Concentration of Bisphenol A in Highly Consumed Canned Foods on the U.S. Market

Abstract

Metal food and drink cans are commonly coated with epoxy films made from phenolic polymers produced from bisphenol A (BPA). It is well established that residual BPA monomer migrates into can contents during processing and storage. While a number of studies have reported BPA concentrations in foods from foreign markets and specialty foods on the U.S. market, very few peer-reviewed data for the BPA concentrations in canned food from the U.S. market were available. This study quantified BPA concentrations in 78 canned and two frozen food products from the U.S. market using an adaptation of a previously reported liquid chromatography-tandem mass spectrometry method. The tested products represented 16 different food types that are from the can food classifications that constitute approximately 65% of U.S. canned food sales and canned food consumption. BPA was detected in 71 of the 78 canned food samples but was not detected in either of the two frozen food samples. Detectable BPA concentrations across all foods ranged from 2.6 to 730 ng/g. Large variations in BPA concentrations were found between different products of the same food type and between different lots of the same product. Given the large concentration ranges, the only distinguishable trend was that fruits and tuna showed the lowest BPA concentrations. Experiments with fortified frozen vegetables and brine solutions, as well as higher BPA concentrations in canned food solids over liquid portions, clearly indicated that BPA partitions into the solid portion of foods.

Full text

Published: May 20, 2011
This article not subject to U.S. Copyright.
Published 2011 by the American Chemical Society 7178 dx.doi.org/10.1021/jf201076f |J. Agric. Food Chem. 2011, 59, 7178–7185
ARTICLE
pubs.acs.org/JAFC
Concentration of Bisphenol A in Highly Consumed Canned Foods
on the U.S. Market
Gregory O. Noonan,* Luke K. Ackerman, and Timothy H. Begley
Center for Food Safety and Applied Nutrition, U.S. Food and Drug Administration, 5100 Paint Branch Parkway, College Park,
Maryland 20740, United States
ABSTRACT: Metal food and drink cans are commonly coated with epoxy films made from phenolic polymers produced from
bisphenol A (BPA). It is well established that residual BPA monomer migrates into can contents during processing and storage.
While a number of studies have reported BPA concentrations in foods from foreign markets and specialty foods on the U.S. market,
very few peer-reviewed data for the BPA concentrations in canned food from the U.S. market were available. This study quantified
BPA concentrations in 78 canned and two frozen food products from the U.S. market using an adaptation of a previously reported
liquid chromatographytandem mass spectrometry method. The tested products represented 16 different food types that are from
the can food classifications that constitute approximately 65% of U.S. canned food sales and canned food consumption. BPA was
detected in 71 of the 78 canned food samples but was not detected in either of the two frozen food samples. Detectable BPA
concentrations across all foods ranged from 2.6 to 730 ng/g. Large variations in BPA concentrations were found between different
products of the same food type and between different lots of the same product. Given the large concentration ranges, the only
distinguishable trend was that fruits and tuna showed the lowest BPA concentrations. Experiments with fortified frozen vegetables
and brine solutions, as well as higher BPA concentrations in canned food solids over liquid portions, clearly indicated that BPA
partitions into the solid portion of foods.
KEYWORDS: bisphenol A, canned foods, LCMS/MS
’INTRODUCTION
Metal, as drink or food cans, caps on glass bottles and jars, and
heating trays and lids, is a commonly used material for food
transport, storage, and protection. Most, if not all, of these metal
surfaces are coated prior to making contact with food.
1
The
internal coating plays a dual role, providing protection of the
contents from metal contamination and protection of the metal
from oxidation or corrosion by the contents.
1
Perhaps the most
common examples of internal coatings are epoxy films, often
composed of phenolic polymers produced from bisphenol A
(BPA) or 2,2-(4,40-dihydroxydiphenyl)propane. These films are
widely used as food can coatings because of their flexibility and
corrosion resistance.
It is well established
2,3
that because a residual amount of
monomer remains after the polymerization process, BPA
46
and
other components
7,8
migrate, at very low concentrations (parts
per billion), from the epoxy coatings into the contents during
processing and storage. Indeed, food has been identified as the
primary source of BPA exposure for humans.
9,10
To characterize
this exposure, the exposure estimates from all dietary sources
were re-evaluated in 2009.
11
During this exposure evaluation, a
number of studies were identified that reported BPA concentra-
tions in canned and other packaged foods.
24,1223
However,
many of these studies sampled only a limited number of
foods
3,18,24
or focused on a specific food category, such as infant
formulas,
2,12,16
baby foods,
15
or soft drinks.
14
The remaining
publications and larger survey efforts focused on foods from
the European Union,
13,17
Asia,
2123
or other markets
6,20,25
and
tested only a small number of U.S. products. It was clear that
there were no large scale studies of the U.S. market and that there
were significant gaps in the data for highly consumed canned
foods, such as chili, pastas, and pork and beans.
More recently, two publications
26,27
have reported BPA
concentrations in foods available on the U.S. market. Both of
these studies were published in mid to late 2010 and were not
available when this study was initiated. Given the large number of
canned foods, different food and can manufacturers, and the
proprietary nature of can coatings,
1
the two previous surveys,
when combined with this survey, only begin to assemble a
representative sample of BPA concentrations in foods on the
U.S. market. Because of the size and diversity of the potential
canned food sample pool, the objective of this study was to
measure BPA concentrations in canned food types representing
the most consumed canned foods. To this end, this study
utilized sales data to determine select food types that, on a
mass basis, were the most highly purchased canned foods in
the United States. Finally, this study investigated the partition-
ing of BPA between solid and liquid food portions
23,28,29
as
a possible explanation for differences between reported BPA
concentrations.
’EXPERIMENTAL SECTION
Reagents. Standards of BPA (>99%, Sigma-Aldrich, St. Louis, MO)
and [
13
C
12
]BPA (99%, Cambridge Isotope Laboratories, Andover, MA)
Received: March 16, 2011
Accepted: May 20, 2011
Revised: May 18, 2011

7179 dx.doi.org/10.1021/jf201076f |J. Agric. Food Chem. 2011, 59, 7178–7185
Journal of Agricultural and Food Chemistry ARTICLE
were tested for purity and suitability by a liquid chromatography
tandem mass spectrometry (LCMS/MS) method upon receipt. The
water, acetonitrile, and methanol used were 99.9% Fischer (Fair Lawn, NJ)
Optima LCMS grade.
Equipment. Highly crystalline 50 and 15 mL polypropylene (PP)
centrifuge tubes (Corning, Corning, NY) were used for sample storage
and processing. Solid foods were homogenized using a stainless steel
homogenizer (PRO250, Pro Scientific, Oxford, CT). Aliquots were
centrifuged on a Fischer Marathon 2100R centrifuge (Thermo Fisher,
Pittsburgh, PA). HPLCMS/MS analyses were performed using an
Agilent 1100 Series HPLC system (Agilent Technologies, Wilmington,
DE) with a micro volume mixing and degassing chamber, a Pursuit XRs
C18 column (Agilent Technologies; 2 mm 150 mm, 3 μm) or a Luna
C8(2) column (Phenomenx, Torrance, CA; 2 mm 150 mm, 3 μm),
and an API-4000 Q-trap MS/MS instrument (Applied Biosystems,
Foster City, CA). Data analysis was performed using Analyst and Excel
(Microsoft, Redmond, WA). A Branson 3510 ultrasonic bath purchased
from VWR (Radnor, PA) was used for all sonication experiments.
Sample Collection. Eighty different products, 78 canned and two
frozen foods, were purchased from retail stores in Washington, DC, and
surrounding Maryland counties. The majority of samples were pur-
chased from large chain grocery stores, but a number of cans were
purchased from local cooperatives and ethnic food markets. Brands were
chosen to represent a variety of available manufacturers, including
national brands, store brands, value brands, organic brands, and specialty
brands. None of the foods collected were labeled as “BPA free”. Food
types and brands chosen for analysis were many of the most commonly
consumed canned food products on the basis of available sales survey
data from 2007 to 2009 (The Nielsen Co., New York, NY). For example,
using the Nielsen data from 2007 to 2009, it was determined that chili is
the most highly consumed canned food, while ravioli and pork and beans
are the second and third most consumed canned foods, respectively
(Table 3). The 16 types of foods analyzed for this study constitute more
than 65% of the canned food sales and canned food diet. All canned
samples were stored at room temperature prior to being opened and
were sampled prior to the “use by”dates stamped on the cans. Any
frozen foods were stored frozen (below 10 °C) and thawed at 4 °C
prior to use. The manufacturer or distributor, food type, can size, can
type, expiration date, and manufacturing code were recorded for all
products prior to opening.
Sample Preparation. The analytical method utilized for this work
is a simplification of the LCMS/MS method previously reported for
the determination of BPA in liquid and powder infant formula.
12
Because of the higher BPA concentrations generally found in canned
foods, the solid phase extraction cleanup and extract concentration steps
were not performed. After the cans had been opened, any brine or liquid
was decanted, processed, and analyzed as an independent sample. The
remaining solid food was placed in a glass beaker and homogenized into
a smooth paste using a stainless steel homogenizer. An aliquot (1 g) of
the liquid or solid food was placed in a 15 mL polypropylene centrifuge
tube, spiked with [
13
C]BPA (50 μL of a 1 ng/μL solution), and
vortexed. Acetonitrile (2 mL) was added, and the samples were
thoroughly mixed by being vortexed for 12 min. After being mixed,
the samples were centrifuged (4000 rcf) for 10 min at room temperature.
An aliquot (0.75 mL) of the supernatant was transferred to a glass
autosampler vial and diluted 1:2 with water. The vials were sealed with
PTFE-lined caps, briefly vortexed to ensure mixing, and then immedi-
ately analyzed or held at 4 °C until analysis.
Instrumental Analysis. Separations were performed at a flow rate
of 0.4 mL/min and a column temperature of 60 °C. A watermethanol
gradient from 40 and 100% methanol during the first 6 min led to elution
of BPA around 6.1 min. Following BPA elution, the analytical column
was washed with methanol for 1 min and returned to 40% methanol for a
9 min equilibration time. Negative ion electrospray at 5 kV, curtain gas
at 20, ion source gas 1 at 40, ion source gas 2 at 70 (arbitrary units), and a
temperature of 600 °C were used. A declustering potential of 70 V, an
entrance potential of 10 V, and a collision cell exit potential of 10 V
were used. Two MS/MS transitions were monitored for each of the
analytes (BPA, m/z227.1 to 211.9 and 133.1; [
13
C
12
]BPA, m/z239.1 to
223.9 and 139.1). Collision energies of 28 and 35 eV were used for the
first and second transitions, respectively. Analyte confirmation required
both mass transitions to peak at a >3:1 signal-to-noise ratio (S/N) and
within (0.02 min of the expected retention time and the relative
intensity of the two transitions to match their ratio in the standard
within (20%. Calibration solutions ranged from 1.0 to 1000 ng/mL
BPA, with 50 ng/mL [
13
C
12
]BPA in each calibration solution. Calibra-
tion solutions were analyzed prior to and throughout daily sets of sample
extracts, and BPA was quantified as the area ratio of one BPA transition
(m/z227.1 to 211.9) to the internal standard [
13
C
12
]BPA transition
(m/z239.1 to 223.9). All resulting calibration curves were linear
between 1.0 and 1000 ng/mL (r
2
> 0.99).
A second quantification method for the confirmation of BPA con-
centrations, which used different chromatographic conditions, was
performed on a subset of sample extracts. The confirmatory analysis
was performed on a Luna C8(2) column (150 mm 2.0 mm, 3 μm)
using a 0.4 mL/min flow rate and 60 °C column temperature. A water
(5% acetonitrile)acetonitrile (5% water) gradient between 40 and 100%
acetonitrile during the first 6 min led to elution of BPA around 3.4 min.
The analytical column was washed by being held at 100% acetonitrile for
2 min and then returned to 40% acetonitrile for an 8 min equilibration
time. The mass spectrometer source and MRM conditions were identical
to those described previously.
Partitioning Experiments. Frozen peas and green beans were
thawed and dried of excess water. Approximately 25 g of the vegetable
was placed in 50 mL glass bottles. An aqueous BPA solution (25 mL of a
100 ng/mL solution) was added to each bottle, and the samples were
held at 100 °C for 4 h or 40 °C for 24 h. Additionally, experimental
controls, including a BPA solution without vegetables and vegetables
covered in LCMS grade water, were also held at 100 and 40 °C. After
the allotted time, the bottles were removed from the heat and allowed to
cool for 1015 min at room temperature. The liquid was decanted from
the vegetables, and the entire solid portion was homogenized to a
smooth paste. Subsamples (1 g) of the solid homogenate and liquid were
prepared and analyzed separately to determine potential partitioning.
Additionally, to determine background BPA concentrations, samples of
the thawed green beans and peas were homogenized, extracted, and
analyzed without incubation.
Fourier Transform Infrared Spectroscopy. Fourier transform
infrared (FTIR) analyses of select can coatings were conducted with a
Nicolet 6700 (Nicolet Analytical Instruments, Madison, WI) spectro-
meter equipped with a Smart Performer single-bounce ATR (attenuated
total reflectance) attachment (Thermo Scientific, Madison, WI) using a
germanium crystal. Spectra were processed with Omnic (Nicolet
Analytical Instruments) with comparisons to Hummel (Thermo Inc.)
and in-house spectral libraries for polymers and additives. After being
emptied, cans were washed with soap and water and air-dried. After the
cans had dried, small pieces (1 cm
2
) were cut from the side and ends of
each can. The exterior of the can was marked to ensure that spectra were
collected (4 cm
1
resolution) using the inside can surface. Using the
empty reflectance cell, background spectra were collected prior to the
analysis of each can piece. To ensure there was no carryover, the ATR
crystal was cleaned with ethanol and allowed to air-dry between each can
sample.
’RESULTS AND DISCUSSION
Method Validation. The method used in this study had
previously undergone extensive single-laboratory validation;
12

7180 dx.doi.org/10.1021/jf201076f |J. Agric. Food Chem. 2011, 59, 7178–7185
Journal of Agricultural and Food Chemistry ARTICLE
however, that work addressed only a single matrix (infant
formula). Therefore, to evaluate the performance of the method
in the wide variety of food matrices tested in this study, we
analyzed a series of reagent blanks, replicates, and spike samples.
BPA was spiked into every food type sampled and also spiked at
various levels (10400 ng/g) in a number of foods, to ensure
that the BPA concentration did not affect the accuracy of the
method. Additionally, extraction using sonication (30 min) in
addition to vortex mixing was evaluated for a number of the solid
food matrices.
BPA was not detected (<2 ng/g) in any of the reagent blanks
or frozen vegetables. These reagent blank data are consistent with the
background concentrations previously reported for the infant formula
method (0.010.14 ng/g) falling well below the limit of detection
(LOD) for the current method (2 ng/g). The nondetectable results
for frozen vegetables are consistent with the theory that epoxy can
coatings and not environmental or other processing conditions are
the primary source of BPA in canned foods.
Spike recovery data across a large concentration range
(10400 ng/g) and a variety of food products are presented in
Table 1. Eighteen of the 19 spiked samples showed BPA
recoveries between 94 and 104%. The final sample had a recovery
of 110% with a spike concentration of 220 ng/g. There was no
correlation between the spike concentration and the recovery,
and none of the food types showed consistently lower or higher
recoveries.
Eleven food samples were analyzed in at least triplicate to
determine the precision of the method. All aliquots were taken
from the same can, but samples were processed and analyzed on
different days. All but two of the samples had a relative standard
deviation (%RSD) of e10, with the remaining two samples, ravioli
and pork and beans, having %RSD values of 12 and 16, respectively.
The majority of samples show good reproducibility, and there
was no correlation between the precision and BPA concentra-
tion. Indeed, samples with BPA concentrations ranging from 13
to 700 ng/g varied between %RSD values of 6 and 8. Addition-
ally, no single matrix consistently yielded poor precision.
Although the largest %RSD for replicate analysis was a pork
and beans sample, another product of this same food type yielded
a %RSD of 7.7. On the basis of the good precision of the method
across the variety of food types, the remaining survey samples
were analyzed in duplicate. All but two samples had a <10%
difference between the duplicate values. One of the two samples
was pineapple (national D), where both values (2.6 and
4 ng/g) were below the limit of quantitation (LOQ). The other
sample was chili (store A), which had BPA concentrations of 26
and 33 ng/g, a difference of 25%. Because both values were well
within the range reported for this food type and other chili
samples showed good precision, no further replicate testing of
this food type was performed.
To ensure that vortex mixing was quantitatively extracting
BPA from different homogenized food types, we investigated a
more rigorous extraction method for a number of solid food
matrices. Sonication (30 min) at room temperature was used in
place of vortex mixing, after the addition of acetonitrile. The use
of sonication did not change, within experimental error, the BPA
concentrations determined for those samples (Table 2). All six
sonicated samples showed BPA concentrations within the 95%
confidence limit of the mean BPA concentration determined
using vortex mixing. On the basis of these data, vortex mixing was
used during the remainder of the sample processing.
Survey Results. It should be noted that the BPA concentra-
tions reported represent the BPA concentration in the canned
food and are not corrected for food preparation steps, which
could change the “as consumed”BPA concentration. Also, no
adjustments were made for can size (i.e., surface area-to-mass
ratio). In general, the can sizes and food amounts throughout the
survey were similar, ranging from 383 to 454 mL, even between
food types. However, there are a few exceptions, including tuna
(142 mL), crushed pineapple (567 mL), a number of condensed
soups (298 mL), one ready-to-eat soup (538 mL), value A products
(794 and 1130 mL), and plum tomatoes (794 mL). Additionally, the
frozen peas and green beans were packaged in polyethylene bags.
The majority (71 of 78) of the canned food samples tested in
this survey contained detectable concentrations of BPA, while
BPA was not detected in either of the two frozen foods (Table 3).
Table 1. BPA Spike Recovery
[BPA] (ng/g)
food description initial
initial and
spike
a
measured
recovery
(%)
green beans frozen <2 10. 11 105
49 46 94
220 240 110
national A 43 91 94 104
pork and
beans
national E 13 64 66 104
420 400 94
chili store E, chicken 98 150 140 94
190 190 102
490 490 99
organic E 60 110 110 102
organic D 78 130 120 94
pasta organic C 40 90. 88 98
store E 40 91 91 100
fruit cocktail store A <2 48 48 99
pineapple national K <2 45 43 97
vegetables import B, liquid <2 50. 50. 101
import B, solid 10 380 390 104
tomatoes national B 24 72 72 100
fish import E 22 390 380 97
a
Nominal 1 g sample aliquots were used for all spike samples.
Table 2. Repeatability and Sonication
food description
average
[BPA] (ng/g)
RSD
(%)
BPA concentration
for sonication (ng/g)
soup 55 1.5
pork and beans national E 13 7.7
national F 46 16
ravioli national F 50 8.4 52
store E 54 12 55
chili store E 98 7.7 110
red beans national J 200 6.4 200
black beans national J 720 5.9 710
green beans national B 24 8.2
national B 310 2.7
spinach national B 22 10

7181 dx.doi.org/10.1021/jf201076f |J. Agric. Food Chem. 2011, 59, 7178–7185
Journal of Agricultural and Food Chemistry ARTICLE
Table 3. Descriptions of Samples and BPA Concentrations in Food Samples
average [BPA] (range)
b
(ng/g)
food (canned diet %)
a
brand (description) volume (mL) can pieces solid liquid
green beans (6.9) store A 411 3 22 8.3
store A (frozen) not available not available <2
organic A 411 2 230 (32440) 21 (4.238)
national A 411 2 38 (3343) 3.6 (2.74.7)
national B 411 2 490 (280600) 30 (20.42)
national B (no salt) 411 2 500 (300730) 28 (1547)
national B (organic) 411 3 33 4.2
value A 794 3 16 <2
corn (5.5) import A 425 3 76 7.6
store B 432 3 4.2 2.6
national B 432 3 25 3.1
tomatoes (0.6) national C (sauce) 425 2 2.6
store B (plum) 794 3 5.8
store A (diced) 411 3 43
organic A (diced) 411 3 <2
national B (diced) 411 3 24
peas (1.2) national B (no salt) 425 2 310 11
national A 425 2 170 12
store A (no salt) 425 2 12 <2
organic A 425 3 2.6 <2
store D 425 3 3 <2
fruit cocktail (4.1) store A 425 3 <2
national B 432 3 19
store D 425 3 2.7
store C 425 3 4.6
pineapple (3.6) national D 567 3 3.1
store C 567 3 <2
national K 567 3 <2
organic B 396 3 13
store A 567 3 <2
sliced peaches (5.8) store A 432 3 9.3
store C 432 3 <2
national B 425 3 6.3
national B 425 3 7.0
store E 425 3 <2
ravioli (7.4) national F 425 2 43 (2254)
value A 1130 3 7.5
organic C 425 2 39
national E 425 3 10.
store E 418 2 50. (4462)
pork and beans (7.1) national F 425 3 46
national E 446 3 13 (1214)
chili (8.3) store E (chicken) 425 2 98 (95110)
store E (turkey) 425 2 76
national G 425 2 61 (3491)
national G (hot) 425 2 30.
national H 425 2 130
national H (country) 425 2 150
organic E 416 2 75
store A 425 2 30.
store A (no beans) 425 2 65 (4581)
organic E 425 2 59

7182 dx.doi.org/10.1021/jf201076f |J. Agric. Food Chem. 2011, 59, 7178–7185
Journal of Agricultural and Food Chemistry ARTICLE
These findings are consistent with previously reported data and
the fact that the epoxy resins used as liners for metal cans result in
migration of BPA and are the main source of BPA in canned
foods.
There was a large range between the lowest detectable (2.6
ng/g) and highest (790 ng/g) BPA sample concentrations. This
high variability was noted between food types, and also within
many of the food types. For example, BPA concentrations
determined for peas showed a 10-fold difference (2.6 to 310
ng/g) between the minimal and maximal values, while green
beans had a 30-fold difference (22 to 730 ng/g) between brands.
Refried beans showed the largest range with more than 2 orders
of magnitude (6.3 to 790 ng/g) between brands. A number of
foods, such as tomatoes and corn, exhibited large relative (20-fold)
but smaller magnitude (40 and 70 ng/g, respectively) differences
between manufacturers. These variations within food types are
consistent with previously reported data. Cao et al.
26,30
reported
60- and 40-fold differences for tuna and condensed soups,
respectively. The large variation in BPA concentration is also
consistent with multiple can manufacturers using different
proprietary coatings and with food producers using different
can styles or coating choices for different products.
1
Although a
number of foods did show large variations in BPA concentra-
tions, approximately half of the food types, including pasta, pork
and beans, chili, soups, ravioli, and fruits, showed smaller relative
and absolute variation, with differences in BPA concentrations
from 3- to 5-fold or from 10 to 80 ng/g.
Multiple lots of the same food type and brand were collected
for 16 of the 78 canned products. Additionally, for 10 of the
products, similar products (e.g., chili with and without beans) in
similar can types from the same manufacturer were purchased. In
all cases, multiple lots were tested only in products that exhibited
detectable levels of BPA. Overall, the lot-to-lot variability was
smaller than the variability between and within foods, but it varied
greatly between food types and manufacturers. For example, tuna
and pork and beans samples showed very little difference between
lots, while refried bean and green bean lots differed from 3- to >10-
fold. Lot-to-lot variability was only evaluated on similar product
information; therefore, it is possible that for certain foods, different
food production lots used the same can type and lot, while for
other foods both the can and food production lot differed.
Additionally, because different lots were collected at different
times and locations, there were no controls or methods to evaluate
if the same can coating or can manufacturer was used for both lots
of food.
As described above, the presence of BPA was confirmed in all
of the samples by the intensity ratio of both mass transitions
compared to that of the standard. However, it has been reported
that comigrants (hydrolyzed forms of the BADGE monomer)
from epoxy can coatings can dissociate during electrospray
Table 3. Continued
average [BPA] (range)
b
(ng/g)
food (canned diet %)
a
brand (description) volume (mL) can pieces solid liquid
refried beans (4.5) import D (black) 454 3 10 (6.321)
import D (red) 454 3 23
national J (black) 454 3 680 (280790)
national J (red) 454 3 210 (160240)
pasta (4.2) organic C (tomato sauce) 425 2 39 (3443)
national E (tomato sauce) 425 3 12
store E (tomato sauce) 418 3 40.
national F (cheese) 425 2 15
national F (tomato sauce) 425 2 21
store D (tomato sauce) 418 2 35
fish (0.4) national I (tuna) 142 2 17 <2
national I (albacore) 142 2 12 (1113) <2
national M (tuna) 142 2 5.8 <2
national M (tuna, oil) 142 2 4.5 <2
national L (albacore) 142 2 11 <2
import E (mackerel) 400 3 22 5.4
soups (0.9) national E (condensed chicken) 298 2 54
national E (condensed chicken) 298 2 74 (55110)
national A (ready-to-eat chicken) 538 2 56
store B (clam chowder) 425 2 32
store B (organic tomato) 411 2 63
national E (chicken broth) 396 3 13
miscellaneous vegetables (0.3) national B (wax beans) 411 2 140 14
national B (spinach) 396 3 23 2.9
import B (stir fry vegetables) 425 3 11 <2
import C (oyster mushrooms) 425 3 13 4.0
miscellaneous import F (almond jelly) 538 3 4.6
a
The percentage of canned food diet based on register receipt data from 2007 to 2009.
b
Range used to show BPA concentrations determined when
multiple lots of a product were analyzed.

7183 dx.doi.org/10.1021/jf201076f |J. Agric. Food Chem. 2011, 59, 7178–7185
Journal of Agricultural and Food Chemistry ARTICLE
ionization and produce the (M H)

ion of BPA (m/z227.1) in
the source and thus the same MRM transitions.
12,31
These
previous reports established that the chromatographic method
used in this survey resolved these comigrants from BPA. However,
because of the large variation in BPA concentrations within food
types and the relatively high BPA concentrations reported for a
number of foods, a secondary chromatographic method was used
to confirm the initial results. Refried beans, green beans, ravioli,
peas, chili, and soups were reanalyzed to evaluate if comigrants or
other compounds were interfering with the quantification of BPA.
The alternate chromatographic conditions, described above, shor-
tened the BPA retention time and changed the elution order of
BPA and known comigrants.
31
Such changes were consistent with
the retention and elution profiles previously reported by Petersen
et al.
32
for bisphenol A diglycidyl ether derivatives. All of the BPA
concentrations determined with the Luna C8(2) chromatographic
conditions were comparable, within the 95% confidence limit, to
those reported using the standard method. On the basis of the
results from the confirmatory quantification method, it is clear that
the BPA concentrations reported (Table 3) are not affected by the
previously identified comigrating epoxy derivatives
31
and unlikely
that they are affected by other coeluting interference.
Given the large variability in BPA concentration combined
with the variety of foods and manufacturers, there are few clear
trends in the data. For example, the survey data do not show a
correlation between the type of manufacturer (national, store,
organic, or import) and the concentration of BPA in the foods.
This lack of correlation is consistent with epoxy linings being the
primary source of BPA and multiple can manufacturers supplying
cans to numerous food producers. One clear trend is that fruits,
excluding tomatoes, and tuna samples had the lowest BPA
concentrations of the samples analyzed. Six of the 14 fruit
samples showed no detectable BPA, and six of the eight remain-
ing samples had BPA concentrations of <10 ng/g. These findings
are in agreement with those of Thomson and Grounds,
25
who
reported canned fruits from Australia, New Zealand, and Asia to
contain <10 ng of BPA/g. Braunrath et al.
13
and Yoshida et al.
23
also report similar results in smaller surveys of canned fruits from
Asia and Australia. Finally, Schecter et al.
27
also reported on a
single pineapple sample from the U.S. market, which had no
detectable BPA concentration (<0.2 ng/g).
The low BPA concentrations found in canned fruits
(pineapple, peaches, and fruit cocktail) are consistent with the
general industry practice of using tin and not epoxy phenolic
films in canned fruit containers.
1
Using FTIR spectroscopy,
organic films were not detected on the side walls of any of the
fruit cans sampled for this survey. Some, but not all, of the cans
did have epoxy phenolic films on the can ends. However, the
presence of an epoxy phenolic coating did not always correlate
with the detection of BPA in the fruit sample. For example, the
spectra of the can ends for store A pineapple were consistent with
the use of an epoxy phenolic coating; however, BPA was not
detected in the contents. Alternatively, organic films were not
detected on the ends or side walls of the national D pineapple
can; however, an average BPA concentration of 3.1 ng/g (below
the LOQ) is reported for this sample (Table 3).
Given the large variation of BPA concentrations within food
types and between manufacturers, the comparison of BPA data
from overseas markets and products to U.S. products is not
overly informative. However, the results presented here are in
good agreement with recent reports on BPA concentrations in
canned foods from the Canadian market.
26,30
Many of the foods
and food manufacturers tested in that survey are products
available on the U.S. market. Specifically, the range and mean
of the BPA concentrations reported for condensed and ready-to-
serve soups, tomatoes, miscellaneous vegetables, and single-
serving canned pasta are consistent with the values reported in
Table 3. The green bean values also are within the range reported
here; however, Cao et al.
26
do not report any of the higher
concentrations found in this survey. Tuna is the only product for
which there is limited agreement between the two data sets.
While there is an overlap in the concentration ranges, the data
from Cao et al. show a wider range (9.0534 ng/g) and a higher
mean (137 ng/g) than the results found in this study (4.517
and 10 ng/g, respectively). The tuna products sampled in the
Canadian study were not available for purchase for this survey.
It is likely the differences in BPA concentration are due to
differences in can coatings used by the manufacturers.
While the values of this study are in good agreement with
those of the Canadian survey, there is less agreement with BPA
concentrations in canned foods recently reported.
27
In general,
the values reported by Schecter et al.
27
are biased low compared
to the ranges reported in Table 3. Pineapple is the only commodity
for which the two surveys report comparable BPA concentrations
(below the LOD). Green beans (50.5 ng/g) and peas (3.12 ng/g)
are the only foods for which the BPA concentrations reported by
Schecter et al. fall within, but at the low end, of the ranges reported
here, 22730 and 3310 ng/g, respectively.
27
BPA concentra-
tions reported here are higher for a number of commodities,
including chili, pasta, soups, and tuna fish.
It is possible that the differences in BPA concentrations
simply represent differences in can size, can lot, or manufacturer.
However, they may also represent differences in sample proces-
sing or analytical methodology. Schecter et al. do not describe the
procedure used for sample homogenization or if solid and liquid
portions were combined or separated.
27
While the entire can
contents were used in foods such as soups, chili, and pasta, the
inclusion of the supernatant in tuna and green bean samples
would lower the BPA concentrations compared to those in
Table 3. Additionally, Schecter et al.
27
used a freeze-drying step,
prior to the addition of internal standard, which was not used by
Cao et al.
26
or in our analysis. It should be noted that while the
overall instrumental method used by Cao et al.
26
is different from
that reported here, the initial sample handling steps were compar-
able. Both methods separated the supernatant from solid food,
homogenized the entire canned solid sample, added internal
standard prior to extraction, and extracted the wet homogenized
aliquot with acetonitrile.
BPA Partitioning. In an effort to measure BPA in the common
edible portion, the sample preparation step of the analysis took
care to separate the solid food from the liquid supernatant. The
supernatant was then analyzed to provide complete data, in case
there is disagreement about the consumption of the liquid portion.
Because of this approach, the survey results clearly showed that in
foods with separate liquid and solid portions, the BPA concentra-
tions are higher, generally g10-fold, inthe solid food (Table 3). A
few samples did have smaller solid:liquid ratios, but these samples
had relatively low levels (<20 ng/g) of BPA in the solid portion
or BPA concentrations at or near the method LOQ. These results
agree with data reported for Japanese canned fruits and vegetables
23
and in a Belgian market survey.
29
However, other researchers
have reported BPA concentrations determined with solid phase
microextraction gas chromatography and mass spectrometry
to be higher in the liquid supernatant of a variety of canned

7184 dx.doi.org/10.1021/jf201076f |J. Agric. Food Chem. 2011, 59, 7178–7185
Journal of Agricultural and Food Chemistry ARTICLE
vegetables.
28
That study did not utilize an internal standard
during extraction and derivatization, which may impact the
accuracy of the recovery from solid foods and thus the quanti-
fication. Additionally, water was utilized as a solvent during the
direct injection solid phase microextraction (SPME) procedure
for the solid food portions. If BPA partitions into the food at a
concentration greater than that at which it partitions into water,
as our results indicate, then the choice of water as an extraction
solvent would impact the accuracies of the SPME quantification.
To further address the question of BPA partitioning with a
controlled sample set, a series of partitioning experiments were
performed using frozen green beans and peas. The green beans
and peas held at 40 °C for 24 h had the same texture and color as
the samples processed directly from the bag. Both were bright
green, with a slight hard, crisp texture that made homogenization
more difficult. However, the samples that had been held at
100 °C for 4 h had a texture and color more consistent with
canned vegetables. The vegetables had a faded green color and a
soft texture, allowing for easy homogenization.
All of the blanks and negative control samples, including
samples taken directly from the package and samples incubated
with water, showed no detectable levels of BPA (Table 4).
Additionally, all of the positive control samples, which consisted
of BPA solutions incubated under the different conditions with-
out the addition of vegetables, showed excellent recoveries. The
results for blanks and positive controls clearly established that
under the incubation conditions investigated, contamination
from storage containers or loss of BPA through degradation or
adsorption to the storage container did not affect BPA concen-
trations. For the incubated vegetable samples, all of the liquid
supernatants contained significantly lower concentrations of
BPA than the corresponding solid portion. The results were
generally similar between green beans and peas and between the
different incubation conditions. The loss of BPA from the liquid
water and the increase in the level of BPA in the vegetables are in
agreement with data reported by Yoshida et al.
23
with similar
experiments using corn. That study used lower temperatures and
longer storage times but reported comparable solid/liquid
partitioning values at the end point of the storage.
It is clear from our findings that the BPA concentrations in
canned foods vary greatly not only between food types but also
within food types and even between production lots from the
same manufacturer. Additionally, the results established that
because of partitioning of BPA into the solid portion of foods,
the inclusion or exclusion of a brine solution can significantly
affect the BPA concentrations reported. Both of these factors
may contribute, solely or in combination, to the discrepancies
between published BPA concentrations. On the basis of these
survey results, no specific type of manufacturer (national, store,
value, or organic) had significantly higher or lower BPA con-
centrations in the food. While it is tempting to correlate BPA
concentrations with food composition (fat content, salt content,
and pH), the lack of knowledge about and control of differences
in can coatings and can processing makes such conclusions
erroneous.
’ACKNOWLEDGMENT
We thank Allan Bailey and Karen Hatwell of the U.S. Food and
Drug Administration’sOffice of Food Additive Safety for provid-
ing and compiling the Nielsen data on purchased canned food.
’REFERENCES
(1) Oldring, P. K.; Nehring, U. Metal packaging materials 7. Metal packag-
ing for foodstuffs; Institute for Life Sciences International: Brussels, 2007.
(2) Biles, J. E.; McNeal, T. P.; Begley, T. H. Determination of
bisphenol A migrating from epoxy can coatings to infant formula liquid
concentrates. J. Agric. Food Chem. 1997,45, 4697–4700.
(3) Brotons, J. A.; Olea-Serrano, M. F.; Villalobos, M.; Pedraza, V.;
Olea, N. Xenoestrogens released from lacquer coatings in food cans.
Environ. Health Perspect. 1995,103, 608–612.
(4) Goodson, A.; Robin, H.; Summerfield, R. W.; Cooper, I. Migra-
tion of bisphenol A from can coatingseffects of damage, storage
conditions and heating. Food Addit. Contam. 2004,21, 1015–1026.
(5) Munguia-Lopez, E. M.; Peralta, E.; Gonzalez-Leon, A.; Vargas-
Requena, C.; Soto-Valdez, H. Migration of bisphenol A (BPA) from
epoxy can coatings to jalapeno peppers and an acid food simulant.
J. Agric. Food Chem. 2002,50, 7299–7302.
(6) Munguia-Lopez, E. M.; Soto-Valdez, H. Effect of heat processing
and storage time on migration of bisphenol A (BPA) and bisphenol
A-diglycidyl ether (BADGE) to aqueous food simulant from Mexican
can coatings. J. Agric. Food Chem. 2001,49, 3666–3671.
(7) Bradley, E. L.; Driffield, M.; Harmer, N.; Oldring, P. K. T.;
Castle, L. Identification of potential migrants in epoxy phenolic can
coatings. Int. J. Polym. Anal. Charact. 2008,13, 200–223.
(8) Schaefer, A.; Simat, T. J. Migration from can coatings: Part 3.
Synthesis, identification and quantification of migrating epoxy-based
substances below 1000 Da. Food Addit. Contam. 2004,21, 390–405.
(9) Kang, J. H.; Kondo, F.; Katamaya, Y. Human exposure to
bisphenol A. Toxicology 2006,226,79–89.
(10) Joint FAO/WHO Expert Meeting to Review Toxicilogcal and
Health Aspects of Bisphenol A Summary Report [Internet]; World
Health Organization: Geneva, 2010.
(11) Bailey, A.; Hatwell, K.; Mihalov, J. Memorandum of 10/22/
2009, Exposure to bisphenol A for infants, toddlers and adults from the
consumption of infant formula, toddler food and adult (canned) food. In
Federal Register;Office of the Federal Register: Washington, DC, 2009;
Vol. 75, pp 1714517147.
(12) Ackerman, L. K.; Noonan, G. O.; Heiserman, W. M.; Roach,
J. A.; Limm, W.; Begley, T. H. Determination of bisphenol A in U.S.
infant formulas: Updated methods and concentrations. J. Agric. Food
Chem. 2010,58, 2307–2313.
(13) Braunrath, R.; Podlipna, D.; Padlesak, S.; Cichna-Markl, M.
Determination of bisphenol A in canned foods by immunoaffinity
chromatography, HPLC, and fluorescence detection. J. Agric. Food
Chem. 2005,53, 8911–8917.
(14) Cao, X.-L.; Corriveau, J.; Popovic, S. Levels of bisphenol A in
canned soft drink products in Canadian markets. J. Agric. Food Chem.
2009,57, 1307–1311.
(15) Cao, X.-L.; Corriveau, J.; Popovic, S.; Clement, G.; Beraldin, F.;
Dufresne, G. Bisphenol A in baby food products in glass jars with metal
lids from Canadian markets. J. Agric. Food Chem. 2009,57, 5345–5351.
Table 4. Partitioning of BPA into Solid Food
[BPA] (ng/g)
40 °C 100 °C
sample liquid solid liquid solid
positive control 110 not available 110 not available
green beans
blank <2 <2 <2 <2
BPA-spiked 7.3 110 18 77
frozen peas
blank <2 <2 <2 <2
BPA-spiked 2.9 90 10 100

7185 dx.doi.org/10.1021/jf201076f |J. Agric. Food Chem. 2011, 59, 7178–7185
Journal of Agricultural and Food Chemistry ARTICLE
(16) Cao, X.-L.; Dufresne, G.; Belisle, S.; Clement, G.; Falicki, M.;
Beraldin, F.; Rulibikiye, A. Levels of bisphenol A in canned liquid infant
formula products in Canada and dietary intake estimates. J. Agric. Food
Chem. 2008,56, 7919–7924.
(17) Goodson, A.; Summerfield, W.; Cooper, I. Survey of bisphenol
A and bisphenol F in canned foods. Food Addit. Contam. 2002,
19, 796–802.
(18) McNeal, T. P.; Biles, J. E.; Begley, T. H.; Craun, J. C.; Hopper,
M. L.; Sack, C. A. Determination of suspected endocrine disruptors in
foods and food packaging. In Analysis of Environmental Endocrine
Disruptors; American Chemical Society: Washington, DC, 2000; Vol.
747,pp3352.
(19) Munguia-Lopez, E. M.; Gerardo-Lugo, S.; Peralta, E.; Bolumen,
S.; Soto-Valdez, H. Migration of bisphenol A (BPA) from can coatings
into a fatty-food simulant and tuna fish. Food Addit. Contam. 2005,
22, 892–898.
(20) Rastkari, N.; Ahmadkhaniha, R.; Yunesian, M.; Baleh, L. J.;
Mesdaghinia, A. Sensitive determination of bisphenol A and bisphenol F
in canned food using a solid phase microextraction fibre coated with
single walled nanotubes before GC/MS. Food Addit. Contam., Part A
2010,27, 1460–1468.
(21) Sajiki, J.; Miyamoto, F.; Fukata, H.; Mori, C.; Yonekubo, J.;
Hayakawa, K. Bisphenol A (BPA) and its source in foods in Japanese
markets. Food Addit. Contam. 2007,24, 103–112.
(22) Yonekubo, J.; Hayakawa, K.; Sajiki, J. Concentrations of bi-
sphenol A, bisphenol A diglycidyl ether, and their derivatives in canned
foods in Japanese markets. J. Agric. Food Chem. 2008,56, 2041–2047.
(23) Yoshida, T.; Horie, M.; Hoshino, Y.; Nakazawa, H. Determina-
tion of bisphenol A in canned vegetables and fruit by high performance
liquid chromatography. Food Addit. Contam. 2001,18,69–75.
(24) Bendito, M. D. P.; Bravo, S. R.; Reyes, M. L. L.; Prieto, A. G.
Determination of bisphenol A in canned fatty foods by coacervative
microextraction, liquid chromatography and fluorimetry. Food Addit.
Contam., Part A 2009,26, 265–274.
(25) Thomson, B. M.; Grounds, P. R. Bisphenol A in canned foods in
New Zealand: An exposure assessment. Food Addit. Contam. 2005,
22,65–72.
(26) Cao, X.-L.; Corriveau, J.; Popovic, S. Bisphenol A in canned
food products from Canadian markets. J. Food Prot. 2010,73, 1085–
1089.
(27) Schecter, A.; Malik, N.; Haffner, D.; Smith, S.; Harris, T. R.;
Paepke, O.; Birnbaum, L. Bisphenol A (BPA) in U.S. food. Environ. Sci.
Technol. 2010,44, 9425–9430.
(28) Vi~nas, P.; Campillo, N.; Martinez-Castillo, N.; Hernandez-
Cordoba, M. Comparison of two derivatization-based methods for
solid-phase microextraction-gas chromatography-mass spectrometric
determination of bisphenol A, bisphenol S and biphenol migrated from
food cans. Anal. Bioanal. Chem. 2010,397, 115–125.
(29) Geens, T.; Apelbaum, T. Z.; Goeyens, L.; Neels, H.; Covaci, A.
Intake of bisphenol A from canned beverages and foods on the Belgian
market. Food Addit. Contam., Part A 2010,27, 1627–1637.
(30) Bureau of Chemical Safety. Survey of bisphenol A in canned food
products from Canadian markets; 978-1-100-16029-0; Health Canada:
Ottawa, ON, 2010.
(31) Ackerman, L. K.; Noonan, G. O.; Begley, T. H.; Mazzola, E.
Accurate mass and NMR identification of bisphenolic can coating
migrants and their interference with LC-MS/MS analysis of BPA. Rapid
Commun. Mass Spectrom. 2011,25, 1336–1342.
(32) Petersen, H.; Schaefer, A.; Buckow, C. A.; Simat, T. J.; Steinhart,
H. Determination of bisphenol A diglycidyl ether (BADGE) and its
derivatives in food: Identification and quantification by internal standard.
Eur. Food Res. Technol. 2003,216,355–364.