Development, Validation and Application of an ICP-SFMS Method for the Determination of Metals in Protein Powder Samples, Sourced in Ireland, with Risk Assessment for Irish Consumers

A method has been developed, optimised and validated to analyse protein powder supplements on an inductively coupled plasma-sector field mass spectrometer (ICP-SFMS), with reference to ICH Guideline Q2 Validation of Analytical Procedures: Text and Methodology. This method was used in the assessment of twenty-one (n = 21) elements (Al, Au, Ba, Be, Bi, Cd, Co, Cr, Cu, Fe, Hg, Li, Mg, Mn, Mo, Pb, Pt, Sn, Ti, Tl, V) to evaluate the safety of thirty-six (n = 36) protein powder samples that were commercially available in the Irish marketplace in 2016/2017. Using the determined concentrations of elements in samples (µg·kg−1), a human health risk assessment was carried out to evaluate the potential carcinogenic and other risks to consumers of these products. While the concentrations of potentially toxic elements were found to be at acceptable levels, the results suggest that excessive and prolonged use of some of these products may place consumers at a slightly elevated risk for developing cancer or other negative health impacts throughout their lifetimes. Thus, the excessive use of these products is to be cautioned, and consumers are encouraged to follow manufacturer serving recommendations.


Importance of Elemental Monitoring and Regulation in Food
Everything around us, from the air we breathe to the food we consume, consists of elements. From a clinical perspective, these elements can be separated into different groups and classified based on their respective roles in human health. As can be seen below in Table 1 (adapted from Ring et al., 2016 [1]), the human body is comprised of these elements with many playing critical roles in metabolic processes, while others can cause negative side effects at certain concentrations. Balanced nutrition is therefore an essential aspect of maintaining health, as it allows our bodies to grow, repair and function and promotes overall health and wellbeing [2].
Over the last number of years, interest in fitness has been on the rise, and consequently, products that supplement dietary nutritional requirements to help build muscle and enhance performance and endurance have become more popular [3,4]. Protein powder is one such supplement, which is manufactured from materials such as plants, eggs and dairy products, whose intended purpose is to support the intake of adequate amounts of protein needed to promote muscle growth and recovery [5,6]. Research has indicated that the early intake of protein following training, in amounts of 0.25-0.30 g/kg bodyweight (17.5-21.0 g

Instrumentation
All standards, controls and samples were analysed by an ELEMENT2™ ICP-SFMS (Thermo Fisher Scientific, Bremen, Germany). Solutions were introduced to the instrument by peristaltic pump using an SC-E2 FAST autosampler (Elemental Scientific Inc., Omaha, NE, USA) equipped with a quartz cyclonic spray chamber. Digestion of samples was performed using a Mars6 iWave microwave digestion unit (CEM, Matthews, NC, USA) equipped with a MarsXpress carousel for forty TFM digestion vessels (55 mL). Further information on the theoretical principles of the ICP-SFMS analysis and microwave-assisted acid digestion can be found in the Supplementary Materials Section S2-ELEMENT2™ ICP-SFMS.

Gases, Reagents and Volumetric Equipment
The selection of reagents, ISTDs and analyte isotopes, and the preparation of the acid diluent and calibration standards is further detailed in the Supplementary Materials Section S3-Method Development. High purity grade argon gas (14-cylinder MCP, 99.996% pure) was purchased from Irish Oxygen (Cork, Ireland). Deionised water, with resistivity of 15.0 MΩ·cm, was obtained from an ELGA Purelab ® Option water purification system (ELGA LabWater, High Wycombe Buckinghamshire, UK). PlasmaPure HNO 3 (67-69% w/w) and HCl (34-37% w/w) were purchased from SCP Science (through QMX Laboratories, Thaxted, Essex, UK). Calibration standards: Solutions were prepared in the concentration range 0.001-50 µg·L −1 through serial dilution of a 5 µg.mL −1 multielement standard traceable to NIST standard reference materials from SCP Science (through QMX Laboratories, Thaxted, Essex, UK). Internal standard: A 100 µg·L −1 solution containing Sc, Rh, Ir and Ga was prepared from two 5 µg·mL −1 multielement standards traceable to NIST standard reference materials from SCP Science (through QMX Laboratories, Thaxted, Essex, UK). This solution was spiked into all standards, samples and blanks to achieve a final concentration of 2.5 µg·L −1 . Certified reference material (Seronorm™ Trace Element Urine L-2) was purchased from SERO (Billingstad, Norway). Daily optimisation of instrument settings was performed using a 1 µg·L −1 Tune-Up solution (ThermoScientific, Bremen, Germany, P/N 1099601), which was diluted 1:10 with 2.82% w/w nitric acid. Grade A polymethylpentene (PMP) volumetric flasks, as well as beakers, graduated cylinders, 15 mL polypropylene (PP) sample tubes and disposable pipettes were purchased from VWR International Ltd. (Blanchardstown, Dublin 15, Ireland).

Quality Assurance
Analytical parameters such as linearity, limit of detection (LOD), limit of quantification (LOQ), selectivity, specificity, precision and measurement of uncertainty were each evaluated in the validation of this method (see Supplementary Materials Section S4-Method Validation). Method accuracy was also verified through the analysis of a certified reference material (Seronorm™ Trace Elements Urine L-2, Lot 1403081), where recoveries of 85-115% were recorded, meeting the recovery acceptance criteria of 100 ± 20% (see Supplementary  Table S23).
The accuracy of analytical determinations of protein powder samples was corroborated through the measurement of matrix-spiked controls. Because none of the protein powder samples analysed had blank backgrounds for the isotopes of interest, one sample was selected (P14) and diluted before being spiked with aliquots of standard and ISTD solutions. Five control levels were included to cover the calibration range (A = 0.2 µg·L −1 , B = 1 µg·L −1 , C = 5 µg·L −1 , D = 15 µg·L −1 and E = 40 µg·L −1 ). An acceptance limit of 100 ± 25% recovery was applied.
High recoveries were recorded for some analytes across these levels, driven by high backgrounds in the sample used for matrix spiking (P14). In particular, Cu contamination was noted across all control concentration levels. In these cases, analysis of calibration readback standards in 2.82% HNO 3 /0.24% HCl diluent at similar concentrations to the matrix-spiked controls served as verification of the calibration and instrument performance. These standards were analysed in the middle and at the end of the analytical sequence.
Due to high background levels in the deionised water used, some controls at the 0.2-5 µg·L −1 levels could not be determined for Mg, Al and Fe. However, in all samples investigated, the concentrations determined for each of these analytes were above these control levels, and on that basis, the sample results were accepted. Cu controls were not assessed at the 0.2 µg·L −1 level because they fell below the LOQ for Cu (0.25 µg·L −1 ). A summary of the recoveries of these controls can be found in Table 2 below. * = recovery determined through analysis of two matrix-spiked controls due to carryover; a = recovery determined through analysis of 0.25 µg·L −1 standard in 2.82% HNO 3 /0.24% HCl diluent (n = 2); b = recovery determined through analysis of 1.5 µg·L −1 standard in 2.82% HNO 3 /0.24% HCl diluent (n = 2); c = recovery determined through analysis of 4 µg·L −1 standard in 2.82% HNO 3 /0.24% HCl diluent (n = 2); d = recovery determined through analysis of 12.5 µg·L −1 standard in 2.82% HNO 3 /0.24% HCl diluent (n = 2); e = recovery determined through analysis of 35 µg·L −1 standard in 2.82% HNO 3 /0.24% HCl diluent (n = 2).

Sample Preparation: Microwave-Assisted Acid Digestion
Thirty-six protein powder samples were purchased online and from local retail outlets in Cork, Ireland, in 2016 and 2017. Fourteen popular brands were investigated in this study, which included different types of protein powder supplements: whey (n = 27), pea (n = 2), soy (n = 2), mixed plant (n = 2), whey/soy/egg blends (n = 2) and casein (n = 1). The sample IDs can be found in Table 3.

Digestion Vessel Preparation
Prior to use, all Mars6 Xpress vessels (TFM, 55 mL final volume) were rinsed in triplicate with deionised water before adding 10 mL of 5% w/w HNO 3 . Vessels were then sealed, and the OneTouch™ Express Clean method was initiated. During this cleaning cycle, temperature was ramped up to 150 • C over 15 min and held at that temperature for 10 min before cooling down. The vessels were then rinsed again in triplicate using deionised water and allowed to air dry.

Pre-Digestion of Protein Powder Samples
On opening the sealed product packages, 0.5 g of powder was accurately weighed out and transferred into the pre-cleaned MarsXpress digestion vessels, followed by 8 mL of HNO 3 (SCP Science PlasmaPure, 67-69% w/w), 0.5 mL HCl (SCP Science PlasmaPure, 34-37% w/w) and 1.5 mL deionised water (ELGA Purelab ® Option 15.0 MΩ·cm). Before sealing, the inner lid was positioned and the vessels were gently swirled to encourage mixing and then left to stand for 15 min to allow the venting of initial reaction gasses. After 15 min, the vessels were capped, and the samples were placed on the Mars6 carousel for digestion.

Digestion and Preparation of Protein Powder Samples for ICP-SFMS Analysis
The internal temperature of each vessel was ramped up to 170 • C over 15 min and held at that temperature for 1 min, before ramping up again to a final temperature of 190 • C over 10 min. The samples were held at this temperature for 20 min at a pressure of 800 psi before cooling. The vessels were opened, and the digested samples (P1-P36) were quantitatively transferred into 15 mL sample tubes (which were previously rinsed in triplicate with deionised water and air dried). Each tube was capped, gently inverted and opened to release any residual build-up of reaction gasses before being recapped. The returned final volume of each sample was recorded, and the samples were stored at −24 • C until required for analysis. Ahead of ICP-SFMS analysis, samples were removed from the freezer and allowed to equilibrate at room temperature before undergoing a 1:5 dilution with the 2.82% HNO 3 /0.24% HCl diluent and being spiked with ISTD.

ICP-SFMS Analysis
Instrumental conditions were assigned as per Table 4. Parameters with asterisks (*) were optimised during daily tuning of the instrument using a 1 µg·L −1 Tune-Up solution from ThermoScientific (diluted 1:10 prior to analysis). In order to be deemed sufficiently sensitive, the instrument needed to achieve an intensity response of at least 100,000 cps for the indium ( 115 In) reference isotope in low resolution (LR). Where this sensitivity was not initially met, sample gas flow rate was adjusted as well at the X-, Yand Z-axis positions of the torch. During instrument tuning, the integrity of the entrance slit was monitored via ion transmission ratios between low (LR), medium (MR) and high (HR) resolutions. Percentage transmission (%T) of the indium ( 115 In) isotope indicated the ability of the instrument to sufficiently move between LR → MR → HR, where the following transmission acceptance criteria applied: MR intensity/LR intensity: %T ≈ 10-12% HR intensity/LR intensity: %T ≈ 1-2% Post-tuning, the autosampler sample probe was placed into a 500 mL container of diluent (2.82% HNO 3 /0.24% HCl), and the sample lines were purged to rinse the lines, removing any residual tune solution as well as trapped air bubbles. The diluent was left to aspirate for 15-30 min to condition the system prior to initiating the sample sequence. Blank solutions (2.82% HNO 3 /0.24% HCl) were analysed at the start of each sample sequence before measuring calibration standards (0.001-50 µg·L −1 ).
The concentrations of analysed solutions were determined by interpolation using standard calibration, and the impact of potential interferences were minimised through the manual addition of suitable ISTDs. To determine the concentration of each analyte in the original protein powder sample, the following formula (1) was applied to the results recorded from the instrumental analysis: where Cfinal = calculated concentration of analyte in the original sample (µg·kg −1 ); Cinst = determined concentration of the sample solution analysed by ICP-SFMS (µg·L −1 ); DF = dilution factor (5 mL/mL); V = returned volume of the digested sample (mL); W = mass of original sample weighed out (kg).

Non-Carcinogenic Risk
Hazard quotient (HQ) looks at the non-carcinogenic risk to human health from exposure to toxic and potentially toxic elements [22]. HQ assesses the relationship between the estimated daily intake (EDI) of potentially harmful elements with respect to the established oral reference dose (RfD). The RfD of a substance is an estimate of the acceptable level of exposure where the risk of negatively impacting human health is negligible. Hence, hazard quotient was used to characterise potential health risks associated with the consumption of the thirty-six protein powder samples under investigation.
A summary of the available RfD values can be found in Table 5. There are no established RfD values for Pt, Bi, Mg, Ti or total Cr. Therefore, these analytes were omitted from the risk assessment. If HQ < 1, the risk of exposure is not expected to pose any adverse health effects, while HQ > 1 indicates that ingestion of the product carries an increased health risk for consumers. To estimate HQ, the following formulae (2) and (3) were used [14]: where Cfinal = calculated concentration of analyte in the original sample (µg·kg −1 ); IR = intake rate of protein powder per respective manufacturer serving sizes (kg·day −1 , 1-3 servings); BW = bodyweight (70 kg); RfD = reference oral dose (µg·kg −1 BW·day −1 ). Where a product contains more than one toxic/potentially toxic element, exposure to the product may incur interactive effects, which can enhance the overall risk to the consumer. The hazard index (HI) estimates this increased risk through the addition of the calculated HQs of each element [23], as per the following Equation (4): H I = HQLithium +HQBeryllium + HQMolybdenum + HQCadmium +HQTin + HQBarium + HQMercury + HQThallium +HQLead + HQAluminium + HQVanadium +HQManganese + HQIron + HQCobalt + HQCopper As with individual HQs, if HI < 1, the product is not expected to cause harm to the consumer. However, if HI > 1, the product is potentially unsafe for consumption.

Carcinogenic Risk
Cancer risk (CR) was calculated to evaluate the long-term risk of developing cancer through exposure to known and potential carcinogenic elements present in the protein powder samples. According to the International Agency for Research on Cancer (IARC), Be, Cd and Cr(VI) are listed as Group 1 known carcinogens, while Pb is classified as a Group 2A probable carcinogen. As such, each element has a unique cancer slope factor (CSF), which is used to estimate future cancer risk. Because a CSF value has not been established for total Cr, Cr was omitted from the carcinogenic risk assessment. A summary of the available CSF values can be found in Table 5. The following Equation (5) was used to determine cancer risk: where EDI = estimated daily intake (µg·kg −1 BW·day −1 ); CSF = slope factor of the carcinogenic element (µg·kg −1 BW·day −1 ) −1 .

Concentration of Metals in Protein Powder Samples
The level of each element found in the protein powder samples was recorded and can be seen in Table 6. Many of the samples tested had concentrations below the limit of detection for Li, Be and Sn, while many of the Al and Fe levels recorded were above the calibration range. The Mg concentration exceeded the calibration range for all samples tested, and therefore Mg was not reported.

Lithium ( 7 Li)
As a therapeutic element, Li is commonly used in the treatment of psychological afflictions such as schizophrenia and depression [31]. The majority of samples tested had Li concentrations that were below the LOQ, resulting in a final concentration of <17 µg·kg −1 in samples. Of those samples with concentrations above the LOQ, a concentration range of 30.3 (P2, pea)-142.0 (P22, soy) µg·kg −1 was recorded for a single serving. EDI of samples ranged from 0.8 to 4.6 x10 −2 µg·day −1 for a 70 kg person. These daily intake values fall well below the recommended dietary allowance (RDA) suggested previously (1000 µg Li·day −1 ) [32] and indicate the intake of Li from protein powder samples is negligible.

Molybdenum ( 95 Mo)
Mo serves an essential role as a cofactor in the active site of mammalian enzymes (Moco), where it acts as a catalyst for substrate redox reactions [33]. The RDA for Mo is 45 µg·day −1 with an average intake by adults between 76 and 109 µg·day −1 [34]. Mo concentrations of 100.2 (P31, whey)-2348.7 (P12, pea) µg·kg −1 were recorded in samples, which are comparable with recent studies: 60-1710 µg·kg −1 [13] and 500-810 µg·kg −1 [15]. The EDI of Mo from samples analysed in this study was calculated to be between 0.04 and 1.13 µg·day −1 for a 70 kg person, which equates to a maximum %RDA of 2.5% and is well below the UL of 2000 µg·day −1 [34].

Platinum ( 195 Pt)
While Pt has been used as a therapeutic agent in the treatment of cancer [1], the risk it poses in food has not been thoroughly examined. A concentration range of 12.4 (P31, whey)-102.7 (P2, pea) µg Pt·kg −1 was determined in samples, with an estimated intake of <0.04 µg·day −1 (70 kg person, single serving).

Gold ( 197 Au)
Au was found in all protein powder samples at concentrations between 5.3 (P20, whey) and 26.2 (P8, whey) µg·kg −1 . Previously Au has been used in the treatment of rheumatoid arthritis [35]; however, its safety and overall biological function has been questioned [36]. Au is used as a food additive (E 175), and the European Food Safety Authority (EFSA) determined in 2016 that due to the low solubility and systemic availability of elemental Au, adverse health effects are not expected for consumers [37]. Intake of Au from consumption of a single serving of these samples for a 70 kg person was estimated to be <0.04 µg·day −1 .

Magnesium ( 24 Mg)
Mg is instrumental in energy metabolism and protein synthesis as well as physiologically supporting brain, heart and skeletal muscle development and repair, making it one of the most essential elements in maintaining overall health [38]. The RDA for Mg is 6000 µg·day −1 [39], and the UL for Mg is 350,000 µg·day −1 [40]; however, accurate determinations for Mg concentration in the protein powder samples could not be made, as the concentration of Mg in all samples exceeded the upper point on the calibration plot, and further dilution of the samples was not possible. Hence, Mg was not reported for these samples (NR*).

Vanadium ( 51 V)
V can be found in a variety of foods including mushrooms, shellfish and processed foodstuffs, and interest in the research of vanadium compounds as therapeutic agents is on the rise [34,41]. A daily intake of <1800 µg V·day −1 has been advised [42]. Concentrations of V in samples were determined in the range of <0.6 (P31, whey)-78.6 (P36, blend) µg·kg −1 , which is in line with results seen in Pinto et al. (2020) of up to 109.8 µg V·kg −1 across 49 samples tested. A corresponding intake range for V in this study was estimated between 0.00 and 0.06 µg·day −1 for a single serving of protein powder (70 kg person). An UL for V has been reported as 18,000 µg·day −1 [34].

Chromium ( 52 Cr)
With an influential role in the metabolism of carbohydrates, lipids and proteins, Cr is an essential component of diet and has an RDA of 20-35 µg·day −1 [43]. Recent studies by

Iron ( 56 Fe)
Fe is an essential component for health, owing to its role in the synthesis of haemoglobin and myoglobin, which are proteins responsible for the transportation of oxygen around the body. A number of enzymes involved in electron transfer and oxidative-reductions also contain Fe [44]. The RDA for Fe is 8000 µg·day −1 (men, post-menopausal women) and 18,000 µg·day −1 (pre-menopausal women) [34]. The EDI of iron from consumption of these samples was calculated to be between 0.69 and 4.17 µg·day −1 (based on a single serving for a 70 kg person), which is well below the UL for Fe of 45,000 µg·day −1 . When analysed, many samples had concentrations that exceeded the highest point on the calibration range, and because samples could not be further diluted, the levels of Fe in these samples were not reported (NR*). Of those samples whose concentrations fell within the calibration range, concentrations between 2550.5 (P35, whey) and 7456.8 (P8, whey) µg Fe·kg

Cobalt ( 59 Co)
Though many Co compounds can have a toxic effect with excessive exposure, Co serves an important biological function as a component of cyanocobalamin (Vitamin B12), which is involved in facilitating red blood cell production, supporting the nervous system and the release of energy from food (respiration) [45,46]. Between 2.5 (P31, whey) and 188.5 (P2, pea) µg Co·kg −1 was recorded in the protein powder samples tested, with an estimated intake of 0.0-0.1 µg Co·day −1 (single serving, 70 kg person). Analysis of protein supplements by Pinto et al. (2020) [13] yielded Co concentrations in the range of 10-134 µg·kg −1 .

Copper ( 63 Cu)
Cu is involved in significant redox reactions in the body and serves in the synthesis of neurotransmitters, the production of melanin, antioxidant defence and the development of bone tissue [43]. The RDA for Cu is approximately 900 µg·day −1 [34], which is significantly higher than the EDI values recorded for a 70 kg person consuming a single serving of the samples investigated in this study (0.2-3.8 µg Cu·day −1 ). The concentrations recorded in samples ranged from 573.5 (P31, whey) to 6523.8 (P17, whey) µg Cu·kg −1 , which is in line with concentration ranges previously reported by Pinto et al. (370-10,500 µg·kg −1 ) and Muller et al. (260-6100 µg·kg −1 ). None of the samples tested exceeded the UL for Cu of 10,000 µg·day −1 .

Tin ( 118 Sn)
The negative influence of Sn on the human body is less to do with its own toxicity and more to do with the its influence on the absorption of Cu, Fe and Zn, where it can lead to deficiency symptoms of those elements [47]. The majority of results for Sn obtained from the analysis of the protein powder samples were below the LOQ, corresponding to a final concentration of <17.4 µg·kg −1 . Four samples with concentrations within the calibration range recorded final Sn concentrations of 17.7 (P1, mixed plant)-329.7 (P12, pea) µg·kg −1 , which is below the reported RfD for Sn of 600 µg·kg −1 BW·day −1 [14,26]. A maximum exposure, based on a single serving for a 70 kg person, was estimated at 0.05 µg Sn·day −1 , which is lower than the results seen in Elgammal et al. (2019), who estimated Sn exposures between 0.6 and 1.3 µg Sn·day −1 .

Bismuth ( 209 Bi)
Though utilised for some medicinal purposes, Bi can be toxic at elevated concentrations, resulting in fever, weakness, rheumatic pains and diarrhoea [48]. Concentrations of Bi in samples ranged from <0.1 (P22, soy and P29, mixed plant) to 38.9 (P14, whey) µg·kg −1 , with a maximum daily exposure to Bi of 0.02 µg·day −1 (based on a single serving of protein powder for a 70 kg person). Given the no-observed-adverse-effect level (NOAEL) of 1,000,000 µg Bi·kg −1 [49], Bi concentrations in the protein powder samples tested are negligible and therefore not a concern.

Titanium ( 47 Ti)
Ti is often used in dental implants and other biomedical devices, owing to its reputation as a relatively inert metal. However, harmful reactions in humans (including hypersensitivity and allergic reactions such as facial eczema) can occur as a result of device failure [50]. In this study, concentrations between 29.7 (P6, whey) and 2313.7 (P34, whey) µg Ti·kg −1 were recorded in samples, with exposures of 0.01-0.99 µg·day −1 , based on a single serving of protein powder for a 70 kg person. Values for RfD or UL have not been established.

Beryllium ( 9 Be)
Be is a known carcinogen and can potentially result in gastrointestinal lesions [25,51]. The majority of protein powder samples analysed were below the LOQ, with resulting final concentrations of <0.9 µg·kg −1 . Two samples, P1 (mixed plant) and P12 (pea), recorded final Be concentrations of 5.6 µg·kg −1 and 5.5 µg·kg −1 , respectively. For a 70 kg person consuming a single daily serving of protein powder, the maximum exposure to Be was estimated at 4.8 × 10 −3 µg·day −1 , which represents 0.24% of the RfD (2 µg·kg −1 BW·day −1 ) [25]. Hence, exposure through consumption of protein powder is not a concern.
3.4.2. Cadmium ( 111 Cd) Like Be, Cd has been classified as a known carcinogen, and its toxicity has been reported to result in additional negative health effects that include damage to the reproductive, renal, skeletal and nervous systems [51][52][53]. Protein powder samples analysed yielded Cd concentrations in the range of 0.6-58.1 µg·kg −1 , with a maximum exposure to a 70 kg person from a single serving estimated to be 0.05 µg Cd·day −1 . Exposure at this level represents just 5% of the reported RfD for Cd (1 µg·kg −1 BW·day −1 ) [14,27]. Previous studies by Pinto [25], which is well above EDI range of 0.04-2.24 µg·day −1 determined for these samples (70 kg person, single serving of protein powder).

Mercury ( 202 Hg)
As a heavy metal, elemental (inorganic) Hg can be very toxic to humans. Acute Hg poisoning has been linked to disorders of the nervous and gastrointestinal systems and can result in death [55]. The concentration of elemental Hg determined in protein powder samples ranged from <0.2 to 18.4 (P1, mixed plant) µg·kg −1 , which corroborates results seen in Pinto et al. (0.7-23.9 µg·kg −1 ). Based on this, a daily exposure of up to 0.02 µg Hg·day −1 was estimated from consumption of a single serving of protein powder, which equates to a maximum of 6.7% of the RfD for Hg (0.3 µg·kg −1 BW·day −1 ) [14,27].

Thallium ( 205 Tl)
Due to its high toxicity, Tl is frequently used as a rodenticide and insecticide. In humans, Tl can enter the system through dermal/inhalation exposure routes as well as through accidental ingestion. Among the symptoms of Tl toxicity are abdominal pain, nausea/vomiting/diarrhoea/constipation, headaches, tremors and seizures. Doses of 10,000-15,000 µg·kg −1 result in death for humans, though lower concentrations can also result in coma and death [56]. In the samples tested, Tl concentrations of 0.02 (P35, whey)-3.4 (P34, whey) µg·kg −1 were recorded. In 2012, a provisional RfD of 0.01 µg Tl·kg −1 BW·day −1 was established by the US EPA as part of the provisional peer-reviewed toxicity values (PPRTVs) assessment. All samples tested were below this provisional limit, with maximum exposure for a 70 kg person estimated to be 26.6 × 10 −4 µg Tl·day −1 .

Lead ( 208 Pb)
Pb is a heavy metal that is highly toxic and has been classified as a probable carcinogen by the IARC [51]. The nervous system is the primary target of Pb poisoning, though other symptoms include anaemia and kidney damage, as well as damage to the immune and reproductive systems [43]. In previous studies, Pb has been found in protein powder samples at concentrations such as <1.0-31. An RfD value of 1000 µg·kg −1 BW·day −1 was previously established for Al [14,26], which means the calculated maximum exposure (4.17 µg Al·day −1 ) represents 0.4% of the RfD, posing little risk to consumers.

Health Risk Assessment
An assessment of the potential risk to human health from oral exposure to the elements investigated in this study was carried out with reference to US EPA recommendations and previous exposure assessment studies [14,26,29,57]. Using oral reference dose (RfD) values listed in Table 5, general health risk was examined through the calculation of the hazard (HQ) and hazard index (HI). Carcinogenic health risk was estimated using the estimated daily intake (EDI) values of elements along with their carcinogenic slope factor (CSF), where applicable (also listed in Table 5).

Non-Carcinogenic Risk
HQ characterises potential risk to health from exposure to toxic substances by relating the EDI of elements in samples with their respective RfD value. As part of the HQ assessment (the results of which can be found in Supplementary Table S39), risks associated with 1 and 3 servings of protein powder were investigated for a 70 kg person. While an RfD value has been established for Cr(VI), one has not been established for total Cr (which was determined in this study). Hence, Cr was not included in the non-carcinogenic risk assessment.
It was noted that none of the samples analysed yielded element HQs >1, indicating that adverse health effects from individual toxic or potentially toxic elements present in the protein powder samples are unlikely.
Because of the potential for interactive/additive effects where more than one toxic/ potentially toxic element is present in a sample, HI was estimated based on the addition of the HQs (see Table 7). With respect to a single serving of protein powder for a 70 kg person, none of the samples tested yielded a HI value > 1. When the number of servings of protein powder is increased to three per day, the number of samples whose HI value is >1 increases to ten, which includes mixed plant (P1), pea (P2), casein (P19) and whey protein samples (P9, P13, P16-18, P21 and P34). Thus, it can be inferred that while the products are generally safe when taken in moderation, excessive consumption of these samples over time may increase the potential risk of non-carcinogenic health implications.

Carcinogenic Risk
Cancer risk (CR) from prolonged exposure to known and probable carcinogenic elements was investigated, to include Be, Cd and Pb. Though Cr(VI) is classified as a known carcinogen and has an established CSF value, only total Cr was determined in samples for this study. Because a CSF value has not been established for total Cr, it was not included in the carcinogenic risk assessment. CR was assessed for a 70 kg person consuming 1 or 3 servings of protein powder, and the results can be found in Table 8. Additionally, samples with the highest recorded CR values for each of these elements are highlighted in Table 9. Given that thirty-four of the thirty-six samples tested returned Be concentrations below the LOQ, it was expected that the carcinogenic risk from Be consumption through protein powder use would be low. All samples tested yielded CR ratios ≈ 10 −6 for both 1 and 3 servings for a 70 kg person, thus supporting the hypothesis that the risk of developing cancer from prolonged exposure to Be through consumption of protein powder is low (1:1,000,000 risk). The highest recorded CR values were observed in sample P1 (mixed plant), with CR values of 1.1 × 10 −6 (single daily serving) and 3.3 × 10 −6 (three daily servings) for a 70 kg person, which again does not indicate any significant risk.
For Cd, all samples recorded CR values ≤ 10 −4 , suggesting that the concentrations of Cd found in protein powder generally does not increase the risk of developing cancer. The highest CR value recorded for a single serving of these powders was for sample P1 (mixed plant), where CR was determined to be 1.3 × 10 −4 . When the number of servings was increased to three per day, four samples recorded CR levels between 1.3 × 10 −4 and 3.9 × 10 −4 , including P1 (mixed plant), P2 (pea), P29 (mixed plant) and P36 (blend). Though each of these CR values are acceptable, this data would suggest that prolonged excessive use of protein powder products by consumers has the potential to increase carcinogenic risks associated with lifetime exposure to low levels of Cd.   CR values of ≤10 −4 were noted in twenty-six of the samples tested for Pb (one daily serving for a 70 kg person), indicating that Pb cancer risk in these samples is negligible. The remaining ten samples (27.8% of samples investigated) recorded slightly elevated CR values for Pb that exceeded the CR ≤ 10 −4 threshold, in the range of 1.1 × 10 −3 -5.0 × 10 −3 . Among these samples were casein (P19), pea (P2), mixed plant (P1, P29), blend (P33, P36) and whey (P18, P20, P28, P34), and the results may suggest a marginally higher cancer risk with prolonged consumption of these ten protein powders. When increasing the number of daily servings to three, the number of samples that exceeded the CR ≤ 10 −4 threshold rose to twenty-four (66.7%). In general, while the determined concentrations of Pb in the investigated protein powder samples were low, the evidence suggests that excessive consumption of these products over long periods of time may increase the possibility of carcinogenic effects.

Conclusions
In this study, various types of protein powder samples were analysed, and the concentrations of elements that are classified as essential/therapeutic (Li, Mo, Pt, Au, V, Cr, Mn, Fe, Co, Cu), non-essential/potentially toxic (Sn, Bi, Ti) and toxic (Be, Cd, Ba, Hg, Tl, Pb, Al) were determined. All samples underwent microwave digestion prior to analysis by ICP-SFMS. The maximum concentrations determined for each element (µg·kg −1 ) were as follows: Fe (7456. Calculation of EDI and HQ revealed that all samples investigated had element levels below tolerable limits, including those established and/or reported by the U.S. EPA/PPRTV, E.U. SCCS and CDC/ATSDR. Similarly, the combination of HQs for elements in each sample showed that none of the samples had an HI value >1 after one serving of protein powder. Increasing the number of daily servings to three resulted in 10 protein powder samples with an HI > 1. The risk of developing cancer throughout a lifetime as a result of prolonged exposure to these protein powders was also assessed as cancer risk (CR), using oral carcinogenic slope factors (CSFs). The results from a single daily serving suggest that the levels of Pb in ten of the samples (while low) may still represent a slightly elevated risk of cancer development when taken for prolonged periods of time (CR values > 10 −4 ). As with the HI assessment, increasing the daily servings to three saw the number of samples with CR values > 10 −4 rise to twenty-four.
Based on the data presented, it can be concluded that the protein powder samples investigated in this study are generally safe for consumption, when taken in moderation. However, as is evidenced by the rising HI and CR values when the number of daily servings is increased to three, it is advised that consumers of these products follow the recommended number of daily servings stated by the manufacturer to avoid incurring negative health effects from prolonged use.   (2)). In addition, the authors acknowledge funding from Cork Institute of Technology/Munster Technological University through the RÍSAM postgraduate scholarship. Finally, the authors acknowledge funding from the Higher Education Authority of Ireland, The Programme for Research in Third Level Institutions (HEA-PRTLI), which supported the following projects: 'ASTOX, ZEBRATOX and AZP' projects (HEA-PRTLI 2, 2002) and the 'Environment and Climate Change: Impacts and Responses' project (HEA-PRTLI 4, 2007) through funding of the research laboratory infrastructure. The funders had no role in the design of the study; in the collection, analyse or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study is contained within this article and is supported by data in the Supplementary Materials.