anton Paar, Austria) and were subjected to heating in two phases: in the first, the samples were heated for 20 min until reaching 180 °C in 400 W; in the second phase, the samples were heated by 180 °C at 800 W for 10 min and, subsequently, cooling was performed for 10 min. After the digestion procedure, the samples were transferred to polyethylene tubes and ultrapure water type I (18.2 MΩ cm resistivity; conductivity: 0.054 µS/cm; TOC: < 5 PPB [< 5 µg/L]), obtained from a Milli-Q purification system (Millipore, USA), was added until they reached 25 mL of volume. Blank solutions were subjected to the same procedure to verify the quality of the reagents. The procedure was performed in duplicate. The samples' preparation by microwave digestion was carried out in the Laboratory of Biorigin Brazil (Lençóis Paulista, Brazil). The preparation of the fat samples was performed using the methodology adapted from Llorent-Martínez 58 . Acid digestion of samples was carried out in a microwave SW-4 model Speed Wave (Berghof, Germany). An aliquot of approximately 0.15 g of sample was weighed directly into the digestion vessel, then 5 mL of diluted HNO 3 (25% HNO 3 and 75% ultrapure water) was added to each oil sample. The HNO 3 used was P.A. (65% m/v) Corn gluten meal 21 6 Corn gluten meal 60 6 Mineral supplement 6 www.nature.com/scientificreports/ (brand: Synth ® ) (Diadema, Brazil), and ultrapure water used was a type I (18.2 MΩ cm resistivity; conductivity: 0.054 µS/cm; TOC: < 5 PPB [< 5 µg/L]), obtained from a Milli-Q purification system (Millipore, USA). Subsequently, the digestion procedure was performed in four stages of heating. First, the samples were heated for 5 min until reaching 90 °C at 700 W. In the second stage, the temperature was maintained at 90 °C for 3 min at 600 W.
In the third stage, the temperature was increased for 10 min to 170 °C at 600 W. In the last stage, the temperature was kept at 170 °C for another 7 min at 600 W. After this stage, the samples were cooled and ultrapure water was added until the volume reached 15 mL. Two blank solutions were included for every 18  For As, Hg, Sb, and Se determination a hydride generator (hydride ICP, Elemental Scientific, Omaha, USA) coupled to the ICP-OES was used. To avoid cross-contamination, ultrapure water type I (18.2 MΩ cm resistivity; conductivity: 0.054 µS/cm; TOC: < 5 PPB [< 5 µg/L]), obtained from a Milli-Q purification system (Millipore, USA) was used between the samples to clean the system and, for every five analyses of samples determined by the ICP-OES, the system was cleaned with 1.0 g/100 mL nitric acid P.A. (65% m/v) (brand: Synth ® ) (Diadema, Brazil). The calibration curves were prepared using multi-element solutions with a certificate of analysis and traceability to NIST (National Institute of Standards and Technology, Gaithersburg, MD, USA) of 100 mg/L for the elements Al, As, B, Ba, Be, Cd, Co, Cr, Fe, Hg, Ni, Pb, Sb, Se, Sn, and V, and monoelementary solutions (with a certificate of analysis and traceability to NIST) of 100 mg/L of U. The reference material used was SpecSol ® (Jacareí, Brazil) and was purchased from the company Quimlab (www. quiml ab. com. br). The metal concentrations of the reference material are traced to the following NIST standards: Al: NIST 928; As: NIST 3103a; B: NIST 3107; Ba: NIST 3104a; Be: NIST 3105a; Cd: NIST 928; Co: NIST 928; Cr: NIST 3112a; Fe: NIST 928; Hg: NIST 3133; Ni; NIST 928; Pb: NIST 928; Se: NIST 3149; Sb: NIST 136f; Sn: NIST 3161a; U: NIST 3164; V: NIST 3165. The curves were prepared with the aid of automatic pipettes and falcon tubes were used. Calibration curves were prepared one day before analysis. For As, Sb, Se, and Hg the calibration curves had the following points: 0.001-0.05-0.1-0.5-1-2 ppm. For the other metals analyzed the calibration curves had the following points: 0.001-0.05-0.1-0.5-1-2-5-10 ppm.
The emission line wavelengths of each element were:  Table 3.

Statistical analysis.
For comparisons between dry and wet food, dog and cat food, categories of dry foods, carbohydrate sources, and protein sources, the Shapiro-Wilk test was performed to assess the normality of the residues and the F test to verify the homogeneity of the variances. For data that did not show a normal distribution, the generalized linear mixed model with a logarithmic link function was used to stabilize the residues. After data transformation, ANOVA was performed and, when there was a difference between the groups, the Tukey test was performed. The analyses described above were performed using the Statistical Analysis System (SAS) software version 9.4 (SAS Institute, North Carolina, USA) and p values below 0.05 were considered significant.
The results found in commercial foods were compared with the MTLs established by the FDA 59 in mg/kg of dry matter, descriptively. For the elements whose values were not indicated by FDA 59 (B, Ba, and Sn), the MTL values of the most sensitive mammal to each of the elements were used, according to the Mineral Tolerance of Animals 60 and, for iron, the legal limit values of European Union legislation were used, according to regulation 1831/2003/EC, expressed in FEDIAF 61 , also in mg/kg of dry matter. www.nature.com/scientificreports/ For the elements that showed concentrations above the MTL of FDA 59 or legal limit expressed in FEDIAF 61 , to better investigate whether the values observed in food are a real health issue, it was calculated how much animals would consume of these elements per kg of body weight (BW) per day if they ate these analyzed foods. Thus, a simulation was carried out with dogs of different sizes (5 kg, 15 kg, 30 kg, and 50 kg) and, in the case of cats, simulations were carried out with variation in body weight between 3 and 5 kg. Simulations were performed with animals of different sizes, as the daily amounts consumed of the metals were calculated per kg of BW, not per kg of metabolic weight. Thus, the amount consumed per kg of BW differs, so that the smaller animals consume greater amounts when compared to larger animals.
The maintenance energy requirement for dogs was calculated using the equation 95 kcal × BW 0.75 , and for cats, through the equation 100 kcal × BW 0.67 . To calculate the daily amount of food to be provided, the metabolizable energy (ME) values declared on the labels of the analyzed foods were used and, for foods that did not contain this information stated on the label, the ME was estimated using the method described by the NRC 62 .
Ethical approval. The experimental protocol was conducted according to ethical principles in human and animal experimentation and was approved by the Commission on Ethics in the Use of Animals of the School of Veterinary Medicine and Animal Science of the University of Sao Paulo (protocol number 6717110219).

Results
Toxic metals concentrations in commercial pet foods. Through ICP-OES methodology, it was possible to determine 17 elements. In dog foods, the elements Sb, Ba, Cr, Sn, Fe, and Hg were detected in all samples (Table 4). In cat foods, the metals Cr, Hg, Sb, Fe, and Sn were detected in all samples analyzed (Table 5). In both dog and cat foods, the elements As, B, Be, and Se were below the detection limit of 0.05 mg/kg in all samples (Tables 4 and 5). Regarding the comparisons of the results with the MTLs by FDA 59 , in dog foods, values above the MTL were observed for the following elements: Al, Pb, Co, Hg, U, and V. In cat foods, values above MTL were observed for Al, Cr, Hg, Pb, U, and V. In addition, one food had Fe concentration above the legal limit.
As for comparisons between dog and cat foods, higher concentrations of Al (P = 0.0033), Ba (P = 0.0116), and Fe (P < 0.0001) in wet dog foods were observed, compared to wet cat foods, which presented higher U concentrations (P = 0.0331) ( Table 7). Dry dog foods had higher concentrations of Al (P < 0.0001), Ba (P < 0.0001), Co (P = 0.0104), and U (P < 0.0001), than dry cat foods, which presented higher concentrations of Fe (P < 0.0001) and Hg (P < 0.0001) ( Table 8). Regarding the comparisons between the different dog food categories, the standard  www.nature.com/scientificreports/ category had the highest concentrations of Ba (P < 0.0001), Fe (P < 0.0001), and U (P < 0.0001), while the superpremium category showed the highest concentrations of Al (P < 0.0001) ( Table 9). As for this same comparison in cat foods, the premium category showed the highest levels of U (P = 0.0084), and the super premium category presented the highest concentrations of Fe (P < 0.0001). Regarding Hg, the premium category showed higher concentrations than the super premium category (P = 0.0359), and the standard category did not differ from the others (Table 10).   www.nature.com/scientificreports/

Toxic metals concentrations in ingredients.
Regarding the comparisons between ingredients used as a carbohydrate source, wheat bran showed higher concentrations of Al, Ba, Cr, Fe, Pb, Sn, and U, compared to whole corn and broken rice (P < 0.05) ( Table 11). Whole corn showed the highest Ni concentrations (P = 0.0007). There was no difference between carbohydrate sources for the other elements analyzed (P > 0.05) ( Table 12). As for comparisons between protein sources, there was a difference between the ingredients for most of the elements analyzed (P < 0.05), except for Co (P > 0.05) ( Table 12). Regarding fat sources, only Al, As, Cd, Fe, Hg, Ni, Sb, and Sn were detected in samples of at least one type of ingredient. Pork fat had higher concentrations of As, Hg, and Sb compared to fish oil and poultry fat. There was no difference between fat sources in the concentrations of the other detected TMs (Table 13). In Fig. 5, the number of times that each type of ingredient exceeded the MTL for the elements found in concentrations above this limit in commercial pet foods is illustrated. The results of the toxic metal concentrations in the analyzed mineral supplements are shown in Table 14. The sample of dicalcium phosphate analyzed was the one that presented the largest number of elements above the detection limit, so that only Se and As were in concentrations below the detection limit.

Discussion
Of all the toxic metals analyzed, it is noteworthy that selenium and iron are considered essential, as they have known biological functions important for the maintenance of vital functions, therefore they must be present in the food so that only the excess is considered harmful. The main findings of this study regarding contamination by toxic metals in commercial pet foods were the high concentrations of Al, Pb, Hg, U, and V.
Aluminum. Concerning aluminum, in dog and cat foods, 31.9% and 10.71% had levels above the MTL, respectively. However, more than 75% of the analyzed foods exceeded the MTL less than 2.5 times. In a study conducted by da Costa 45 , which evaluated commercial foods for dogs and cats available in the Brazilian market, Al concentrations also determined by ICP-OES were above MTL in several pet foods. Fernandes 38 also observed high Al concentrations (determined by instrumental neutron activation) in commercial dog foods, which ranged from 58 to 11900 mg/kg food, which corresponds to 59.5 times the MTL. In a study conducted by Paulelli 46 , Al concentrations in commercial dog and cat foods ranged from 12 to 519 mg/kg food, and dog foods had higher Al concentrations than cat foods, which is consistent with the present findings. It is worth mentioning that there is little information in the literature regarding the Al toxicity in small animals, so the maximum intake that these animals can tolerate is not known. The MTL used in the present study as a parameter was extrapolated from the species most sensitive to this element, however, dogs and cats may be more tolerant to Al intake, consequently, a smaller proportion of the analyzed foods would be exceeding the aluminum tolerance for these species.
It is known that the Al absorption in the gastrointestinal tract is low (less than 1.0% in animals) and can be influenced by some factors such as solubility, pH, and chemical presentation 63 . In addition, it has been shown that the presence of citric acid in the diet results in increased Al absorption in humans 64 , as well as increased Al retention in tissues of laboratory animals [65][66][67][68] . This inclusion of citric acid could increase the absorption of www.nature.com/scientificreports/ aluminum present in food by the gastrointestinal tract of dogs and cats and, especially in foods with high concentrations of this element, it would increase the risk of adverse effects. In a study conducted by Katz 69 , four groups of Beagle dogs received sodium aluminum phosphate, in dosages of 0.0%, 0.3%, 1.0%, and 3.0% for 6 months. Complete blood count, biochemical analyses, and urine tests were performed, in which no changes suggestive of toxicity were observed. The authors concluded that the "No Observed Adverse Effect Level" (NOAEL) was 70 mg/kg BW/day. Pettersen 70 evaluated the effects of consuming high Al doses (0-1143 mg/kg BW for males; 0-1251 mg/kg BW for females) in Beagles. Toxicity was limited to an acute and transient decrease in food consumption and a concomitant decrease in body weight, observed only in males. There were no changes in serum biochemistry, variables obtained from blood counts, and urinalysis. In the present study, it was calculated how much animals would consume of Al per kg of BW per day if they were fed these diets, and it was observed that none of the analyzed foods provided the amount of 70 mg/kg BW/day. This suggests that the Al concentrations observed in this study do not imply risks of intoxication. Based on data from the present study, no food provided more than 45 mg/kg BW/day.
Regarding the Al concentrations in the analyzed ingredients, higher concentrations were found in the wheat bran, however, none of the samples of this ingredient exceeded the MTL for this element. The beef meal was the protein source with the highest Al concentrations, and 25% of the samples exceeded more than twice the MTL.    Charbonneau 71 with cats, in which different oral doses of methylmercury, the most toxic form of Hg, were tested and it was concluded that 0.02 mg Hg/kg BW/day did not cause adverse effects after 2 years. Thus, the FDA 59 estimated that for a dog or cat to consume this daily amount, the food must have a Hg content of 0.27 mg/kg dry matter, which was established as the MTL for Hg. Most foods in the present study, both for dogs and cats, provided amounts of mercury above 0.02 mg/kg BW/day. Among the adverse effects of high Hg intake, Charbonneau 71 observed ataxia, loss of balance, and motor incoordination in the group of cats that consumed 0.176 mg of Hg/kg BW/day after 14 weeks, and these same signs were observed after 40 weeks in the group that consumed 0.074 mg of Hg/kg BW/day. Histopathological findings demonstrated neurological injuries.
It should be noted that the Hg MTL was determined based on the methylmercury consumption, the most toxic form of Hg found mainly in aquatic organisms. In the present study, the methodology employed did not allow quantifying the different Hg forms present in the samples, only the total Hg concentrations. It is assumed, however, that methylmercury is not the predominant form in the analyzed foods, since this organometallic compound Table 7. Comparison of detected toxic metals concentrations (mg/kg) between wet foods for dogs and cats. SEM standard error mean, P probability of significance. a,b Means followed by different letters in the lines differed (P < 0.05).  www.nature.com/scientificreports/ www.nature.com/scientificreports/ is found in fish and other aquatic organisms, in which over 90% of the Hg is in the form of methylmercury 72,73 .
In most of the analyzed foods, fish by-products or other aquatic organisms were not present, according to what was stated on the labels. In addition, none of the samples of fish meal and fish oil analyzed in our study had Hg concentrations above the detection limit. This suggests that the chemical Hg form in the analyzed foods is not methylmercury, i.e., it may be a less toxic form, which in the concentrations found in this study, may not have adverse effects on the health of pets. However, this aspect needs to be further investigated. In our study, dry foods had higher Hg concentrations than wet foods, results opposite to those observed by Luippold 42 and Paulelli 46 . Concerning dry foods, those intended for cats had higher Hg concentrations than those found in dog food. This could be attributed to the greater inclusion of protein to meet the higher protein requirement of the feline species, since animal protein sources, more specifically beef meal and chicken by-products, were the ingredients with the highest Hg contamination and are widely used in dry pet food formulation.
As previously mentioned, no fish meal samples had Hg concentrations above the detection limit, although fish is considered the main Hg poisoning pathway in humans 74,75 . In the study conducted by Kim 44 , when determining toxic metals in commercial dry dog foods, the authors found higher Hg concentrations in food whose primary protein source was based on fish by-products, compared to those foods whose main protein source was by-products derived from red meat and poultry. A variety of factors can influence Hg contamination in fish, such as species, place of origin, whether it originates from pisciculture or the natural environment, feed, life span, and size 73 . It is known that fish meal can come from two sources, fishery by-products and fish caught exclusively for the production of this ingredient 76 . Therefore, all of these factors can influence Hg contamination in the fish meal used in pet food. Some studies have shown that fish from pisciculture has less Hg contamination than fish from natural environments 77,78 .  www.nature.com/scientificreports/ In Brazil, salmon (scientific name: Salmo salar) is widely used in fish meal employed by the pet food industry and, according to Ginsberg 79 , regarding the risk of Hg poisoning in humans, salmon is not worrisome. Furthermore, in the study conducted by Olmedo 80 , in which toxic metals were evaluated in several fish species, low Hg concentrations were found in salmon samples, of which the median and interval between 5 and 95th percentiles were 0 (0-0.004). This could explain the Hg concentrations below the detection limit in the fish meal samples analyzed in the present study. In addition, fish meal may come from waste from the tilapia filleting industry 81 , the most produced fish species in Brazil 82 , and low Hg concentrations in this species have been reported. In the study by Kitahara 74 , where samples of 11 different species of freshwater fish were analyzed, tilapia was the one with the lowest concentrations, ranging from 0.01 to 0.02 mg/kg. This could also justify the non-detection of mercury in the fish meal analyzed in our study.
In reference to the Hg contents in the analyzed carbohydrate sources, this toxic metal was not quantified above the detection limit in any whole corn sample, and wheat bran did not differ from broken rice. Among protein sources, in addition to fish meal, no Hg was present above the detection limit in any sample of corn gluten meal 60. Beef meal and chicken by-products meal showed higher concentrations, compared to other sources.
Other ingredients could be considered potential contaminants, such as broken rice since more than 50% of the samples exceeded the MTL more than ten times, as well as corn gluten meal 21, which had values up to 20 times higher than the MTL. These ingredients, in addition to having high Hg concentrations, can be included Of the analyzed ingredients, wheat bran was the carbohydrate source with the highest Pb levels, while the beef meal was the protein source that presented the highest concentrations of this toxic metal. In addition, values more than two times above the Pb MTL were observed in samples of chicken by-products meal, feather meal, and fish meal, which can contribute to the contamination of the final product if these ingredients are used in high quantities. The higher Pb levels found in beef meal corroborate the results found in the study performed by Kim 44 , higher Pb concentrations were observed in dog foods, in which red meat was the main protein source based on the list of ingredients stated on the label, compared to foods composed mainly of chicken and fish proteins. In that study, the maximum Pb concentration was also found in red meat-based foods, which was 270 times higher than the average human daily intake in units per megacalories estimated by Thomas 83 . The medians of the results found in the chicken-and fish-based foods exceeded the average daily human intake by 6 and 8 times, respectively.
As for the mineral supplements analyzed, high Pb concentrations were observed in the samples of calcium carbonate and dicalcium phosphate, which exceeded the MTL by 6.63 and 9.25 times, respectively. Although these ingredients are used in small quantities, compared to carbohydrate and protein sources, they could also contribute to the contamination of the final product. Environmental contamination by Pb is mainly due to the burning of fossil fuels (coal, natural gas, and oil) and the mining industry. In the past, tetraethyl lead was added to increase the gasoline octane rating, and the burning of this fuel was considered the main source of environmental contamination by Pb. Although this practice was banned in 1989, a large part of the Pb present in Brazilian soils is still attributed to tetraethyl lead 81 , which may justify the results observed in the present study.
With regard to the adverse effects of Pb excess, it is known that its toxicity may involve gastrointestinal signs 50,51 , neurological disorders 50,52 , damage to the hematopoietic system 53,54 , and kidney injuries 52 . Pb MTL was established based on a study in which there were no adverse effects related to the consumption of a diet with 10 mg/kg of Pb in dogs for a period of 2 years 84 . It is not known whether the Pb levels in the foods evaluated in the present study that exceeded the MTL can cause adverse effects, noting that three foods exceeded this limit by more than two times. In the study conducted by Steiss 85 , neuropathy or histological changes in the central nervous system were not observed in dogs that consumed an oral dose of 5 mg of Pb/kg BW/day for 40 weeks. Two dogs, however, showed evidence of non-regenerative anemia after 24 and 26 weeks of consumption. In our study, through the simulation of Pb consumption per kg of BW, the highest values observed were approximately 0.4 mg/kg BW for dogs and 0.5 mg/kg BW for cats, much lower than the dose studied by Steiss 85 , suggesting a Table 14. Toxic metals concentrations (mg/kg) of in the analyzed mineral supplements. SEM standard error mean, P probability of significance. a,b Means followed by different letters in the lines differed (P < 0.05). www.nature.com/scientificreports/ safety margin in the analyzed foods. In addition, according to Wismer 86 , the chronic accumulative toxic dose of Pb for dogs is 1.8-2.6 mg/kg BW/day, much higher than the amount estimated in our study.
Uranium. Of the analyzed dog and cat foods, 85.71% and 95.85% exceeded the MTL for uranium, respectively. In addition to this large portion having exceeded the MTL, the observed values exceeded this limit by up to 14 times. High U concentrations were also found in homemade foods for dogs and cats in the study performed by Pedrinelli 57 , in which 92% of the recipes for dogs and 100% of the recipes for cats exceeded the MTL, with values up to 16 times higher. As for the analyzed ingredients, wheat bran was the carbohydrate source that showed the highest U concentrations and, among the protein sources, higher U concentrations were found in the beef meal. Regarding the analyzed mineral supplements, high U concentrations were observed in the samples of calcium carbonate (281.24 mg/kg) and dicalcium phosphate (673.24 mg/kg), which corresponds to 28.12 and 67.32 times the MTL of that element, respectively.
In addition to the high concentrations observed in these ingredients, it is worth mentioning that most of the ingredients of animal and vegetable origin had concentrations much higher than U MTL, such as beef meal, of which all samples evaluated exceeded more than 17.5 times the MTL. All samples of fish meal, soybean meal, and wheat bran exceeded the MTL by more than ten times. In the case of chicken by-products meal, all samples exceeded more than 7.5 times the MTL. Due to the widespread use of these ingredients in the manufacture of commercial pet foods, these results may justify the high U concentrations observed in the commercial pet foods analyzed in this study.
Brazil has the fifth largest U reserve in the world, totaling 309,000 tonnes of this element, representing more than 5% of the total in the world 87 . This could justify the high U concentrations observed in the present study, since the exploration of uranium mines may result in contamination of water and soil and, consequently, of the ingredients used by the pet food industry. Furthermore, according to Prado 88 , contamination with uranium in foods occurs mainly as a consequence of its presence in phosphate rocks, from which fertilizers and mineral supplements used in animal feed, such as phosphate, are extracted. Of all the samples of ingredients analyzed, the sample of dicalcium phosphate showed the highest U concentration (623 mg/kg). The U contamination in plant-based products could be attributed to the use of phosphates as soil fertilizers, whereas the contamination of animal by-products could occur due to the consumption of phosphate by animals as a mineral supplement, as well as the consumption of vegetables contaminated by the continuous use of phosphate as a fertilizer.
The U concentrations in the analyzed foods are worrying, as this metal is considered one of the heaviest toxic elements present in the environment and is a precursor to natural radionuclides, therefore it emits alpha and gamma radiation 89 . In the study conducted by these latter authors, the effects of U consumption by growing dogs on bone U deposition were evaluated, as well as on clinical and histopathological changes. In that study, one Beagle consumed a diet containing 20 mg/kg of uranium and three other dogs of the same breed consumed a diet with 100 mg/kg of uranium for 279 days. On the histopathological examination, glomerular degeneration was observed in the three animals that consumed the food with the highest U content. However, no changes were observed in the animal that consumed the diet with 20 mg/kg of uranium. In our study, two cat foods and 12 dog foods had U concentrations above 100 mg/kg, which could imply risks of glomerular injury. However, it is worth noting that in the study by Arruda-Neto 89 , growing dogs were evaluated, which are perhaps more susceptible to the harmful effects of uranium, in addition to consuming a larger amount of food (consequently more uranium) per kg of metabolic weight, due to the greater energy and nutritional requirements in this stage of life.
It is noteworthy that for the determination of MTL, due to the lack of information in the literature regarding the U toxicity to dogs and cats, the FDA 59 used the MTL of the most sensitive mammal to this toxic metal, which in this case are rodents, and reduced this value by ten times, as suggested when extrapolating between species. According to Vicente-Vicente 90 , when comparing dogs and rodents with respect to sensitivity to U inhalation, less sensitivity was observed in dogs. This suggests that perhaps dogs are also less sensitive to U intake than rodents, in this case, there would be no need to reduce the U MTL by ten times when extrapolating rodents to dogs. Therefore, a smaller portion of the analyzed foods would be above this limit. Concerning cats, there is no study to date that has investigated the effects of U consumption or inhalation.
Vanadium. As for V, 75% and 28.57% of dog and cat foods exceeded the MTL for this element, respectively. However, the majority exceeded less than two times that limit. Due to the lack of information regarding the V toxicity in dogs and cats, the MTL proposed by the FDA 59 was the V MTL for the mammal known to be more sensitive to this element, divided by 10 (safety factor). Therefore, it is possible that dogs and cats are less sensitive and, in this case, a smaller portion of the samples would have exceeded the MTL. No food exceeded the V MTL for the most sensitive mammal (10 mg/kg) in the present study.
The highest toxic metals concentrations, in general, in animal-based ingredients than plant-based ingredients could be attributed to the potential for the accumulation of these elements in the animal organism. According to NRC 60 , some elements can accumulate in animal tissues, such as bone tissue, skeletal muscle, liver, kidneys, and spleen, as well as in animal products such as milk, reaching critical concentrations that may imply adverse effects on the health of human beings who consumes these products. Also according to NRC 60 , this could happen even in situations in which the levels of these minerals in animal feed are below the MTL for the respective species (considered safe). Therefore, it is possible that food animals, such as broilers and beef cattle, consume a diet with plant-based ingredients contaminated with toxic metals in addition to supplements with a large contamination potential, such as dicalcium phosphate, and accumulate these metals in various body tissues, which results in high contamination of their by-products, such as poultry by-products meal and meat meal. The animal tissues www.nature.com/scientificreports/ mentioned above are part of the composition, in general, of animal-based ingredients used in the formulation of pet foods, which perhaps justifies the greater contamination of these by-products.

Conclusion
The analyzed foods presented high concentrations of the following elements: Al, Cu, Hg, Pb, U, V, and Zn. In general, animal-based ingredients have a greater potential for contamination than plant-based ingredients. Further studies are needed to assess the effects of chronic ingestion of the elements mentioned above in the quantities found in this study and if under the same circumstances, these toxic metals pose risks to the dogs' and cats' health.
Regarding the concentrations of the elements As, B, Ba, Be, Cd, Co, Sn, Sb, Ni, and Fe, in general, values above the respective MTLs and/or legal limits were not observed, both in commercial foods and in the ingredients analyzed in the present study, so it is not considered that there are intoxication risks with these elements in small animals through long-term consumption of the analyzed foods. It is noteworthy that it is necessary to establish legal limits for all metals with toxic potential.

Data availability
The datasets generated during and/or analyzed in the current study are available from the corresponding author on reasonable request.