**3. Conclusions**

The proposed analytical methodology, sample pre-treatment with sodium bicarbonate and PBS followed by IAC clean-up and LC-FD, enabled low detection and quantification levels and good results regarding accuracy and precision.

The application of this method to 85 samples showed that 10.59% were contaminated, two of which were homemade and presented considerable concentrations (9.21 and 11.25 μg/L). The discrepancy found can be explained by the storage of cereals already prepared for homemade brewing that may not be as strictly controlled as the others.

Three risk assessments were carried out based on three di fferent scenarios. In the first two, the ingestion of OTA through the consumption of beer presents no risk to the respective consumers. The inverse situation was observed for the worst case scenario, where the most contaminated sample was considered.

The EU established maximum limits for OTA in some products. Current legislation does not include limits for the occurrence of OTA in beer, but the identified concentrations, especially in homemade beers, should be considered. For this reason, it is important to adopt preventive measures and develop control programs, reviewing the critical points where OTA production can occur in order to minimize human exposure to OTA.

#### **4. Materials and Methods**

#### *4.1. Chemicals and Materials*

HPLC grade acetonitrile and methanol and PBS tablets were purchased from Sigma Chemicals Co. (St. Louis, MO, USA). Toluene was acquired from Carlo Erba (Milan, Italy). Acetic acid was obtained from Merck (Darmstadt, Germany) and 98% purity degree OTA was obtained from Sigma Chemicals Co. (St. Louis, MO, USA).

The OTA stock solution was prepared at 250 μg/mL in toluene-acetic acid (99:1) and stored at −20 ◦C. The intermediate solution was prepared at 10 μg/mL, in mobile phase, and diluted accordingly to obtain the external calibration solutions.

Bi-distilled water was obtained from a Milli-Q System (Millipore, Bedford, MA, USA). A mixture of acetonitrile–water–acetic acid (49.5:49.5:1 *v*/*v*/*v*) was used as mobile phase. All chromatographic solvents were filtered through a 0.20 μm membrane filter (Whatman GmbH, Dassel, Germany) and degassed.

Immunoa ffinity columns (IACs) OchraTestTM (Vicam/Waters, Milford, MA, USA) were used for clean-up. Micro-glass fiber paper (150 mm, Munktell & Filtrak GmbH, Bärenstein, Germany), cellulose nitrate (0.45 μm, Sartorius Stedim Biotech GmbH, Göttingen, Germany), and Durapore membrane filter (0.22 μm, GVPP, Millipore, Ireland) were also used.

#### *4.2. Sampling and Sample Characterization*

In 2018, 84 bottled commercial beer samples and one draft beer, representing 59 brands, were randomly acquired from di fferent retail outlets and supermarkets located in Coimbra (Coimbra, Portugal). The samples were classified based on the type of production, type of beer, color, fermentation, alcohol content, and country of origin.

In total, 61 samples were industrial manufactured, 21 were craft beers, and 3 were homemade beers. Regarding the type of beer, 44 were ale with top-fermentation, and 41 were lager, two of which were fruit/vegetable beers with bottom-fermentation. Of all the samples, 59 were of pale color, 2 were pale-red, and 24 were dark beers. Of the 85 beer samples, 30 were strong beers, with an alcohol content ≥6% and 55 contained alcohol <6% (3 samples were non-alcoholic, with <1% alcohol).

The majority of the samples (79) were of European origin and six were from abroad. Of all the beers, 36 were produced in Portugal. The imported beers originated from Belgium (n = 15), the Czech Republic (n = 3), Germany (n = 11), Ireland (n = 1), Mexico (n = 1), the Netherlands (n = 5), New Zealand (n = 2), Poland (n = 1), Russia (n = 3), Scotland (n = 2), and Spain (n = 5).

Among these samples, seven were brewed with organically produced materials and were labelled as organic beers. None of the samples was analyzed beyond their expiration date. Until analysis, they were stored in the dark at 4 ◦C and all the information available on the labels was assembled.

### *4.3. Experimental Procedure*

Based on a previously reported analytical methodology [41], degassed and consecutively filtered beer samples (10 mL) were added, of 4% sodium bicarbonate (1.25 mL) and 10 mL of PBS. After centrifugation, the extract was loaded into an IAC cartridge for clean-up. After a washing step with 5 mL of water, OTA was eluted with 4 mL of methanol. Afterwards, the solvent was evaporated at 40 ◦C under a gentle nitrogen stream, and the dried residue was stored at −20 ◦C until analysis. For liquid chromatography with fluorescence detection (LC-FD), redissolution was accomplished with 1 mL of mobile phase. Following filtration through a Durapore membrane filter, 20 μL were injected into the HPLC system that consisted of a 805 manometric module Gilson, and a fluorimetric detector from Jasco (Tokio, Japan) FP-2020 Plus. Excitation and emission wavelengths were 336 nm and 440 nm, respectively. A C18 Nucleosil 5 μm (4.6 × 250 mm i.d.) column (Hichrom, Leicestershire, UK) was used and the flow rate was set at 1 mL min −1. The total run time was 15 min.

#### *4.4. Validation and Quality Control Assays*

Validation and quality control assays were performed as set by European guidelines [31]. Di fferent parameters were evaluated, including linearity, limit of detection (LOD) and limit of quantification (LOQ), matrix e ffect (ME), accuracy and precision. Linearity was assessed using standards (2.5–25 μg/L), and matrix-matched solutions (0.25–2.5 μg/L). Sensitivity was evaluated through the matrix-matched calibration curve.

The LOD was set as |3.3Sy/x|/b and the LOQ as |10Sy/x|/b, respectively, knowing that b corresponds to the slope and Sy/x corresponds to the residual standard deviation of the linear function.

The ME (%) corresponds to the percentage of the ratio of matrix-matched calibration curve slope (B) and the slope of the standard calibration curve (A). The results were interpreted as follows: 100% signifies an absence of ME; a result higher than 100% corresponds to a signal enhancement; and a result lower than 100% corresponds to a signal suppression.

Accuracy and precision were evaluated using blanks and fortified samples at three levels (0.5, 1.0, and 2.0 μg/L). Three replicates were made (n = 3), in three di fferent days for each fortification level. The relative standard deviation (RSD) of intra-day and inter-day repeatability were assessed to evaluate the precision of the analytical methodology.

#### *4.5. Calculation of the Human Estimated Daily Intake and Risk Assessment*

A deterministic method [42] was used to calculate the OTA estimated daily intake (EDI) through the consumption of beer through the following equation (*1*):

$$\text{EDI} = (\sum \text{c)}. (\text{CN}^{-1} \text{ D}^{-1} \text{ K}^{-1}), \tag{1}$$

where c is the OTA sum in the positive samples (μg/L), C corresponds to the mean annual beer consumption estimated per inhabitant, N is the samples' number, D corresponds to the days of a year, and K is the body weight (kg). According to Statistics Portugal (INE) data, the beer consumption in 2016 was 50.9 L/inhabitant (C) [43]. The mean body weight considered for the Portuguese adult population was 69 kg (K) [44].

Three di fferent scenarios were used to perform three EDI evaluations. In the first scenario, the OTA contamination levels were taken into consideration for the total of all analyzed samples. For the second scenario, the mean OTA content was considered. Finally, the worst case scenario was observed using the highest OTA level.

**Author Contributions:** Conceptualization of this study was made by C.M.L. and A.P.; methodology was optimized by A.C.T. and L.J.G.S., investigation and data collection were performed by A.C.T., A.M.P.T.P., L.J.G.S., writing—original draft preparation by A.C.T.; writing, review and editing, C.M.L. and L.J.G.S.; overall supervision by C.M.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** Fundação para a Ciência e a Tecnologia: UIDB 50006/2020

**Acknowledgments:** The authors gratefully acknowledge the Portuguese governmental Fundação para a Ciência e a Tecnologia—FCT for funding support through the project UIDB 50006/2020.

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