Abstract
Concerns have recently been raised regarding the presence of arsenic in California wine. Total arsenic concentrations in 101 wines produced or bottled in California were characterized using inductively coupled plasma mass spectrometry. Of these wines, 28 were wines identified in media reports as containing arsenic concentrations greater than the U.S. Environmental Protection Agency Maximum Contaminant Level for drinking water of 10 μg/L. The remaining 73 wines were randomly purchased at local retailers. Blush wines were found to contain the greatest total arsenic concentration (mean = 27.2 μg/L; SD = 16.9 μg/L) regardless of sampling group, followed by white wines (mean = 10.9 μg/L; SD = 11.0 μg/L) and red wines (mean = 6.75 μg/L; SD = 7.33 μg/L). Moreover, the publicized group of wines was found to have significantly (p < 0.05) greater total arsenic concentrations (mean = 25.6 μg/L; SD = 12.0 μg/L) than the random wines (mean = 7.42 μg/L; SD = 10.2 μg/L). The concentrations of total arsenic in all wines evaluated in this analysis were less than the two currently used arsenic guidelines for wine of 100 μg/L and 200 μg/L. Results from the statistical analysis suggest that no more than 0.3% of California wines (if any) may contain arsenic concentrations greater than the 100 μg/L guideline. A significant inverse association between total arsenic concentration and unit wine price was also identified in this analysis. Chronic daily intake of arsenic as a result of wine consumption was estimated to account for a small fraction (<8.3%) of a typical adult’s dietary arsenic intake, indicating that wine consumption is not a significant source of total arsenic exposure. These results indicate that the presence of arsenic in wine does not represent a health risk for consumers.
Arsenic (As) is an environmentally-stable metalloid that is ubiquitous in food, water, soil, and air from both natural and anthropogenic sources (Klaassen 2013). Increased scientific study of the As content of foods and beverages, such as apple juice and rice, has led to intensified regulatory and consumer interest regarding the As content of food (Jara and Winter 2014, Moreno et al. 2000, Tvermoes et al. 2014, Uneyama et al. 2007). Recent media reports have raised concern regarding elevated As concentrations in California wines (CBS News 2015, Haelle 2015). The independent commercial laboratory that measured the As concentrations in wine cited by the media reportedly tested over 1300 different wines (BeverageGrades 2015, Tainted Wine 2015). Some of these reports have noted that as many as 83 of the wines tested contained As concentrations that exceeded the U.S. Environmental Protection Agency (USEPA) drinking water Maximum Contaminant Level (MCL) for As (Haelle 2015, BeverageGrades 2015, Tainted Wine 2015). Media reports have also suggested that price is potentially correlated with As concentration, with less expensive wines containing greater concentrations of As (CBS News 2015, Haelle 2015). The resulting publicity has led some Americans to question whether the As present in these wines poses a health hazard.
The presence of As in wine has been known for decades (Aguilar et al. 1987, Crecelius 1977, Schroeder and Balassa 1966). The As content in wine may result from a number of sources, including naturally-occurring As in soil, industrial emissions, and the historical use of As in a wide range of agricultural practices (Handson 1984, Hopfer et al. 2015, Moreno et al. 2000, Rodrigues et al. 2011, Tariba 2011). Arsenic concentrations in environmental media, such as groundwater used for winemaking, have also been reported to impact the final As content in wine (Hopfer et al. 2015, Kunkee and Eschnaur 2002). In addition to environmental sources, winemaking practices may influence the concentration of As in wine. These practices include the use of bentonite as a filtering agent, grape-pressing processes, contact with juice lees, and contact with metals in the production equipment (Barbaste et al. 2003, Bertoldi et al. 2013, Hopfer et al. 2015, Kunkee and Eschnaur 2002).
Arsenic naturally exists in different chemical forms, including the inorganic species arsenite [As(III)] and arsenate [As(V)], and the organic species, monomethylarsonic acid (MMA) and dimethylarsinic acid (DMA). These species all differ in their biological properties and toxicity; specifically, inorganic forms of As are known to be more toxic to humans than organic forms of As (Herce-Pagliai et al. 2002). Previous research has established that “inorganic and organic arsenicals exhibit distinctly different toxicokinetic characteristics” that result in inorganic As forms being more toxic than organic forms of As (ATSDR 2007, p. 17). The toxicity of As depends on the form of As (organic versus inorganic) that is present, as well as the matrix (e.g., air, food, water), the route of exposure (inhalation versus ingestion), and magnitude of exposure (Herce-Pagliai et al. 2002).
A number of federal and international agencies currently regulate the amount of As in drinking water and beverages to protect public health because of the known toxicity of As following ingestion of high doses. In 2001, the USEPA published an MCL for total As of 10 μg/L for drinking water (USEPA 2001). In addition, a limit of 10 μg/L inorganic As was adopted by the FDA for bottled water and apple juice in 2005 and 2013, respectively (USFDA 2013, USFDA 2014).
To date, the U.S. has not established a maximum acceptable concentration for As in wine. However, international organizations, including Health Canada and the International Organisation of Vine and Wine (OIV), have set limits for total As in wine. The OIV is an intergovernmental scientific and technical organization comprised of 46 member states throughout Europe, Africa, South America, Australia, and Asia that works to inform its members and integrate existing standards and practices regarding vine and wine products (OIV 2015). The OIV has recommended a maximum acceptable limit for total As in wine of 200 μg/L, and Health Canada has established a maximum total As standard for beverages, including wine, of 100 μg/L (OIV 2011, VQA 2015).
Given the previously published reports of elevated As concentration in wine and the accompanying recent media attention, the objectives of the current study were to: 1) characterize the total As concentrations in a random sample of commercially available California wines, as well as the subset of publicized California wines identified in media reports; 2) estimate the contribution of As in wine to total dietary As consumption; and 3) assess whether there is a relationship between total As concentration and price in California wines. Findings were catagorized by wine type (red, white, or blush), and total As concentrations were compared to international maximum acceptable limits for As in wine.
Materials and Methods
Sample collection
A total of 101 different labels (winemaker and varietal) of bottled and boxed wines produced or bottled in California were purchased in Pennsylvania and New York. Of these wines, 28 were referred to as the “publicized wines” group because they were identified in a number of recent media reports as containing elevated concentrations of As. A total of 83 types of wine were cited in the media as “high arsenic wines,” and the 28 wines selected were a convenience sample based on availability in Pennsylvania and New York, and represented the population of the 83 publicized wines.
After collecting the publicized wines, 73 additional California wines available in Pennsylvania stores were selected so as to have a diverse sample of commercially available California wines (herein referred to as the “random wines” group). The 73 random wines were purchased in two batches. The first 52 samples were selected as a convenience sample stratified by type of wine based on in-store availability at local retail outlets. The remaining 21 samples were selected as a random sample from a list of wines for sale in stores stratified by price range and type of wine. The random wines were selected in this manner to determine whether the As concentrations cited in media reports were representative of the As concentrations in California wines as a whole. The random wine sample consisted of 42% red, 44% white, and 14% blush wine labels, which is a similar distribution to recent consumer market research that indicated that red, white, and blush wines account for 47, 40, and 13% of wines sold at retail stores, respectively (Hogden 2011).
The randomly sampled wines were priced between $3.00/L and $213.31/L, with an average unit price of $34.03/L (median = $22.65/L). Publicized wines were priced between $3.60/L and $21.32/L, with an average unit price of $9.22/L (median = $8.32/L). Three to four separate bottles for nine of the wines were purchased to assess the variability of As concentration within identical labels. A total of 121 boxes or bottles of wine were thus evaluated in this study. All wines were stored in their original containers at room temperature until analyzed. Each wine was assayed within five days of purchase.
Sample preparation
A total of 121 samples (wine bottles or other containers) were analyzed by the RJ Lee Group (Monroeville, PA). For each sample, wine aliquots of 10 mL were digested and analyzed using a Perkin Elmer NexION 300X inductively coupled plasma mass spectrometry system (ICP-MS). Sample preparations were performed in laminar flow clean hoods free of trace metal contamination. From each bottle of wine, a 10 mL sample was collected into a 50 mL metal-free polypropylene centrifuge vial (Labcon North American, ref. no. 3194-335-001). This 10 mL sample was then evaporated to 5 mL over a two-hour period using a CPI International brand MODBlock hot block set at 110°C. After evaporation, 2 mL of concentrated nitric acid (HNO3; Trace Metal Grade 67–70% HNO3, Fisher Brand no. A509-P212) and two to three drops of hydrogen peroxide (H2O2; Certified ACS, 30–32% H2O2, Fisher Brand no. H325-500) were added to the polypropylene vials. The samples were placed back in the same hot block, still set at 110°C, covered with a watch glass, and allowed to digest for 45 min. As part of the preparation, the following quality control samples were prepared for every 20 samples: a reagent blank (LRB), a fortified reagent blank (LFB), a matrix duplicate (MD), and a matrix spike (MS). The LFB and MS quality control samples were spiked with a known amount of an As National Institute of Standards and Technology (NIST)-traceable standard so that the final concentration of the spike in the sample would equal 10 μg/L As. A standard reference material that represents As in a wine matrix was not available for this study.
Quantification of total arsenic
The resulting digestates were diluted to a final volume of 10 mL with double-deionized water and were analyzed by ICP-MS following the principles of the EPA SW846-6020/6020A protocols. The NexION 300X ICP-MS used in this study was equipped with a collision cell that utilizes kinetic energy discrimination with helium as the collision gas. Data for total As were collected using that mode along with the standard mode in which interference correction equations were applied. Samples were diluted four-fold before analysis, similar to other studies that have assayed trace contaminants in wine (Stacey et al. 2014). The four-fold dilution mitigated matrix interferences that could arise in the digested wine sample. The calibration standards, verification standards, and verification samples were prepared with purchased NIST-traceable stock solutions. An internal standard was monitored throughout the analysis of all samples; the internal standard contained germanium, indium, and rhodium. Each internal standard was applied to the As mass in both the standard and kinetic energy mode. Internal standard recovery in the samples was monitored to be within 60 to 125% of the calibration blank. An average of three measurement readings was reported for all samples, and calibration, calibration verification, sample analysis, and sample-analysis verification were performed. A set of five calibration standards were prepared from the purchased NIST-traceable stock solutions and were analyzed for each analytical run. Calibration curves were validated at a correlation coefficient of 0.998 or greater. The lowest calibration standard of 0.5 μg/L limited measurement of total As in the diluted digests to 0.5 μg/L and the minimum reporting limit in the wine samples to 2.00 μg/L.
Quality assurance and quality control
A 50 μg/L initial calibration verification standard was prepared from an independent NIST-traceable source and was analyzed at the beginning of the analytical runs, with a ±10% acceptance criteria. A reporting limit or low-level verification standard of 0.5 μg/L was also analyzed at the beginning of the analytical runs, with an acceptance criteria of ±30% of the nominal value. The As concentration of the initial calibration blank sample was also analyzed at the beginning of the analytical runs and was verified to be below the reporting limit of 0.5 μg/L. Unspiked and spiked interference check standards were analyzed at the beginning of each calibration. The interference standards were monitored to ensure that the interference correction equations and the kinetic energy discrimination removed interferences that might occur in the matrix, the acceptance criteria for the interference standards was ±20%. A continuing calibration verification standard, prepared from the same stock as the calibration standards, was analyzed at a rate of one for every 10 sample analyses. Acceptance criterion for the continuing calibration verification standard was ±10% of the nominal value of 50 μg/L. A continuing calibration blank solution was also analyzed every 10 samples. This blank solution was verified to have a result below the reporting limit of 0.5 μg/L.
A total of nine sets of quality control samples were prepared. The nine LRB quality control samples were verified to have results that were below the analytical limit of detection of 0.5 μg/L. The nine LFB quality control samples showed total As recoveries between 80.5 and 93% of the spiked nominal value of 10 μg/L total As. For the nine MD samples (sample duplicates consisted of two samples taken from a single bottle of wine), the relative percent difference was less than 22%. For the nine MS samples (sample matrix spikes were an aliquot of a wine sample that was spiked with a NIST-traceable solution), spike recovery ranged between 78.7 and 124.2% of the spiked nominal value of 10 μg/L As. Additional quality control samples were analyzed in four wines that included postdigestion spikes (PDS), by which the four-fold diluted aliquot of the digested wine sample was spiked with a NIST-traceable As standard so that the nominal value of the spike in solution would be 2.5 μg/L, and also dilution samples, by which the diluted four-fold sample was additionally diluted by five-fold. The four PDS sample recoveries varied between 89.3 and 110.3%, while the dilution test sample recovered at 90% of the original sample value at the four-fold dilution.
Statistical analysis of As concentrations in wine
The total As concentration was reported for each wine label included in the study. Statistical analysis was used to compare the total As concentrations measured for the random and publicized wines. The coefficient of variation for As concentrations measured in the nine wine labels analyzed in triplicate ranged between 0.03 and 0.34, indicating that the As concentrations were consistent between samples within each wine label. The average concentration of total As in wine labels purchased in triplicate was therefore used for all subsequent statistical analyses. Additionally, statistical analysis was performed within each group to compare total As concentration by wine type (red, white, and blush). Data analysis and transformation was completed using Microsoft Excel in order to determine the distribution of As concentrations (Ver. 2013, Microsoft Corporation).
Simple distribution analysis demonstrated that total As concentrations in the random wine group followed a log-normal distribution. Significant differences in average total As concentrations between wine type were determined by unpooled Student’s t tests, assuming unequal variance using Microsoft Excel. The t tests were performed with the log-transformed As concentration values for each sample. Linear regression analysis was performed to assess the correlation between unit price ($/L) and total As concentration for wines grouped by type (red, white, and blush) and by sampling group (random, publicized). Statistical significance was defined as p < 0.05. For statistical purposes, As concentrations less than the detection limit were treated as if the concentration was one-half of the minimum reporting limit (MRL) in order to avoid underestimation of the As content (1/2 MRL = 1.00 μg/L) (Ignacio and Bullock 2006).
Total As exposure estimation
The chronic daily intake (CDI) was used to characterize the exposure to As resulting from wine consumption. This approach is consistent with what has been used to evaluate exposure to metals from other beverages (Tvermoes et al. 2014). The following equation was used to determine the CDI based on the total As concentrations determined in this study (USEPA 1992):
where CDI is the chronic daily intake (μg/kg-day), C is the total As concentration found in the wine (μg/L), DI is the average daily intake rate of wine (L/day), and BW is the body weight (kg) of an individual. The product of C and DI represents daily As intake (mg/day). An annual wine consumption rate of 9.99 L (roughly 67 five-ounce glasses of wine per year) was used to estimate the daily intake of wine. This volume was determined based on data reported in the most recent National Institutes of Health Surveillance Report (LaVallee et al. 2014). The body weight for the various age groups was adapted from the USEPA Exposure Factors Handbook (USEPA 2011).
Results
Total As concentration by wine type
The arithmetic mean, SD, geometric mean, and geometric standard deviation (GSD) of total As by group and wine type are provided in Table 1. Random wines contained significantly lower total As concentrations than publicized wines. The overall mean As concentrations for blush, white, and red wines were 27.2 μg/L (SD = 16.9 μg/L), 10.9 μg/L (SD = 11.0 μg/L), and 6.75 μg/L (SD = 7.33 μg/L), respectively. Random wines contained significantly lower total As concentrations than publicized wines (p < 0.0001). The mean total As concentration in random wines was 7.42 μg/L (SD = 10.2 μg/L), while the average total As concentration in publicized wines was 25.6 μg/L (SD = 12.0 μg/L). Furthermore, the total As concentration in all wines assayed were found to follow a lognormal distribution. The geometric mean total As concentration in the random group of wines was 3.73 μg/L (GSD = 3.27). Using these results, three out of every 1000 California wines (0.3%) would be anticipated to contain total As concentrations of 100 μg/L or greater. However, in this analysis, no wine was found to contain total As at a concentration greater than 69 μg/L; therefore, this result is extrapolation beyond the maximum, and it is possible that no wines would contain As concentrations greater than 100 μg/L.
Total arsenic concentration by wine group.
Total As concentration by wine group
Statistical t tests on the log-transformed As concentration values indicated that random blush wines contained significantly greater concentrations of total As than random white wines and random red wines (p < 0.05 in both instances), and random white wines contained significantly greater concentrations of total As than random red wines (p < 0.05). Within the group of publicized wines, blush wines contained significantly greater concentrations of total As compared with red wines (p < 0.05). Differences between the means of the publicized wine groups were observed, but were not statistically significant. The publicized blush wines contained greater concentrations of total As compared with publicized white wines (p = 0.10), and publicized white wines contained greater concentrations of total As compared with publicized red wines (p = 0.12).
Total As concentration as a function of price
Figure 1 presents the total As concentrations measured in the 101 brands of California wine as a function of unit price (price per liter of wine). Arsenic was not detected in eight random white wines and 18 random red wines (25% of all wines sampled). Total As concentrations ranged from nondetect (MRL = 2.00 μg/L) to 68.4 μg/L. The correlation between As concentration and price was assessed for: 1) all wines in this analysis (R2 = 0.571, p < 0.05); 2) random wines (R2 = 0.463, p < 0.05); and 3) publicized wines (R2 = 0.0364, p = 0.33). A statistically significant correlation between total As concentration and price existed when: 1) all wines were included in the analysis; and 2) the wines in the randomly selected group were analyzed. No association between total As concentration and price was observed in the publicized wines.
Total arsenic concentration in sampled wines as a function of price per unit volume ($/L). The arsenic content of random wines was found to follow a statistically significant (p < 0.05) relationship with unit price. No association was observed in the publicized wine group. One very high-cost wine ($213.31/L, 2.69 μg/L) was omitted from the figure, but included in the power law analysis.
Figure 2 presents the distribution of total As concentration of wines in both the publicized and random groups for similar price ranges. Because all publicized wines in this study cost less than $22/L, wines that cost less than $22/L in the randomly sampled group were considered separately from the rest of the randomly sampled wines. As shown in Figure 2, the total As concentration in the publicized wines is significantly higher than wines in the random sample that cost less than $22/L in red, white, and blush wines assessed in this analysis (p < 0.05). Furthermore, the total As concentrations in red and white randomly sampled wines that cost more than $22/L were significantly lower than both the random and publicized wines that cost less than $22/L.
Distribution of total arsenic concentration in sampled wines as a function of price per unit volume ($/L) per wine type category. All publicized wines had a price per unit volume of $22/L or less, and more than 50% of wines purchased in the United States cost less than $20/L (Thach et al. 2014). Therefore, $22/L was selected as the cutoff between low- and high-cost wines to facilitate comparison of the publicized and random groups.
Exposure estimate for total As in wine
Table 2 contains the estimated CDIs for total As from wine consumption that were calculated based on the mean As concentrations. The daily dose for the population’s average total dietary As intake was estimated on a per-body-weight basis using the average male and female values presented by Tao and Bolger (1999) and Jara and Winter (2014), and taken as presented by Xue et al. (2010). Estimated daily-dose values are presented for the publicized wines, low-cost random wines, and high-cost random wines. The estimated total As daily-dose values for each wine group ranged between 0.00061 and 0.012 μg/kg-day.
Chronic daily intake (CDI) of total arsenic in sampled California wines and comparison to literature-based mean arsenic intake rates.
Discussion
Numerous studies have assessed the presence of As and other heavy metals in food and beverages, including juices, wine, and beer (Aguilar et al. 1987, Barbaste et al. 2003, Lynch et al. 2014, Tvermoes et al. 2014, USFDA 2013). The presence of metals in beverages can arise from natural sources in the soil, other practices including the application of pesticides on crops, or inadvertent contamination from the manufacturing process.
Analysis of total As content by wine type
Growing practices, enological practices, and environmental contamination may contribute to total As content in wine, as well as to the differences in As content in different types of wine (Almeida and Vasconcelos 2003, Barbaste et al. 2003, Bertoldi et al. 2013, Hopfer et al. 2015). In California, concentrations of As in both soil and groundwater have been reported to follow a lognormal distribution, indicating that inadvertent environmental contamination may influence the total As content in California wines (Bradford et al. 1996, USGS 2000). The distribution of total As concentrations in all California wines assessed in this analysis were found to be lognormal. The distribution of total As in these California wines is not surprising, given that the presence of minerals in soil and plants has been well established and is known to be closely tied to the geological composition of the underlying rock, the physical and chemical properties of the soil, and the ability of a plant to take up and accumulate various elements (Singh et al. 1997). Bertoldi et al. (2011) noted that the concentrations of metals in grapes from two different vineyards in Italy were “probably related to the different availability of minerals, particularly trace elements, in the two soils” (p. 7232). Other researchers have also noted that the elemental composition of wine is influenced by the solubility of inorganic compounds in the soil of vineyards (Greenough et al. 1997).
Most data indicate that the type of winegrape impacts the elemental composition of wines. Yang et al. (2010) reported that the grapes for red and white wines have significantly different elemental concentrations, which in part, likely explains the difference in total As concentration between the white and red California wines found in this analysis. The ripeness of the grapes may also affect the total As content in wine. Bertoldi et al. (2011) reported that the total As content progressively increases in the grape as it develops and matures; at ripeness, As was primarily localized in the pulp and skin of the grape, with ~10% of the As in the seeds.
Previous analyses have also reported that blush and white wines typically contain a higher total As concentration than red wines, regardless of the wine’s country of origin (Aguilar et al. 1987, Barbaste et al. 2003, Baxter et al. 1997, Crecelius 1977, Escudero et al. 2013, Fiket et al. 2011, Galani-Nikolakaki et al. 2002, Kment et al. 2005, Rodrigues et al. 2011). This finding suggests that factors other than inadvertent environmental contamination and naturally occurring As may also be responsible for the variation in total As content among the California wines reported in the current analysis. Processes that may impact As content include the fermentation process, as well as the addition of clarifying and refining agents to the wine (Aguilar et al. 1987, Almeida and Vasconcelos 2003, Crecelius 1977, Huang et al. 2012, Tariba 2011, Bertoldi et al. 2013). Aguilar et al. (1987) assessed the influence of winemaking techniques on As content in wines. The authors reported that a decrease in As content in wine occurs during the fermentation and maceration stages of wine production. They suggested that the decrease in As content results from the volatilization or the formation of colloidal substances that at first fluctuate, but later settle as a sediment that adheres to the yeast present in the grape skins. Moreover, the authors specifically noted that rose and blush wines contained more total As than red wine, “because they require a shorter period of contact with the skins” (Aguilar et al. 1987, p. 185). Other researchers have also reported that the winemaking process influences the elemental composition of wines (Almeida and Vasconcelos 2003). Variations in these practices may in part be responsible for the differences in total As content among blush, red, and whites found in our analysis. Results from the current analysis on California wines are consistent with these other findings, with blush wines having a higher total As concentration relative to red wines.
Total As concentration as a function of wine price
The potential relationship between total As concentration and price was also examined in this study. A recent consumer survey indicated that ~73% of wines are priced between $8 and $20/bottle ($10.67 to $26.67/L) (Thach et al. 2014). In this analysis, 44 of the 101 different bottles, boxes, or other containers purchased fell within this price range when price was normalized per liter (51 of the containers, regardless of volume, cost between $8 and $20).
A statistically significant inverse association between total As concentration and the price of a wine was found when all wines, regardless of type or group, were pooled, as well as within the randomly sampled wine group. No association was identified between total As concentration and wine price in the publicized wine group, however. The publicized wines represented less than 28% of the wines assessed in this analysis. Furthermore, the average cost of the publicized wines (mean = $9.22/L) was substantially lower than the wines in the random group (mean = $34.03/L). As a result, the subset of publicized wines occupies one end of the distribution presented in Figure 1, which may in part be because the publicized samples were not randomly selected and were known to contain elevated total As concentrations prior to this analysis. Hence, these wines were expected to occupy a narrower portion of the distribution of total As when included with a larger number of randomly selected California wines.
When price is controlled for, the publicized wines were found to contain significantly higher concentrations of As relative to each type of the randomly selected California wines (Figure 2). The As concentrations in the publicized wines (which were all low cost) are therefore not likely to be representative of all lower cost California wines. As discussed earlier, multiple factors, including soil conditions, growing practices, and enological methodologies, all contribute to total As content in wine, and may in part be responsible for the differences between the low-cost random and publicized wines in the current analysis.
Arsenic in wine exposure estimate
The maximum total As daily intake from wine consumption estimated in this analysis was 0.012 μg/kg-day for an adult. The largest source of As exposure in the general population is dietary intake, which was estimated to be 52.6 μg/day (~0.658 μg/kg-day) among US adults (Yost et al. 1998, USEPA 2011). Results from this analysis suggest that no more than 6.4 to 8.3% of the dietary intake of total As is attributable to wine (Jara and Winter 2014, Xue et al. 2010). These results confirm an estimate presented by Xue et al. (2010), who found that ~12% of inorganic As intake is attributable to beer and wine, suggesting that As intake from wine represents a fraction of the average dietary intake.
The estimated As exposure from wine reported in the current analysis is thus less than health-based values for inorganic As, such as the USEPA reference dose (RfD) and MCL. The estimated CDI for publicized wines was 25- to 50-fold less than the RfD for inorganic As, and the calculated CDI for random wines was 40- to 500-fold lower than the RfD. Dietary As intake from ingesting the USEPA mean per capita estimate of water intake (2.4 L/day), containing 10 μg/L of As, results in a daily dose of 0.30 μg/kg of As for adults (USEPA 2011, USEPA 2015). This dose (intake) is 25 to 490 times larger than the daily dose of As from wines sampled in the current study for the same age group, all wine groups (publicized and random), and wine types (red, white, and blush). Taken together, then, results from this study suggest that As exposure from wine consumption would not pose a public health hazard.
Comparison of results to Health Canada and OIV criteria for total As in wine
Similar to the current analysis, recent studies have consistently reported concentrations of As in wines that exceed the current USEPA MCL for drinking water (Campillo et al. 2008, Escudero et al. 2013, Herce-Pagliai et al. 2002, Huang et al. 2012, Moreira et al. 2011, Moreno et al. 2000, Rodrigues et al. 2011, Tašev et al. 2005). Comparing As concentrations in wine to As limits in water, however, does not appropriately characterize the potential health risk associated with wine consumption. Drinking-water standards are set as a fraction of the tolerable or acceptable daily intake value for a given contaminant. The international limits for total As in wine are 10- to 20-fold greater than the USEPA drinking water MCL for As, in part because the USEPA’s drinking-water consumption rate is considerably greater than the typical daily intake rate of wine. The recommended maximum intake for wine is one to two 5-ounce glasses per day, whereas the recommended daily intake for water is 2.7 to 3.7 L (91 to 125 oz) per day, and the USEPA estimated consumption rate for adults is 2.4 L (81 oz) per day (ACE 2008, USDHHS 1995, USEPA 2011, USEPA 2015). As discussed above, the total As dose that an adult would receive from wine consumption is a fraction of the dietary dose of the US population. These results offer further support that the safety of beverages should be assessed not only by the presence and amounts of various chemicals, but also in conjunction with an assessment of estimated background exposures and comparison to health-based standards.
The total As concentration for all wines sampled in this study were well below the limits for total As in wine set by Health Canada (100 μg/L) and OIV (200 μg/L). To date, these are the only standards for As in wine. The OIV recommended guideline is based on “long studies and inquiries” performed in various countries, and accounts for acceptable daily doses of As in food, as “fixed by the toxicologists of the large international authorities” (OIV 1994, p. 188). The Health Canada limit (100 μg/L) is based on the “as low as reasonably achievable” principle because of concerns about the possible oral carcinogenicity of inorganic As. This limit is therefore intended to consider both the toxicity of As to humans, as well as the best available agricultural and manufacturing practices (Health Canada 2014, p. 4). As previously mentioned, the distribution of total As concentrations in the randomly sampled wine group indicated that 0.3% (or one out of every 1000) wines produced in California would be expected to exceed this Health Canada limit. However, in an analysis of 1300 wines reported by BeverageGrades, “50 parts per billion of arsenic” was the “highest level found in one of the bottles,” indicating that no wines contained over 100 μg/L in the BeverageGrades study (CBS News 2015). Taken together, these results suggest that the probability of a California wine having total As concentrations exceeding the Health Canada limit of 100 μg/L may actually be lower than the estimate provided in the current analysis (CBS News 2015).
Conclusions
This study presents an assessment of the total As content in California red, white, and blush wines. California blush wines were found to contain significantly higher total As concentrations than red and white wines among both the publicized and random wine groups. Further, the wines in the publicized group contained higher total As concentrations relative to wines in the randomly selected group, even when controlling for price. Therefore, it appears that the total As content in wines publicized recently by the media are not representative of all low-cost California wines. The CDI for total As resulting from wine consumption indicates that the intake of As from wine represents less than 8.3% of a person’s total dietary consumption of As from food and beverages, regardless of wine group. This finding, then, indicates that wine is not a significant source of As exposure for adults.
Future research would be useful to further understand: 1) the relationship between total As concentration and the cost of California wines; 2) the relationship between the distributions of As in the environment and in wine, and the extent to which wine processing influences As concentration, particularly as it differs between white and red wines; and 3) the concentrations of individual species of As in wine to conduct a formal risk assessment that appropriately considers the lower toxicity of organic (compared with inorganic) As species.
Acknowledgments
All time and funding invested in this work was provided by Cardno ChemRisk and RJ Lee Group. Neither firm received any outside funding from any organization to perform any of the work associated with this research. Some Cardno ChemRisk authors have consulted with the Wine Institute on various matters, including the issue of arsenic in wine. RJ Lee Group is an AIHA-LAP– and TNI-NELAC–certified laboratory in metals analysis. RJ Lee Group has conducted analytical chemistry work for hundreds of firms over a 30-year period, including some in the food, beverage, and wine industry. Some RJ Lee Group senior chemists have provided expert opinions in court regarding analytical chemistry, and have provided expert services in food/beverage cases. RJ Lee Group also contributed support in the form of labor to this project. No funding was received from any wine-related industry. The authors thank Brooke Simmons for project leadership and management, Carley McCormick and Rachel Reid for assistance in purchasing and organizing samples, and Carrie Kahn and Maya McKeown for editorial review of the manuscript.
- Received May 2015.
- Revision received October 2015.
- Revision received November 2015.
- Accepted November 2015.
- ©2016 by the American Society for Enology and Viticulture
Literature Cited
Vol 67 Issue 2
