*Article* **The Removal of Meat Exudate and** *Escherichia coli* **from Stainless Steel and Titanium Surfaces with Irregular and Regular Linear Topographies**

**Adele Evans <sup>1</sup> , Anthony J. Slate <sup>2</sup> , I. Devine Akhidime 1,3, Joanna Verran <sup>1</sup> , Peter J. Kelly <sup>1</sup> and Kathryn A. Whitehead 1,3,\***


**Abstract:** Bacterial retention and organic fouling on meat preparation surfaces can be influenced by several factors. Surfaces with linear topographies and defined chemistries were used to determine how the orientation of the surface features affected cleaning efficacy. Fine polished (irregular linear) stainless steel (FPSS), titanium coated fine polished (irregular linear) stainless steel (TiFP), and topographically regular, linear titanium coated surfaces (RG) were fouled with *Escherichia coli* mixed with a meat exudate (which was utilised as a conditioning film). Surfaces were cleaned along or perpendicular to the linear features for one, five, or ten wipes. The bacteria were most easily removed from the titanium coated and regular featured surfaces. The direction of cleaning (along or perpendicular to the surface features) did not influence the amount of bacteria retained, but meat extract was more easily removed from the surfaces when cleaned in the direction along the linear surface features. Following ten cleans, there was no significant difference in the amount of cells or meat exudate retained on the surfaces cleaned in either direction. This study demonstrated that for the *E. coli* cells, the TiFP and RG surfaces were easiest to clean. However, the direction of the clean was important for the removal of the meat exudate from the surfaces.

**Keywords:** *Escherichia coli*; bacterial retention; surface topographies; meat exudate; wipe cleaning; conditioning film

#### **1. Introduction**

Modern food processing and production facilities provide an environment that promotes bacterial retention due to a myriad of factors, which include the surface properties of the equipment and the matrix of the food being processed [1,2]. The removal of bacteria and/or organic material from food production surfaces is important since its build up can result in microbial contamination of food products, which can have a significant effect on consumers, food companies, and food suppliers, for example, cross-contamination of food with pathogenic bacteria can result in food-borne illnesses [3–6]. The Food Standard Agency estimates that foodborne illness in the UK alone result in a financial loss of £1.5 billion per annum [7]. As such, biofouling in the food industry is a significant problem [8]. For certain bacteria, some of which are important human pathogens, there can be contamination of raw meat due to biofouling [9]. Contamination of beef with *Escherichia coli* O157:H7 has been linked to outbreaks of foodborne illnesses and concerns about *E. coli* O157:H7 contamination have resulted in a zero tolerance towards this microorganism in the food industry [10,11].

**Citation:** Evans, A.; Slate, A.J.; Akhidime, I.D.; Verran, J.; Kelly, P.J.; Whitehead, K.A. The Removal of Meat Exudate and *Escherichia coli* from Stainless Steel and Titanium Surfaces with Irregular and Regular Linear Topographies. *Int. J. Environ. Res. Public Health* **2021**, *18*, 3198. https://doi.org/10.3390/ijerph18063198

Academic Editor: Ki Hwan Park

Received: 18 February 2021 Accepted: 15 March 2021 Published: 19 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Once a surface is introduced into an environment, it will adsorb a variety of organic and inorganic matter, resulting in the formation of a conditioning film [12–14]. Surface conditioning is a process which starts within seconds of the substratum becoming immersed into liquids [15]. The structure and composition of a conditioning film is ultimately dependent upon the surrounding products and the properties of the surface and can result in physicochemical, chemical, and topographical alterations, affecting both the rate and extent of bacterial retention and therefore surface contamination [16–18]. With regards to the meat industry, the exudate of frozen raw meat has been identified as an important source of bacterial contamination on food processing surfaces [19]. It has also been shown that sterilized chicken juice is an ideal environment for survival of *Campylobacter jejuni* [20] and its presence may also increase biofilm formation [21].

The method and type of physical cleaning methods used will be dependent on the food industry and the surfaces involved [22]. Product contact surfaces may typically be cleaned several times per day, while environmental surfaces such as walls and hoods may be cleaned less frequently [23]. In the meat industry, chilled beef carcasses are cut into smaller pieces, which are deboned and made into cuts; such work takes place on flat surfaces that are regularly cleaned [24]. However, it has been suggested that bacterial recontamination during this meat fabrication process results in higher numbers of *E. coli* on the cuts and trimmings [25,26]. Hence, a better understanding is required of the mechanisms involved in the attachment and detachment of bacteria to meat processing surfaces and their removal following cleaning. To simulate more realistic conditions, cleaning assays need to be carried out in the presence of a meat exudate (or relevant conditioning film) to increase the understanding of surface hygiene and decrease transmission and hence potential public health risks [27].

The ideal conditions for a hygienic surface have been defined as easy to clean, able to resist wear and maintain their hygienic qualities over time [28]. The hygienic quality and cleanability of a surface has been linked to the surface properties including the topography [28–30], chemical composition [31] and physicochemical properties [32,33]. Thermodynamics are thought to play a central role in initial bacterial: substrata interactions where it has been suggested that bacterial cells will attach preferentially to hydrophobic materials (i.e., materials with a low surface energy), when the surface energy of the bacteria is greater than the surface energy of the surrounding liquid [34]. Due to the complexity of bacterial-substratum interactions, further research is required to fully elucidate the underpinning mechanisms of bacterial attachment, adhesion, and retention [35].

An approach to reduce microbial contamination, which is a prerequisite for biofilm formation, is the modification of surface topography. Microscale surface topographic features have been shown to both inhibit or promote bacterial retention depending on the size, shape, and density of the topographical features [36]. It has also been shown that surfaces with features on the same scale as bacterial cells (e.g., cocci-shaped *Staphylococcus aureus*; ~1 µm diameter) promote the strongest retention due to maximum binding at the cell-substrate contact areas [37,38]. In an industrial setting, the wear of the surfaces may introduce random features (i.e., scratches) of different dimensions and it has been suggested that an increase in the surface roughness may cause the entrapment of microorganisms within the surface features, which in turn will affect the cleanability and hence the hygienic status of the surface [39]. Bacteria and organic material that become entrapped in the topographical features of a surface are difficult to remove using standard cleaning procedures [40], and it has been proposed that the development of the micro-pattern materials may help in the reduction of viable bacteria on food contact surfaces [41]. However, most studies have not determined the effect of the presence of the conditioning film on surface cleaning, especially with regards to the influence of surface topographical features [20], or with regards to the direction of cleaning compared to the linear surface features.

Stainless steels are used widely throughout the food and beverage industry due to their resistance to corrosion, thermal conductivity, and their ability to be produced with a smooth surface finish [33]. Stainless steel grade 304 is most commonly used in the food

industry [42]. Due to the production process of stainless steel, 'microniches' of heterogenous chemical composition may result in varying bacterial retention patterns [43]. Titanium has been incorporated into stainless steel alloys in the food industry to improve corrosion resistance because it forms stable carbides [44,45]. Titanium surfaces may also have a more homogenous chemical composition than stainless steel since it is comprised mainly of TiO<sup>2</sup> [46]. This work aimed to determine how surface attributes (chemistry and topography) and the direction of cleaning affected bacteria and meat exudate removal from surfaces.

#### **2. Materials and Methods**

#### *2.1. Equipment and Material Suppliers*

The following reagents and materials were used; stainless steel sheets (Outokumpu Stainless Ltd., Helsinki, Finland), sodium hydroxide, di-potassium hydrogen phosphate, potassium di-hydrogen phosphate, tri-sodium citrate ammonium sulphate, magnesium sulphate (Merck, Darmstradt, Germany), tryptone soya agar and tryptone soya broth (Oxoid, Basingstoke, UK) rolled beef brisket (Co-op, Manchester, UK), *Escherichia coli* CCL410 (Agence Francaise de Securite Sanitaire des Aliments, Paris, France), cleaning clothes (WYPALL® <sup>×</sup>80 Kimberley-Clark, West Malling, UK), Rhodamine B, DAPI and glycerol (Merck, Darmstradt, Germany). The following equipment was purchased: Atomic force microscope (Quesant Instruments, Santa Cruz, CA, USA), Crockmeter (A.A.T.C.C Crockmeter, Model CM1, NC, USA), Epifluorescence microscope (Nikon, Tokyo, Japan), F-View II camera (Soft Imaging System Ltd., Olympus, Tokyo, Japan), and Cell F Image Analysis package (Olympus, Tokyo, Japan).

#### *2.2. Production of Surfaces*

Three different surfaces were used in this study, including stainless steel 304 with a fine polished finish (FPSS), 304 fine polished finished stainless steel coated with titanium (TiFP) and a linear, regular finished (RG) titanium surface. Fine polished, grade 304, stainless steel sheets were prepared as 10 mm × 10 mm sample squares using a guillotine. To ensure that the samples were examined in a pristine "as-manufactured" state, the manufacturer's protective plastic coating was only removed directly before experimentation.

The titanium surfaces with a regular topography were unwritten digital video discs stripped of their protective coats. The samples were cut into 10 mm × 10 mm squares using metal cutting shears and soaked overnight in 30% sodium hydroxide solution, followed by rinsing thoroughly with sterile distilled water and drying in a class 2 microbiological cabinet prior to coating with titanium.

Samples of the fine polished stainless steel surfaces and the stripped digital video discs were coated using titanium. The substrata were coated with titanium via magnetron sputtering in a modified Edwards E306A coating system rig using a single 150 mm diameter × 10 mm thick, 99.5% pure titanium target attached to an unbalanced magnetron (argon gas at a working pressure of 0.15 Pa; magnetron power of 0.5 kW; base pressure 10−<sup>4</sup> Pa; time 15 min; substrate biased at −50 V) [47].

#### *2.3. Atomic Force Microscopy (AFM)*

The shape and depth of the surface features was determined using atomic force microscopy. The analysis was carried out in in contact mode using triangular shaped silicon nitride tips, with a spring constant of 0.12 N m−<sup>2</sup> . The height and shape of the features were determined from five areas taken from different replicate surfaces.

#### *2.4. Sample Organisms*

This study was conducted with *Escherichia coli* strain CCL410. This strain was recovered by the laboratory of Dr C. Vernozy-Rozand (Unité de Microbiologie alimentaire et prévisionnelle, Ecole vétérinaire de Lyon, France) from heifers fecal samples. This strain was selected due to it being a non-pathogenic variant of *E. coli* O157:H7 (wild type strain). The pathogenicity of the bacteria was reduced due to the loss of *stx1* and *stx2* [48].

#### *2.5. Bacterial Stock and Working Cultures*

Stock cultures of *E. coli* were stored at −80 ◦C in a freezer mix, which was composed of a sterilised salt solution containing a mixture of autoclaved 12.6 g L−<sup>1</sup> di-potassium hydrogen phosphate, 3.6 g L−<sup>1</sup> , potassium di-hydrogen phosphate, 0.9 g L−<sup>1</sup> , tri-sodium citrate 1.8 g L−<sup>1</sup> ammonium sulphate and 300 g L−<sup>1</sup> glycerol combined with a litre sterilised solution of 1.8 g L−<sup>1</sup> magnesium sulphate [49]. In preparation for the cleaning assays, cultures of *E. coli* were prepared by inoculating *E. coli* onto Tryptone soya agar (TSA), at 37 ◦C overnight. A single colony of *E. coli* was inoculated into 10 mL of Tryptone soya broth (TSB) and incubated at 37 ◦C overnight. One hundred microlitres of overnight culture was inoculated into 100 mL TSB and incubated at 37 ◦C for 18 h with shaking (200 rpm). Following incubation, the bacterial cells were harvested by centrifuging at 1721× *g* for 10 min, washed once, and re-suspended in sterile distilled water using a vortex mixer for 30 s. The suspension was centrifuged at 1721× *g* for 10 min and the cells were resuspended to an optical density (OD) of 1.0 (±0.1) at 540 nm in sterile distilled water. This corresponded to ca. 1.88 <sup>±</sup> 0.22 <sup>×</sup> <sup>10</sup><sup>8</sup> CFU mL−<sup>1</sup> .

#### *2.6. Meat Exudates*

The production of meat exudates was adapted [50]. Commercially available, fresh rolled beef brisket was cut into 10 mm × 10 mm pieces, placed in a stainless steel tray and covered in aluminium foil. The meat was covered by another stainless steel tray and weighed down with 8.4 kg of stainless steel sheets and frozen at −20 ◦C for 24 h. The diced meat pieces were defrosted at room temperature, and the meat exudate produced was collected and stored at −20 ◦C until use.

#### *2.7. Cleaning Assays*

The substrata were inoculated with a bacterial/meat exudate mixture and dried in a microbiological class 2 cabinet. For the bacterial/meat exudate mixture, 100 µL of bacteria and 100 µL of meat exudate was placed into an Eppendorf tube, vortexed for 5 s and 10 µL of the preparation was pipetted onto the substratum, spread across the surface with a sterile plastic spreader, and dried in a class 2 flow hood at room temperature. A crockmeter was used for the wipe clean method to ensure that each wipe across the stainless steel surface was standardised. The substrata were placed on the steel specimen stage and a 45 mm × 45 mm piece of blue wipe cloth was folded and attached to the 16 mm diameter test finger. Sterile distilled water (1 mL) was pipetted onto the cloth and the hand crank was turned to simulate one wipe. The wipe cycles compromised one, five, or ten repeats. Following each cleaning cycle, the substrata were dried for 2 h in a class 2 microbiological cabinet. Three replicates were taken at each cleaning cycle point for each surface, and for each direction of clean (along or perpendicular to the linear features).

Following the cleaning assays, the percentage coverage of the bacteria and meat extract retained on the surfaces per field of view was analysed following differential staining and epifluorescence microscopy.

#### *2.8. Preparation of Stains*

[9-(2-carboxyphenyl)-6-diethylamino-3-xanthenylidene]-diethylammonium chloride (Rhodamine B) was prepared as a stock solution of 0.1 g mL−<sup>1</sup> in ethanol (absolute) and used at a working concentration of 0.1 mg mL−<sup>1</sup> . 40 , 6-diamidino-2-phenylindole (DAPI) was prepared as a stock solution of 0.3 g mL−<sup>1</sup> in sterile distilled water and used at a working concentration of 0.1 mg mL−<sup>1</sup> . Prior to use, the stains were refrigerated (4 ◦C) and stored in a dark environment.

#### *2.9. Differential Staining of Meat Exudate and E. coli*

A dual staining procedure was conducted as described previously [51]. Ten microlitres of DAPI was added to the samples and spread across the surface using a sterile plastic spreader to detect the bacteria and then 10 µL of Rhodamine B was applied to the substrate

in the same manner to detect the retained meat extract [51]. Following staining, the samples were dried in the dark at room temperature in a microbiological class 2 flow hood.

The samples were viewed, and images obtained using an epifluorescence microscope with black and white digital camera and a Cell F Image Analysis package to measure the percentage coverage of the area of the stained material and to determine the percentage surface coverage of the bacteria and organic material. A filter wavelength of 330–380 nm was used to detect the DAPI stained cells, and a 590–650 nm filter was used to detect the Rhodamine B stained organic material. The retained material on the surfaces was measured using percentage coverage of the field size for randomly selected areas across the test substratum. Each of the three samples had 15 areas independently selected and analysed for the percentage coverage of bacteria and meat extract (*n* = 45).

#### *2.10. Statistical Analysis*

Statistical analysis was conducted by performing two-way ANOVA coupled with Tukey's multiple comparison tests for post hoc analysis using GraphPad Prism (version 8.4.2; GraphPad Software, San Diego, CA, USA) to determine significant differences at a confidence level of 95% (*p* < 0.05). Error bars represent the standard error of the mean. Asterisks denote significance, \* *p* ≤ 0.05, \*\* *p* ≤ 0.01, \*\*\* *p* ≤ 0.001, and \*\*\*\* *p* ≤ 0.0001.

#### **3. Results**

Three surfaces were prepared to determine the effect of a linear surface topography (irregular and regular), and defined surface chemistry (stainless steel and titanium) on the removal of bacterial and meat exudate using a wipe clean assay. Atomic force microscopy (AFM) of the fine polished stainless steel (FPSS), titanium coated fine polished stainless steel (TiFP), and the regular linear featured titanium coated surface (RG) revealed that the surface features of the FPSS and TIFP surfaces demonstrated irregular, linear topographies. The Z height of the TiFP surface (Figure 1b) was higher than the FPSS surface (0.338 ± 0.017 µm and 0.284 ± 0.014 µm, respectively) (Figure 1a). Regular linear features were evident on the titanium coated surface (RG) and the z height of the titanium coated regular surface was 0.420 ± 0.021 µm. The FPSS demonstrated valley widths of ~1 µm to 5 µm, whilst the TiFP demonstrated valley demonstrated valley widths of ~0.5 µm to 5 µm. The RG surface demonstrated valley widths of 1.02 µm. The contact angles of the three surfaces were 82 ± 3 ◦ , 84 ± 4.5◦ , and 91 ± 3.7◦ for the FPSS, TiFP and RG surfaces, respectively, and this indicated that the FPSS and TiFP were marginally more wettable than the RG surface.

The percentage coverage of the bacteria on the surfaces following initial fouling of the substrata before cleaning demonstrated that cells were retained in significantly higher amounts of bacteria on the FPSS (15.86%) or TiFP (18.52%) compared to the linear finished RG surface (0.81%) (*p* < 0.0001) (Figure 2).

Following one clean, fouling of the surfaces with different features and chemistries (FPSS, TiFP, RG), the amount of bacteria when cleaned along the linear features was significantly reduced (FPSS 6.98%; TiFP 1.91%; RG 0.17%) (*p* < 0.0001) (Figure 2a), whereas following one clean in the direction perpendicular to the linear features, there was only a significant difference in the amount of cells removed from the FPSS and RG surfaces (FPSS 5.49%; TiFP 1.51%; RG 0.21%) (*p* > 0.05) (Figure 2b). After five or 10 cleans, there was no significant difference in the amount of bacteria retained when the surfaces was cleaned along or perpendicular to the surface features (FPSS 6.98%, 5.49%; TiFP 1.91%, 1.51%; RG 0.17%, 0.21%) (*p* > 0.05). Overall removal of the cells from the surfaces in the direction of the linear features or perpendicular to the linear features demonstrated the same trend whereby the FPSS surface retained more bacteria than the TiFP surface, and the lowest amounts of bacteria was retained on the RG surface (Figure 2a,b).

**Figure 1.** Atomic force microscopy (AFM) demonstrating the (**a**) fine polished stainless steel (FPSS), (**b**) titanium-coated fine polished stainless steel (TiFP), and (**c**) titanium coated regular linear featured surface (RG) (*n* = 15).

Detection of the meat exudate on the surfaces following the initial application demonstrated no significant differences in the amount of conditioning film retained on the different surfaces (FPSS, 76.2%; TiFP, 76.67% and RG, 83.20%) (*p* > 0.05) (Figure 3). The meat exudate was increasingly removed from the surfaces with increased number of cleans and this was evident for all surface types (Figure 3a,b). Following one and five cleans, there was a significant difference in the amount of meat exudate removed from the surfaces when cleaned along the linear features (*p* < 0.0001) and perpendicular to the linear features (*p* > 0.05). There was no significant difference in the amount of meat exudate retained on the different surfaces after ten cleans along (FPSS 1.4%, TiFP 0.7%, RG 0.9%), or perpendicular to (FPSS 3.6%, TiFP 1.5%, RG, 1.6%) the linear features (*p* > 0.05). However, when cleaned along the linear surface features, the overall trend was that most of the meat exudate was retained on the FPSS > TiFP > RG surface demonstrating the same trend as the removal of cells. When cleaned in the direction perpendicular to the linear features, the amount of meat exudate retained on the surfaces did not follow the same trend (one clean, FPSS > TiFP > RG; five cleans, TiFP > FPSS > RG; ten cleans, FPSS > RG > TiFP).

**Figure 2.** Percentage coverage bacteria retained on fine polished stainless steel (FPSS), titanium polished stainless steel (TiFP) and the regular linear featured titanium coated surface (RG) surface following 0, 1, 5 and 10 cleans (**A**) along the direction and (**B**) perpendicular to the surface features (*n* = 45). Asterisks denote significance, \* *p* ≤ 0.05, \*\* *p* ≤ 0.01 and \*\*\*\* *p* ≤ 0.0001.

**Figure 3.** Percentage coverage of meat exudate retained on fine polished stainless steel (FPSS), titanium polished stainless steel (TiFP) and the regular linear featured titanium coated surface (RG) surface following 0, 1, 5 and 10 cleans (**A**) along the direction and (**B**) perpendicular to the surface features (*n* = 45). Asterisks denote significance, \* *p* ≤ 0.05, \*\* *p* ≤ 0.01 and \*\*\*\* *p* ≤ 0.0001.

The amount of bacteria and meat exudate removed from the surfaces following cleaning along linear features compared to cleaning in a perpendicular direction to the linear features, demonstrated that there was no significant difference (*p* > 0.05) in the removal of cells (with the exception five cleans on the FPTi). However, the meat exudate demonstrated a different trend whereby by ten cleans, the meat exudate was significantly more removed when the surfaces were cleaned in the direction along the surface features (*p* < 0.05). This result may have occurred due to the size of the bacterial cells and organic components of the meat exudate with respect to the size of the surface features (Figure 4).

**Figure 4.** Meat exudate (red) and *E. coli* cells (blue) remaining on titanium coated fine polished stainless steel surface (TiFP) following (**a**) pre-cleaning procedure, (**b**) one wipe clean along, (**c**) five wipe cleans along, (**d**) ten wipe cleans along and (**e**) one wipe clean across, (**f**) five wipe cleans across and (**g**) ten wipe cleans across. Scale bar: 20 µm. Differential staining was conducted to visualise bacterial cells and the meat exudate on the surfaces and an example of the images on the TiFP surfaces is demonstrated (Figure 5). Prior to the cleaning procedure, the meat exudate (red) and bacterial cells (blue) can be observed in abundance (Figure 5a). The concentration of organic material and *E. coli* declined as the number of wipe cleans increased, both in the direction of, and perpendicular to the linear surface features (Figure 5).

**Figure 5.** Schematic demonstrating how (**a**,**b**) the size of the bacteria (cylindrical) and (**c**,**d**) meat exudate (circles) influenced the efficacy of cleaning in the (**a**,**c**) direction of cleaning along the linear surface features or (**b**,**d**) in a direction perpendicular to the surface features.

A schematic representation of the bacteria retained, and the effect of the surface topography was produced. The bacteria were initially retained in higher amounts on the irregularly polished surfaces (FPSS and TiSS) being entrapped in the irregular surface features (Figure 6a). However, on the surfaces with regular surface features, the bacteria sat on the top, rather than inside the surface features (Figure 6b). This resulted in a lower binding of the bacteria on the surfaces and less bacterial retention.

**Figure 6.** The bacteria retained on the surfaces were bound in the highest amounts on the surfaces with (**a**) irregular topographies rather than on (**b**) topographically regular surfaces.

#### **4. Discussion**

Product contact surfaces may contaminate meat products directly with microbial or organic material contaminants [23]. The properties of a surface play a pivotal role in bacterial and organic material retention, but nevertheless, the way in which the substrata can mediate such binding remains unclear [28,52,53].

The surfaces with regular surface topographies demonstrated clearly defined features or regular size, shape, depth, and periodicity. However, surfaces with irregular topographies, such at the fine polished stainless steel (FPSS) and the titanium coated fine polished stainless steel (TiFP), contained features of different sizes with irregular frequencies, which dependent on their size may contribute to increased or decreased bacterial binding. In this study, more cells were retained initially on the topographically irregular surfaces, which suggests that these irregular features enhanced the bacterial cell: surface interaction. The amount of bacteria retained on the irregular surface features was higher following the initial inoculation and cleans. In agreement with these findings, micropatterned topography films were utilised to determine the attachment and survival of *Escherichia coli* and *Listeria innocua* and it was demonstrated that initial bacteria attachment to the micro-pattern topography films were significantly lower in the short term [41]. In addition, after incubation with a methicillin-resistant *Staphylococcus pseudintermedius,* it was determined that bacterial biofilms tended to form in crevices [54]. However, such findings were carried out using retention and biofilm assays and were not subject to cleaning or physical forces.

Throughout this study, bacterial retention and meat exudate (e.g., the conditioning film) was quantified via differential staining and epifluorescence microscopy. The samples were prepared by adding DAPI and Rhodamine B directly to the surface and spread across the surface and dried. Although it may be considered that the methodology used in the staining method may affect the distribution of the retained material, previous studies in our laboratories have demonstrated that this is not the case since the material retained is dried onto the surface and is extremely well retained [51]. In addition, all the samples in this study were prepared using the same method; any effect which may be due to the staining process is negligible. In order for epifluorescence microscopy to be utilised effectively, samples must be prepared in a consistent manner, as was the case in this study [55].

Surfaces with features of microbial dimensions similar to those of microbial cells have been shown to promote bacterial binding, whilst the morphology of the bacterial cell can also influence such mechanisms [37,38]. All the surfaces used in this study contained surface topographies with microbial dimensions. The findings in this research demonstrated that surfaces with periodically regular dimensions decreased bacterial retention regardless of the direction of clean and removed the greatest amount of meat exudate following cleaning along the linear surface features. Although features of microbial decisions may readily retain bacteria, when a physical force is applied, it may be that the shape of the topographical feature is of importance, with the periodic regularity of the surface combined with the cell size enabling the bacteria to be easily rolled across the surface. Thus, in the context of cleaning, surface with regular topographies may enhance surface hygiene following cleaning procedures.

In addition to the surface topography, the surface chemistry may affect bacterial retention. The results demonstrated that the bacteria and meat exudate were retained in lower amounts and coverage on the titanium surfaces. In agreement with our findings, Jeyachandran et al. (2007) demonstrated that a titanium oxide film retained fewer bacteria than other materials [56]. Furthermore, Ma et al. (2008) demonstrated that the heterogeneous chemistry of a surface may provide specific contact points for bacterial retention; such points may be found on stainless steel surfaces [43]. Hence, the more homogeneous surface chemistry of the titanium coating may have resulted in a reduced number of chemically different sites, resulting in lowered bacterial and meat exudate retention. Surface wettability can interact with other surface parameters, resulting in preferential or disadvantageous bacterial retention [57,58]. In the current study, the FPSS and the TiFP surfaces were more wettable than the RG surfaces. However, the bacteria and meat extract were deposited directly onto the surfaces and hence the physicochemical effects may have been negated.

The processing of meat products results in high level of organic material remaining on food contact surfaces which conditions the underlying substrata, and it is onto the proteinaceous conditioning film to which the bacteria become retained [9,27]. It has been demonstrated that the attachment of *Pseudomonas fragi* to beef resulted in the bacteria

becoming entrapped within the collagen fibres of the raw meat [59]. It has been suggested that contamination of the meat product by bacteria could be transferred to a surface, therefore thorough cleaning of surfaces and meat residues during meat production is critical to reduce the bacterial load [60]. The results from this study demonstrated that all the surfaces retained similar levels of meat extract initially, but following cleaning, the meat exudate was more difficult to remove from the surfaces with the irregular topographies. When the surfaces were cleaned in the direction along the surface features, the meat exudate was also easier to remove from the titanium coated regular surface (RG) than the titanium coated irregular surface (TiFP) or the stainless steel (FPSS). However, a clear trend on the effect of the surface properties, on the amount of meat exudate removal was not demonstrated when the surfaces were cleaned perpendicular to the linear features. Although only small amounts of organic material were retained, the difference in the trends in the effects of the surfaces properties on meat exudate retention may be due to the composition of the meat exudate, which will consist of much smaller molecules than the bacterial cells. It may be that although the bacteria can be removed by the physical force due to their larger size, the smaller organic molecules can only be pushed out of the linear features when cleaned in the direction along the linear features, as this will offer little resistance. In contrast, when the cleaning action is perpendicular to the surface features, the organic material is pushed against the wall of the surface feature where it becomes retained. This may explain the differences observed in the results.

By ten cleans, the surfaces demonstrated similar amounts of bacteria and meat exudate retained on the surfaces. One of the reasons for this is that a key component of the meat exudate is protein [50]. Protein adsorption on surfaces is a major issue in the food industry and the adsorption of proteins onto surfaces is a complex phenomenon influenced by many factors [61,62]. Protein adsorption to a surface occurs due to a range of forces and will continue until a state of equilibrium occurs [63]. It may be that as the number of cleans increased a state of equilibrium of the protein binding that occurred on the surfaces, masking the original surface properties, albeit at levels of concentrations below the detection limits of the analyses used in this study. This would effectively make the surfaces similar in terms of their characteristics.

The fouling of surfaces with proteins derived from organic foulants such as meat exudates can change the properties of a surface. A recent study conducted by Slate et al. (2019) demonstrated that the surface properties of Ti-ZrN/Ag became more hydrophilic with greater anti-adhesive properties following the introduction of a conditioning film [15]. Furthermore, the presence of a conditioning film may alter the properties of the bacterial cells themselves. When *Staphylococcus* spp. was exposed to a 10% solution of bovine serum albumin (BSA), the bacteria were demonstrated to have a reduction in their hydrophobicity and their propensity to donate electrons [64]. A linear correlation between the negative charge on the bacterial cell surface and the initial attachment to beef lean muscle and fat tissue has also been reported [65]. Such differences in the surface and bacterial properties will influence the interactions between the cell:organic material and the interface.

Standard operating procedures, which include regular cleaning, are used in the food industry to eliminate foodborne pathogens and to reduce contamination, yet despite such measures, surface contamination in food processing facilities still occurs [65]. A fundamental understanding of bacterial attachment to meat surfaces should be the basis for the development of procedures for physical removal of microorganisms that contaminate meat surfaces [11]. The determination of the removal of bacteria in the presence of meat exudate is important since although pathogens have been demonstrated to be easily destroyed by commercial sanitizers in water, the presence of organic matter may significantly affect the function of sanitizers [27,66]. A key aspect of this work is the uneven distribution of fouling across the surface. When surfaces are tested in pristine condition, this allows for easily comparative data between laboratories. However, such methodology although comparable, does not reflect a true environmental situation. The uneven distribution of the conditioning film across the surface demonstrates that the surface in a real environment

will be subjected to very different material-biological interface interactions than occur when a pristine surface is used in such studies. Hence, the use of such organic material in surface-biological interactions is imperative to understand such systems.

The results from our work demonstrated that repeated cleaning of the surfaces resulted in residual organic fouling. When meat processing plants were sampled for biofilms by placing stainless steel and cast iron chips in or on floor drains and food contact areas, it was found that biofilms were formed on the drain samples but were not formed on chips placed on food contact surfaces [67]. Gibson et al. (1999) found that bacterial attachment to surfaces in the food processing environment readily occurred; however, extensive surface colonization and biofilm formation only occurred on environmental surfaces that were not regularly cleaned [23]. In addition, surfaces that were not cleaned daily, resulted in the occurrence of biofilm formation; the bacteria established in a biofilm could not be eradicated by using one single treatment or one single detergent or disinfectant, and the most effective cleaning methods were shown to require scrubbing of the surfaces [68]. With specific regards to a wipe clean, Lopez et al. (2015) showed that using a disinfectant-wipe intervention to clean a contaminated work area that was used in the preparation of chicken fillets decreased the exposure to *Campylobacter jejuni* by 2 to 3 orders of magnitude [69]. Hence, understanding the physical actions of cleaning systems is an important factor in the maintenance of hygienic systems. The cleaning process throughout the food industry is in debate over the best methods, equipment, monitoring, frequency, benchmarks, and standards to be used [70]. Thus, it is important to understand the effects that surface properties have on the cleaning efficacy of the substrata.

#### **5. Conclusions**

This work demonstrated that more bacteria were retained in higher amounts, initially on the stainless steel (FPSS) and titanium coated surfaces with the irregular topographies (TiFP). With subsequent cleaning, the amount of bacteria decreased and was most easily removed from the surfaces that had regular surface features and/or were titanium coated. The direction of cleaning (along or perpendicular to the linear features of the surface) did not have an effect on the amount of bacteria but did affect the amount of meat exudate retained whereby surfaces cleaned along the linear features removed more organic material. After ten cleans, the bacteria and meat exudate retained on the surfaces was not significantly different and suggested that a steady state of the surface properties had been reached. This study highlights the importance of surface properties and cleaning method selection to be utilised within the meat production industry to reduce microbial contamination and surface biofouling.

**Author Contributions:** J.V. devised the project and the main conceptual ideas presented in this study; J.V., I.D.A. and A.J.S. contributed to the manuscript preparation; K.A.W. drafted the final manuscript; A.E. was involved in data generation and analysis; and P.J.K. led the surface design. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The datasets generated during the current study are available from the corresponding author on reasonable request.

**Acknowledgments:** The authors would like to thank Brigitte Carpentier for her kind gift of the *E. coli.* This research was part of the project FOOD-CT-2005-007081 (PathogenCombat) supported by the European Commission through the Sixth Framework Programme for Research and Development.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Toxic Metals in Cereals in Cape Verde: Risk Assessment Evaluation**

**Carmen Rubio-Armendáriz 1,\*, Soraya Paz <sup>1</sup> , Ángel J. Gutiérrez <sup>1</sup> , Verena Gomes Furtado <sup>2</sup> , Dailos González-Weller 1,3, Consuelo Revert <sup>4</sup> and Arturo Hardisson <sup>1</sup>**


**Abstract:** Consumption of cereals and cereal-based products represents 47% of the total food energy intake in Cape Verde. However, cereals also contribute to dietary exposure to metals that may pose a risk. Strengthening food security and providing nutritional information is a high-priority challenge for the Cape Verde government. In this study, toxic metal content (Cr, Ni, Sr, Al, Cd, and Pb) is determined in 126 samples of cereals and derivatives (rice, corn, wheat, corn flour, wheat flour, corn gofio) consumed in Cape Verde. Wheat flour samples stand out, with the highest Sr (1.60 mg/kg), Ni (0.25 mg/kg) and Cr (0.13 mg/kg) levels. While the consumption of 100 g/day of wheat would contribute to 13.2% of the tolerable daily intake (TDI) of Ni, a consumption of 100 g/day of wheat flour would contribute to 8.18% of the tolerable weekly intake (TWI) of Cd. Results show relevant Al levels (1.17–13.4 mg/kg), with the highest level observed in corn gofio. The mean Pb average content in cereals is 0.03–0.08 mg/kg, with the highest level observed in corn gofio. Al and Pb levels are lower in cereals without husks. Without being a health risk, the consumption of 100 g/day of wheat contributes to 17.5% of the European benchmark doses lower confidence limit (BMDL) of Pb for nephrotoxic effects; the consumption of 100 g/day of corn gofio provides an intake of 1.34 mg Al/day (13.7% of the TWI) and 8 µg Pb/day (20% of the BMDL for nephrotoxic effects). A strategy to minimize the dietary exposure of the Cape Verdean population to toxic metals from cereals should consider the continuous monitoring of imported cereals on arrival in Cape Verde, the assessment of the population's total diet exposure to toxic metals and educational campaigns.

**Keywords:** Cape Verde; cereals; metals; dietary intake; risk assessment

### **1. Introduction**

The Macaronesian region consists of a collection of four volcanic archipelagos in the North Atlantic Ocean (Cape Verde, Azores and Madeira in Portugal and the Canaries in Spain). The four archipelagos share features such as a volcanic origin, a contrasting landscape, a gentle climate and a particularly rich biodiversity. The archipelago of Cape Verde is located on the West African coast, 500 km from Senegal, and comprises ten islands, nine of which are inhabited and one of which is uninhabited. The population of the island of Santiago is approximately 260,000 inhabitants, while that of São Vicente is 76,000. The Cape Verdean diet is characterized by the consumption of significant amounts of cereals and cereal-based products. According to the preliminary results of the 2015 Ínquérito Ás Despesase e Receitas Familiares (IDRF), the ingestion of cereals occupies the highest annual per capita consumption expenditure (about 11,611\$) compared to

**Citation:** Rubio-Armendáriz, C.; Paz, S.; Gutiérrez, Á.J.; Gomes Furtado, V.; González-Weller, D.; Revert, C.; Hardisson, A. Toxic Metals in Cereals in Cape Verde: Risk Assessment Evaluation. *Int. J. Environ. Res. Public Health* **2021**, *18*, 3833. https:// doi.org/10.3390/ijerph18073833

Academic Editor: Paul Tchounwou

Received: 24 February 2021 Accepted: 31 March 2021 Published: 6 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

other food products consumed. However, internal cereal production satisfies only 6.9% of the population's consumption needs, contributing to the highly vulnerable state of the country regarding food security. Food security in Cape Verde is also affected by agroclimatic variations and external market fluctuations. National cereal production in 2019 was estimated at about 1000 tons, almost 70% below the mean average of the previous five years [1]. Therefore, about 85% of the domestic cereal demand (mostly rice and wheat for human consumption) was covered by imports. The cereal import requirements in the 2019/2020 marketing year (November to October) were forecasted at an aboveaverage level of 87,000 tons [1]. From 2016 to 2020, the cereal imports reached a total of 419,749.30 tons, with an emphasis on corn (159,979.30 tons), rice (144,799.33 tons) and wheat grain (91,623.39 tons). The market supply of cereals stems both from food aid through cooperative relations with development partners and through commercial imports [2]. Current domestic corn production does not meet the internal demand, and so the cereal must be imported for food and fodder [3]. Moreover, the main drivers of food insecurity in Cape Verde are the effects of dry weather events (such as drought) and pest attacks on cereal and fodder production [1]. As mentioned above, food insecurity in Cape Verde has a structural and multifactorial nature: It demonstrates a structural deficit in national food production, strong dependence on the international market and economic accessibility weaknesses. Strengthening the Food Security and Nutrition Information System (FSNIS) is an important challenge for the Cape Verde government [4].

According to the Food and Agriculture Organization (FAO), in 2017, about 13% of the population was undernourished. The data available indicate that 20% of rural families lived in a situation of food insecurity, with 13% in a moderate position and 7% in a severe position [2]. Cape Verde is in a nutritional transition period characterized both by the high consumption of fat, refined carbohydrates, cholesterol and sugar, and by the low consumption of fruit and vegetables, causing a rapid and significant increase in the prevalence of being overweight and obese [5]. However, the consumption of cereals and cereal-based products is still relevant, representing 47% of the total food energy intake. In Cape Verde, the cereal balance for 2002/2003 estimated a cereal consumption of 242 kg/year per person, comprising 123 kg of corn (337 g/day), 67 kg of rice (184 g/day) and 52 kg of wheat (142 g/day).

Although the nutritional value of cereals is noteworthy, cereals may also contain elements that are harmful to health [6,7], as is the case with elements such as Al, Cd, Cr, Ni, Pb and Sr. Each of these elements has standards of tolerable daily/weekly intake (TDI/TWI) and/or benchmark dose (lower confidence limit) (BMDL) levels set by reference bodies in food safety, such as the European Food Safety Authority (EFSA) and the World Health Organization (WHO) (Table 1).


**Table 1.** Reference intakes of the analyzed elements.

TDI, tolerable daily intake; TWI, tolerable weekly intake; BMDL, benchmark dose level; bw, body weight; Nephrotoxicity <sup>1</sup> and Cardiovascular effects <sup>2</sup> .

Al is a neurotoxic metal with no function in the human body [14]. Prolonged exposure to Al is related to neurodegenerative diseases such as Alzheimer*'*s, and the estimation of its dietary exposure is the subject of previous studies [15–17]. In 2008, the EFSA estimated the dietary intake of Al in the European population to be 0.2–1.5 mg/kg of body weight per week for an adult weighing 60 kg, and concluded that cereals and cereal derivatives are among the main foods that contribute to Al dietary intake [18]. In 2010, González-Weller estimated the total intake of Al in the Canary Islands to be 10.171 mg/day [15].

Cd is a toxic element with a long half-life and a tendency to bioaccumulate [19]. Its presence in cultivation soils favors its transfer to and accumulation in cereals [20]. Known to compete in the body with other essential divalent cations, it affects the renal system, causing irreversible damage to the renal tubules [21,22]. In 2006, Rubio et al. [23] assessed dietary exposure to Cd in the Macaronesian archipelago of the Canary Islands, estimating the intake of Cd from cereals at 1.065 µg/day, and identifying cereals as one of the food categories contributing the most to the dietary intake of Cd. In 2012, the EFSA also identified cereals as one of the food categories that contributes most to the dietary intake of Cd in the European population [24].

Cr is mainly found in the trivalent ion form in food. Although oral Cr (III) is not particularly toxic [25], high intakes of Cr can trigger chronic kidney failure, dermatitis, bronchitis and asthma [26,27]. While cereals were found to contribute most to the dietary intake of Cr (0.087 mg/day) in the Canary Islands archipelago [28] compared to other food categories, a study by Filippini et al. [29] concluded that beverages, cereals and meat provided the highest dietary contributions of Cr in a northern Italian population.

Ni is essential for plants [30], and grains and grain-based products are considered the most important contributors to Ni exposure in the European diet, even though Ni is only regulated in drinking water and not in other food groups [9]. Individuals with hypersensitivity to Ni or with kidney disease are susceptible to damage from a high dietary intake of Ni [26].

Sr is an element that is found in food; however, there are no reported cases of food poisoning from Sr to date. Nevertheless, Sr competes with essential elements such as phosphorus [31], and recent studies in experimental animals reported hepatotoxic effects associated with Sr [32]. The total intake of Sr in the Canary Islands archipelago was estimated at 1.923 mg/day, and cereal intake was estimated at 1.276 ± 0.711 mg/kg w.w. [28].

Pb is a neurotoxic metal that accumulates in the body, causing serious damage to the central nervous system (CNS) as well as contributing to kidney disease, gastrointestinal tract disorders and Alzheimer*'*s [13]. Pb traces can be found in large quantities in food and drinking water [33,34], especially in fruits, vegetables and cereals due to the deposit of Pb particles from the atmosphere. Bread and rolls (8.5%), tea (6.2%) and tap water (6.1%) are among the food categories found to contribute to high Pb exposure in Europe [35]. While Pb intake of the Canarian population was estimated at 72.8 µg/day in 2005 [33], in 2012, mean lifetime dietary exposure in the European population was estimated at 0.68 µg/kg b.w. per day based on middle bound mean lead occurrence [35].

Food risk surveillance and food safety strategies encourage the monitoring of metal in each of the food groups consumed by different populations. The aims of the present study are to determine the levels of Al, Cd, Cr, Ni, Pb or Sr in commonly consumed cereals and cereal-based products in the Cape Verde islands, and to assess their subsequent risk.

#### **2. Material and Methods**

#### *2.1. Samples*

A total of 126 samples of cereals (rice, corn and wheat) and cereal-based products (corn flour, wheat flour and corn gofio) (Table 2) that are marketed and consumed in Cape Verde were acquired from two different islands of the Cape Verde archipelago, specifically, Santiago and São Vicente (Figure 1). Gofio is a traditional artisan food derived from cereals, mainly corn, that is made by first roasting the cereal in its husk and then grinding it until a powder similar to flour is obtained [36–38].


**Table 2.** Analyzed cereal and derived product samples. **Type No. Samples Sampling Location Origin**  56 Santiago Brazil, Vietnam, Thailand, Japan,

**Table 2.** Analyzed cereal and derived product samples.

*Int. J. Environ. Res. Public Health* **2021**, *18*, x FOR PEER REVIEW 4 of 13

cereals, mainly corn, that is made by first roasting the cereal in its husk and then grinding

**Figure 1.** Map of the Cape Verde islands showing the sampling areas (São Vicente and Santiago) (Source: Google Maps).

**Figure 1.** Map of the Cape Verde islands showing the sampling areas (São Vicente and Santiago) (Source: Google Maps). Sampling took place from 2017 to 2019 at establishments that import and sell cereal on the Santiago and São Vicente islands. Because most of the samples were not commercialized in packages, but instead, were mainly sold by weight in local markets, it was not possible to obtain the origin of each individual sample. Nevertheless, according to Entidade Regulatora Independiente da Saúde (ERIS) from Cape Verde, the origins of the cereal samples distributed in Cape Verde are diverse (Table 2). Sampling took place from 2017 to 2019 at establishments that import and sell cereal on the Santiago and São Vicente islands. Because most of the samples were not commercialized in packages, but instead, were mainly sold by weight in local markets, it was not possible to obtain the origin of each individual sample. Nevertheless, according to Entidade Regulatora Independiente da Saúde (ERIS) from Cape Verde, the origins of the cereal samples distributed in Cape Verde are diverse (Table 2).

#### *2.2. Sample Treatment*

One gram of each sample was added to pressure vessels (HVT50, Anton Paar, Graz, Austria) previously washed with laboratory detergent and Milli-Q quality distilled water. Then, 4 mL 65% nitric acid (Sigma Aldrich, Darmstadt, Germany) and 2 mL hydrogen peroxide (Sigma Aldrich, Darmstadt, Germany) were added to the samples. The pressure vessels were closed and placed in a microwave oven (Multiwave Go Plus, Anton Paar, Graz, Austria) for subsequent digestion according to the conditions described in Table 3. After the samples were digested, they were transferred to 10 mL volumetric flasks and made up

with Milli-Q quality distilled water. Finally, they were transferred to airtight jars with a lid for later measurement.


**Table 3.** Microwave digestion process instrumental conditions.

Microwave processing power: 850 W; Limit temperature: 200 ◦C; Cooling temperature: 50 ◦C.

#### *2.3. Analytical Method*

The determination of metal content was conducted by Inductively Coupled Plasma Atomic Emission Spectrometry (ICP-OES) model ICAP 6300 Duo Thermo Scientific (Waltham, MA, USA), with an Auto Sampler automatic sampler (CETAX model ASX-520).

The instrumental conditions of the method comprised the following: RF power of 1150 W; gas flow (nebulizer gas flow, make up gas flow) of 0.5 L/min; injection of the sample to the 50-rpm flow pump; stabilization time of zero s [39,40]. Instrumental wavelengths (nm) of the analyzed elements were Al (167.0), Cd (226.5), Cr (267.7), Ni (231.6), Pb (220.3) and Sr (407.7).

The quantification limits of the toxic metals, calculated as ten times the standard deviation (SD) resulting from the analysis of 15 targets under reproducibility conditions [41], were: 0.012 mg/L (Al), 0.001 mg/L (Cd), 0.008 mg/L (Co), 0.003 mg/L (Ni), 0.001 mg/L (Pb) and 0.003 mg/L (Sr).

The quality control of the method (Table 4) was based on the recovery percentage obtained with reference material (SRM 1515 Apple Leaves, SRM 1548a Typical Diet, SRM 1567a Wheat Flour) under reproducible conditions. The recovery percentages obtained with the reference material were above 94% in all cases. The statistical analysis did not detect significant differences (*p* < 0.05) between the certified concentrations and the concentrations obtained.


**Table 4.** Recovery study results and reference materials used.

#### *2.4. Statistical Analysis*

The IBM Statistics SPSS 24.0 computer software for Windows was used for statistical analysis. Two studies were conducted in order to check the significance of the differences (*p* < 0.05) in the metal contents both between cereals and derived product types and between locations. Kolmogorov-Smirnov and Shapiro-Wilk tests were used to check normality, and Levene's test was applied to check the homogeneity of the variances based on the mean, median and trimmed mean. Data followed a non-normal distribution, and consequently, the Kruskal-Wallis nonparametric test was applied [42]. A one-way study was conducted with the fixed factor *"*Cereal type*"* and six levels of variation: *rice, corn gofio, corn flour, wheat flour, corn, wheat*. The Mann-Whitney test was also conducted (95% confidence interval) to determine significant differences in the concentrations of elements according to the cereal type or product. Another one-way study was conducted with the fixed factor *"*Location*"*

and two levels of variation: *Santiago, São Vicente*. Finally, another Mann-Whitney test was used, and 166 data were analyzed with a 95% confidence interval. *2.5. Calculation of Dietary Intake*  The assessment of dietary exposure was based on the calculation of the estimated

Mann-Whitney test was used, and 166 data were analyzed with a 95% confidence interval.

The IBM Statistics SPSS 24.0 computer software for Windows was used for statistical analysis. Two studies were conducted in order to check the significance of the differences (*p* < 0.05) in the metal contents both between cereals and derived product types and between locations. Kolmogorov-Smirnov and Shapiro-Wilk tests were used to check normality, and Levene's test was applied to check the homogeneity of the variances based on the mean, median and trimmed mean. Data followed a non-normal distribution, and consequently, the Kruskal-Wallis nonparametric test was applied [42]. A one-way study was conducted with the fixed factor *"*Cereal type*"* and six levels of variation: *rice, corn gofio, corn flour, wheat flour, corn, wheat*. The Mann-Whitney test was also conducted (95% confidence interval) to determine significant differences in the concentrations of elements according to the cereal type or product. Another one-way study was conducted with the

#### *2.5. Calculation of Dietary Intake* daily intake (EDI) and the subsequent obtained percentage contribution to the reference value (TDI for Cr, Ni and Sr; TWI for Al and Cd; BMDL for Pb) of each of the metals under

The assessment of dietary exposure was based on the calculation of the estimated daily intake (EDI) and the subsequent obtained percentage contribution to the reference value (TDI for Cr, Ni and Sr; TWI for Al and Cd; BMDL for Pb) of each of the metals under study (Table 1). study (Table 1). EDI (mg/day) = Mean consumption (kg/day) × Element concentration (mg/kg fresh weight)

EDI (mg/day) = Mean consumption (kg/day) × Element concentration (mg/kg fresh weight) Contribution (%) = [EDI/Reference value] × 100

*Int. J. Environ. Res. Public Health* **2021**, *18*, x FOR PEER REVIEW 6 of 13

Contribution (%) = [EDI/Reference value] × 100

#### **3. Results and Discussion** Figure 2 shows box plots with the mean concentrations (mg/kg fresh weight), stand-

**3. Results and Discussion** 

Figure 2 shows box plots with the mean concentrations (mg/kg fresh weight), standard deviations (SD) and comparisons of the concentrations between the different cereals and the derived products. ard deviations (SD) and comparisons of the concentrations between the different cereals and the derived products.

**Figure 2.** Box plot of mean trace element concentrations (mg/kg) by cereals and derived products. **Figure 2.** Box plot of mean trace element concentrations (mg/kg) by cereals and derived products.

Al was found in the highest concentrations in all analyzed cereal samples, most clearly in corn gofio, where it reached a mean average concentration of 13.4 ± 12.7 mg/kg fresh weight. This concentration differs significantly from the rest of the cereals (*p* < 0.05). Liu et al. [43] concluded that cereal husks contain higher concentrations of metals than the grain. Accordingly, the differences in the Al content recorded here in corn gofio may be due to the use of the whole cereal, including the husk, in the manufacture of this corn-derived product [35], which may explain the higher Al content. However, despite the toxicological considerations of this neurotoxic element, current European legislation does not include maximum levels of Al in food.

The wheat flour samples are worth mentioning, as they presented the highest levels of Sr (1.60 mg/kg fresh weight), Ni (0.25 mg/kg fresh weight) and Cr (0.13 mg/kg fresh weight). The Second French Total Diet Study (TDS) had a mean level of Sr in breakfast cereals of 0.842 mg/kg fresh weight [44]; this value was lower than the level obtained in the wheat samples of the present study. In addition, Cubadda et al. [45] reported lower Ni levels in flour and wheat (0.035 mg/kg) than those observed in this study. However, Mathebula et al. [46] observed a mean Cr level in wheat of 2.629 mg/kg fresh weight, higher than the mean level recorded in this study.

As observed for Sr, Ni and Cr, the wheat flour samples also presented the highest mean concentration of Cd (0.02 ± 0.01 mg/kg fresh weight). Tejera et al. [47] recorded mean Cd concentrations of 0.027 mg/kg fresh weight in wheat flour, values similar to those recorded in the present study. However, regarding wheat grain, Škrbi´c et al. [48] observed Cd levels in Serbian wheat of 2.4–252 µg/kg fresh weight, higher than those registered in the wheat analyzed here (0.01 ± 0.01 mg/kg fresh weight).

As for Pb, the highest mean level was observed in the corn gofio samples, with a mean concentration of 0.08 ± 0.05 mg/kg fresh weight. Furthermore, this concentration may indicate that Pb tends to accumulate in the husk of cereals, since in cereal-based products manufactured without the husk, the Pb levels were lower. A study conducted by Bilo et al. [49] on rice and rice husks concluded that rice husks accumulated higher concentrations of toxic metals than rice. This suggests that gofio, being a derivative produced from whole-grain cereal, including the husk, may have higher Pb levels than flours produced from dehusked cereal.

The statistical analysis showed significant differences (*p* < 0.05) in the Pb content between wheat and the rest of the samples, in the Al content between the rice and wheat samples and in the Sr and Ni content of the rice and corn samples when compared to the wheat samples.

Figure 3 presents box plots with the mean concentrations (mg/kg fresh weight), standard deviations (SD) and the comparisons of the concentrations between the sampling locations. The samples from São Vicente presented the highest mean concentrations of Al, Cd, Cr, Ni, Sr and Pb. Considering that these differences may be due to multiple factors [48,50], it is suggested that in future risk-assessment studies, correlations between metal levels and the origin of the imports are calculated. Minimizing the dietary exposure of the Cape Verdean population to metals of toxicological relevance involves risk management actions, including continuous monitoring of these metals in the different food commodities upon arrival in Cape Verde, as well as importing higher-quality cereals that also have lower concentrations of Al, Cd, Cr, Ni, Sr and Pb. In addition, cereals with higher levels of metals, such as Pb and Al, should not be used for the manufacture of cereal-based products containing the husk, but rather, should be used in the manufacture of flours after being dehusked.

In Cape Verde, the cereal balance for 2002/2003 estimated a cereal consumption of 242 kg/year per person, made up of 123 kg maize (337 g/day), 67 kg rice (184 g/day) and 52 kg wheat (142 g/day). However, since there are no additional current data on the consumption habits of cereals and cereal-based products, the estimations here of the dietary exposure (Estimated Daily Intake, EDI) of the Cape Verdean population to the metals under study were performed using a mean ration of 100 g/day of each cereal and its derivatives (Table 5). The European reference limits (Table 1) were used for the evaluation of the EDI of the Cape Verde population. The TDI, TWI, and BMDL were used, along with an estimated mean average weight of an adult individual of 68.48 kg (similar to that of the Spanish population) [51].

*Int. J. Environ. Res. Public Health* **2021**, *18*, 3833


*Int. J. Environ. Res. Public Health* **2021**, *18*, x FOR PEER REVIEW 8 of 13

**Figure 3.** Box plot of mean trace element concentrations (mg/kg) by sampling location. **Figure 3.** Box plot of mean trace element concentrations (mg/kg) by sampling location.

In Cape Verde, the cereal balance for 2002/2003 estimated a cereal consumption of 242 kg/year per person, made up of 123 kg maize (337 g/day), 67 kg rice (184 g/day) and 52 kg wheat (142 g/day). However, since there are no additional current data on the consumption habits of cereals and cereal-based products, the estimations here of the dietary exposure (Estimated Daily Intake, EDI) of the Cape Verdean population to the metals under study were performed using a mean ration of 100 g/day of each cereal and its derivatives (Table 5). The European reference limits (Table 1) were used for the evaluation of the EDI of the Cape Verde population. The TDI, TWI, and BMDL were used, along with an Thus, the consumption of 100 g/day of wheat represents a contribution percentage of 13.2% to the TDI (tolerable daily intake) of Ni, i.e., 13 µg/kg bw/day. In the case of sensitive individuals or people with kidney problems, a high intake of Ni may be a dietary hazard and health risk [9]. The consumption of 100 g/day of wheat was found to provide a contribution percentage of 17.5% of the European BMDL of Pb set at 0.63 µg/kg bw/day for nephrotoxic effects [13]. This percentage may represent a relevant contribution to the total intake of Pb with the consequent risk to health. Similarly, the consumption of 100 g/day (700 g/week) of corn gofio contributes 13.7% of the TWI (tolerable weekly intake) of Al set in Europe at 1 mg/kg bw/week [11].

estimated mean average weight of an adult individual of 68.48 kg (similar to that of the Spanish population) [51]. The Al levels detected in the corn gofio differed between the Santiago and Sào Vicente islands; in the case of Sào Vicente (39 mg Al/kg fresh weight), the consumption of 100 g/day with an Al content of 39 mg/kg fresh weight would mean an intake of 3.9 mg Al/day from this food alone, i.e., almost 39.9% of the TWI for Al.

Assuming that food risk management needs to be accompanied by a communication plan, the authors believe that the nutritional re-education campaigns and actions provided in the PERVEMAC2 Project could contribute to communicating and disseminating this knowledge to the Cape Verdean population, risk managers and policy regulators. Previous studies carried out in Cape Verde [52] have pointed to the success of involving women in health promotion because of their decision-making power; their multidimensional role in purchasing, processing and preparing food as the pillar of familial food security; and their contribution via nonformal economic activities for their families. Focus group discussions

and intensive fieldwork reinforced the higher participation of residents in the informal unit and women in all stages, suggesting the practicability of health-promotion campaigns; this work also showcases the potential of the social capital of the informal settlements and the role of the woman in the family and society in Cape Verde [52].

#### **4. Conclusions**

In this study, the existence of significant differences in the content of elements analyzed between different cereals is confirmed, which reaffirms the need for continuous monitoring of both locally produced and imported cereals upon arrival in Cape Verde as risk management and minimization strategies, while also continuing to monitor the population's total dietary exposure to toxic metals. Furthermore, cereals with higher levels of metals such as Pb and Al should not be used with the husk for the manufacture of cereal-based products, but rather, should be used in the manufacture of flours only after removing the husk. In the case of Al, it would be advisable for the food safety authorities to set a maximum limit for this element in cereals and cereal-based products, thus allowing quality control and minimizing the population's exposure to this neurotoxic element. The evaluation of dietary exposure to the toxic metals studied here in cereals and their cerealbased products should undoubtedly be complemented with future studies targeting other groups of basic foods in the diet of the Cape Verde population.

**Author Contributions:** Conceptualization, C.R.-A. and Á.J.G.; Data curation, V.G.F.; Formal analysis, C.R.-A., Á.J.G. and D.G.-W.; Funding acquisition, C.R.-A. and A.H.; Investigation, C.R.-A., S.P., Á.J.G. and A.H.; Methodology, C.R.-A., S.P., Á.J.G., V.G.F. and A.H.; Project administration, C.R.-A., Á.J.G. and A.H.; Resources, A.H.; Software, Á.J.G.; Supervision, C.R.-A., S.P., Á.J.G., C.R. and A.H.; Writing—original draft, C.R.-A., S.P. and C.R.; Writing—review & editing, C.R.-A., Á.J.G., V.G.F. and A.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by PERVEMAC II: Programa de Cooperación INTERREG V-A España-Portugal MAC (Madeira-Azores-Canarias) 2014–2020 grant number MAC/1.1a/049. Project "Sustainable Agriculture and Food Security in Macaronesia: Investigation of the benefits and risks of the intake of plant products for the health of consumers and development of minimization strategies"

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The datasets generated during the current study are not publicly available but are available from the corresponding author on reasonable request.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

