Bisphenol A Risk Assessment Document

(AIST Risk Assessment Document Series No. 4)

Summary

November 2007

Edition
New Energy and Industrial Technology Development Organization (NEDO)
and
Research Center for Chemical Risk Management (CRM)
National Institute of Advanced Industrial Science and Technology (AIST)

Authors
Junko Nakanishi, NEDO Project Leader, CRM, AIST
Ken-ichi Miyamoto, CRM, AIST
Hajime Kawasaki, CRM, AIST
 Chapter I. Introduction

Chapter I summarizes the background of the bisphenol A (BPA) issue. In addition, it provides
general information such as identification data, physicochemical properties, and reviews of 12 risk/hazard
assessment documents published in the last 7 years.

Around 1996, BPA began to attract considerable interest as a suspected endocrine-disrupting

chemical. Since 1998, the issue of low-dose effects has raised considerable concern. In Japan, the Ministry
of Health and Welfare (1998a), the Ministry of Health, Labour and Welfare (2001a), the Ministry of
Economy, Trade and Industry (2002), the Ministry of the Environment (2004a), etc. convened panels of
experts. Although they did not recommended prohibiting or restricting the use of BPA, some of them
recommended a comprehensive assessment of the risks posed by BPA. In this risk assessment, a large
amount of data regarding the exposure and hazards of BPA were reviewed and reanalyzed. The risk was
characterized not only by conventional approaches, such as the hazard quotient method or the margin of
exposure (MOE) method, but also by more sophisticated approaches including Monte Carlo simulations
that propagate uncertainties in exposure parameters and a population-level ecological risk assessment
technique.

BPA is a white solid at room temperature with a molecular weight of 228.29 and a melting point
of 150°C to 155°C. It vaporizes slightly at room temperature due to its low vapor pressure of 5 × 10–6 Pa at
20°C. It is specified as a “Class I Designated Chemical Substance” according to the “Law Concerning
Reporting, etc. of Releases to the Environment of Specific Chemical Substances and Promoting
Improvements in Their Management.” Based on the Food Sanitation Law, its migration limit in
food-contact polycarbonate plastics is specified as 2.5 ppm.

During the past 3 years, major risk assessments of BPA have been implemented by the (1) OECD
(2002), (2) EU Scientific Committee on Food (2002), (3) Chemicals Evaluation and Research Institute,
Japan (2002), (4) European Commission (2003), (5) Ministry of the Environment (2004a), and (6) the
Ministry of the Environment (2004b). During the past 7 years, hazard assessments have been conducted by
the (7) Ministry of Health and Welfare (1998a), (8) Ministry of Health, Labour and Welfare (2001a), (9)
National Toxicology Program (2001), (10) US EPA (2002), (11) Ministry of Economy, Trade and Industry
(2002), and (12) Ministry of the Environment (2004c). Each of these assessments has been summarized in
this chapter.

1
 Chapter II. Production, uses, and environmental releases

Chapter II summarizes the data on the production, uses, and environmental releases of BPA.
Domestic BPA manufacturers supplied approximately 580,000 tons, and domestic consumption was
approximately 430,000 tons in 2003. Approximately 72% of the domestic consumption was associated with
the production of polycarbonate (PC), and approximately 16% with the production of epoxy resins (EXRs).
The other applications include the manufacture of polyester resin intermediates, flame retardants, and
hydrogenated bisphenol A.

Polyvinyl chloride (PVC) industries voluntarily reduced the amount of BPA used as an additive in
the production of PVC; in 2003, the amount reduced to 100 tons. Thermal paper manufacturers also
voluntarily substituted BPA used as a developing agent; according to available information, it appeared that
the substitution was almost completed in 2003.

Chapter III. Environmental fate characteristics and monitoring data

Chapter III describes environmental fate characteristics and environmental concentrations

obtained from various monitoring programs. The fate characteristics comprise degradation and distribution
properties. The half-life for the reaction of BPA with hydroxyl radicals in the atmosphere is calculated as
approximately 0.2 days. According to aquatic biodegradation studies, BPA appears to be readily
biodegradable.

The environmental monitoring of BPA in surface water, sediment, ground water, the atmosphere,
soil, aquatic and terrestrial organisms, sewage treatment facilities, rainwater, drinking water treatment
plants, and tap water was conducted extensively by the Ministry of the Environment, Ministry of Land,
Infrastructure and Transport, local governments, and many researchers. BPA concentrations were measured
in 1,120 areas with fresh surface water. The average concentrations were 0.005 µg/L or less in
approximately 30% of these areas and 1 µg/L or less in 99% of the areas. BPA concentrations were also
measured in 187 areas with sea water. The average concentrations were 0.005 µg/L or less in almost half
the areas and 1 µg/L or less in all the areas with the exception of one. No apparent trend was observed in
the temporal change patterns of the surface water concentrations.

The highest concentrations were reported as 22 µg/L in surface water, 1,100 µg/kg-dry in

2
sediment, 3.29 µg/L in ground water, 28 ng/m3 in the atmosphere, 30 µg/kg in fish and shellfish, 70
µg/kg-wet in terrestrial organisms, 3.9 µg/L in influent to sewage treatment plants, 0.42 µg/L in water
released from sewage treatment plants, 0.04 µg/L in rainwater, 0.06 µg/L in influent to drinking water
treatment plants, 0.01 µg/L in purified water in drinking water treatment plants, and 0.007 µg/L in tap
water.

Chapter IV. Human health risk assessment

1. Hazard identification and dose-response assessment

This section presents an overview of the toxicological profile of BPA, reviews of published
hazard assessment documents, and no observed adverse effect levels (NOAELs) and a benchmark dose
lower confidence limit (BMDL) that are used for characterizing human health risk. The key toxicological
endpoints for BPA include photosensitization, reduction in body weight gain, effects on the liver, and
reproductive toxicity.

For photosensitization, no established method is available for extrapolating animal toxicological

data to effects on humans. However, the photosensitization will be insignificant in ambient exposure,
although careful control of occupational exposure, which is beyond the scope of this assessment, may be
required.

The NOAEL for the reduction in body weight gain by oral administration was identified as 5
mg/kg/day based on the results of a three-generation study in rats. During the evaluation of the MOEs
based on this NOAEL, it was concluded that there was no unacceptable risk when the MOEs exceeded 100,
which represented uncertainties in species differences (10) and individual differences (10).

The BMDL for the observed liver effect (multinucleated giant hepatocytes) was identified as 23
mg/kg/day based on a reanalysis of the results of a continuous breeding study in mice. During the
evaluation of the MOEs based on this BMDL, it was concluded that there was no unacceptable risk when
the MOEs exceeded 500, which represented uncertainties in species differences (10), individual differences
(10), and extrapolation from short-term to long-term exposure durations (5).

In relation to the reproductive toxicity, the NOAEL for effects on fertility was identified as 50
mg/kg/day based on the three-generation study in rats. During the evaluation of the MOEs based on this
NOAEL, it was concluded that there was no unacceptable risk when the MOEs exceeded 100, which

3
represented uncertainties in species differences (10) and individual differences (10).

An additional uncertainty due to the low-dose effects was not incorporated because the findings
in the low-dose studies were not robust, while those in negative studies were consistent.

2. Exposure assessment
Daily BPA intakes were estimated by employing two different approaches. In the first approach,
exposure levels from possible sources (atmosphere, water, food, tableware, toys, etc.) were estimated and
aggregated. Since the parameters required in this approach, such as the amount of food consumption,
breathing volume, and frequency of PC tableware use, depended on age, the daily BPA intakes were
separately estimated for 6 age groups: infants aged 0–5 months, infants aged 6–11 months, children aged
1–6 years, students aged 7–14 years, young persons aged 15–19 years, and adults aged >19 years. In the
second approach, the daily BPA intakes were determined from urinary excretion by backward calculations.
In both approaches, the parameters required for the calculation of the BPA intakes were characterized as
distributions, and Monte Carlo simulations were performed to propagate uncertainty and variability in the
parameters. Furthermore, it was possible to analyze uncertainty in the functional relationship between
sources and exposure by comparing the results of the two approaches.

The first approach estimated that children aged 1–6 years had the highest level of exposure; in
1998, their average daily intake was 1.2 µg/kg/day. This was due to relatively high dietary consumption per
body weight and the use of PC tableware for this age class. However, in recent years, the BPA exposure
from PC tableware reduced because in 1998, social concern about endocrine-disrupting chemicals
increased in Japan and certain proportions of PC tableware as well as PC feeding bottles were substituted
with non-PC articles thereafter. In students aged 7–14 years, the level of exposure was the second highest
with an average daily intake of 0.55 µg/kg/day. The average daily intakes for infants aged 0–5 months,
infants aged 6–11 months, young persons aged 15–19 years, and adults aged >19 years were estimated as
0.028–0.055, 0.16–0.18, 0.36, and 0.43 µg/kg/day, respectively (the daily intake indicates the average for
males in 1998).

For individuals aged 6 months and above, food was the most significant BPA source. For
individuals aged 1–14 years, the intake levels of BPA from canned food and drinks were almost equivalent
to those from non-canned food and drinks. On the other hand, for individuals aged 15 years and above, the
intake levels from canned food and drinks were approximately twice those from non-canned food and
drinks. This would reflect a difference in the consumption of canned beverages that contained a relatively

4
large amount of BPA.

In the second approach, the 95% confidence intervals of the daily intake for high-exposure
populations (i.e., 95th percentiles) were estimated as 0.037–0.064 µg/kg/day for adult males and
0.043–0.075 µg/kg/day for adult females. Further, the values for average-exposure populations were
0.028–0.049 µg/kg/day for adult males and 0.034–0.059 µg/kg/day for adult females.

The average and 95th percentile intakes estimated by the first approach were 4–7 and 7–13 times
those estimated by the second approach, respectively. These differences resulted from the assumption of the
first approach that BPA concentrations in food items, which were in fact highly skewed toward 0, were
characterized by uniform distributions. Therefore, the values estimated by the second approach appeared to
be more realistic than those of the first approach.

3. Risk characterization
Human health risk was characterized by using the MOEs that were calculated by dividing the
NOAEL or BMDL by the daily BPA intakes. For children aged 1–6 years who had the highest exposure
level, the MOEs calculated by using the average daily intake were 4,200 for the reduction in body weight
gain, 19,000 for multinucleated giant hepatocytes in the liver, and 42,000 for the reproductive toxicity. The
MOEs estimated by using the 95th percentile of the daily intake were 1,200–1,300 for the reduction in body
weight gain, 5,600–5,900 for multinucleated giant hepatocytes in the liver, and 12,000–13,000 for the
reproductive toxicity. The MOEs were sufficiently large for all three endpoints even in the highest exposure
group. The MOEs for the other age classes were in the range of several thousands to several hundreds of
thousands. Larger MOEs, which were in the range of several tens of thousands to a little over one million,
were obtained by using the daily intakes estimated from the urinary excretion (Table 1).

Based on the following results, it is concluded that the current exposure levels of BPA will not
pose any unacceptable risk to human health.
1. The MOEs were sufficiently large for all three endpoints even in the highest exposure group.
2. It is highly likely that the daily BPA intake for the highest exposure group was overestimated due to
conservative assumptions.
3. No potential sources appeared to increase the BPA exposure, although some activities such as the
substitution of PC tableware used for school lunches reduced it since 1998.

5
Table 1 MOEs for each endpoint (reduction in body weight gain, multinucleated giant hepatocytes in the liver, and reproductive toxicity)
Multinucleated giant hepatocytes
Reduction in body weight gain Reproductive toxicity
Exposure in the liver
Age class (sex) Period NOAEL = 5 mg/kg/day NOAEL = 50 mg/kg/day
estimation BMDL = 23 mg/kg/day
Average 95th percentile Average 95th percentile Average 95th percentile
0–5 months (m) 1998 91,000 45,000 420,000 210,000 910,000 450,000
0–5 months (f) 1998 81,000 31,000 370,000 140,000 810,000 310,000
6–11 months (m) 1998 28,000 15,000 130,000 68,000 280,000 150,000
6–11 months (f) 1998 25,000 13,000 120,000 59,000 250,000 130,000
1–6 years (m) 1998 4,200 1,300 19,000 5,900 42,000 13,000
1–6 years (f) 1998 4,200 1,200 19,000 5,600 42,000 12,000
7–14 years (m) 1995–2000 8,600–10,000 3,600–4,200 40,000–46,000 16,000–19,000 86,000–100,000 36,000–42,000
7–14 years (m) 2001–2002 14,000–15,000 6,300–6,500 64,000–68,000 29,000–30,000 140,000–150,000 63,000–65,000
1st 7–14 years (f) 1995–2000 9,400–12,000 3,800–5,000 43,000–53,000 18,000–23,000 94,000–120,000 38,000–50,000
approach 7–14 years (f) 2001–2002 15,000 6,500–6,700 68,000–70,000 30,000–31,000 150,000 65,000–67,000
15–19 years (m) 1995–2000 13,000–17,000 4,500–6,500 58,000–77,000 21,000–30,000 130,000–170,000 45,000–65,000
15–19 years (m) 2001–2002 25,000 11,000 120,000 50,000–52,000 250,000 110,000
15–19 years (f) 1995–2000 15,000–17,000 5,900–7,400 68,000–79,000 27,000–34,000 150,000–170,000 59,000–74,000
15–19 years (f) 2001–2002 24,000–25,000 10,000 110,000–120,000 47,000 240,000–250,000 100,000
>19 years (m) 1995–2000 11,000–13,000 4,200–5,000 51,000–61,000 19,000–23,000 110,000–130,000 42,000–50,000
>19 years (m) 2001–2002 26,000 11,000 120,000 52,000 260,000 110,000
>19 years (f) 1995–2000 14,000–16,000 5,400–6,200 64,000–72,000 25,000–28,000 140,000–160,000 54,000–62,000
>19 years (f) 2001–2002 22,000 8,900–9,100 100,000 41,000–42,000 220,000 89,000–91,000
2nd Adult (m) Recent years 100,000–180,000 78,000–140,000 470,000–820,000 360,000–620,000 1,000,000–1,800,000 780,000–1,400,000
approach Adult (f) Recent years 85,000–150,000 67,000–120,000 390,000–680,000 310,000–530,000 850,000–1,500,000 670,000–1,200,000

6
 The conclusion of the risk assessment by the European Commission is “there is need for further
information and/or testing in relation to developmental toxicity” and “there is at present no need for further
information and/or testing and for risk reduction measures beyond those which are being applied already in
relation to all other endpoints.” The other human health risk assessments summarized in Chapter I conclude
that the risks are below the levels of concern. Our conclusion is the same as that of most existing risk
assessments. They assume worst-case scenarios and calculate the point estimates of exposure, while we
consider more realistic scenarios and the distribution of exposure with two different approaches.

With regard to the nature of uncertainty in risk assessment, it is possible to distinguish between
stochastic variability and “lack of knowledge” uncertainty (NCRP 1996). This assessment incorporated the
stochastic variabilities of most exposure parameters; however, we could not characterize the individual
differences in canned-food consumption and those in the elimination half-life of BPA in the body, whose
nature is categorized as stochastic variability. The major influential components of the “lack of knowledge”
uncertainty in this assessment includes uncertainties that originated from the assumptions of the shapes of
the concentration distribution in exposure media, such as food items, and the frequency of the use of
EXR-coated chopsticks. Most of these concentration distributions were characterized by uniform
distributions that would result in the overestimation of the frequency of higher concentrations. As a
worst-case scenario, the frequency of the use of EXR-coated chopsticks was assumed to be 100%. Detailed
information on these parameters will improve the estimation of the daily BPA intake; however, it will not
change the current conclusion of this risk assessment.

Chapter V. Ecological risk assessment

1. Problem formulation
An assessment endpoint of ecological risk assessment is an explicit expression of the actual
environmental value that is to be protected, operationally defined by an ecological entity and its attributes.
The following three assessment endpoints were selected to assess the impact of BPA on the sustainability of
local populations of aquatic life, particularly fish: (1) survival, reproduction, growth, and development of
susceptible aquatic species; (2) the growth rates of local fish populations including white-spotted char
(Salvelinus leucomaenis), pale chub (Zacco platypus), Japanese dace (Tribolodon hakonensis), barbel steed
(Hemibarbus barbus), and nekogigi (Pseudobagrus ichikawai); and (3) the presence and conditions of fish
species confirmed by field observations in highly contaminated areas in Japan (Fig.1).

7
 Objective
Assessing risks to local populations of aquatic life, particularly fish

Assessment endpoint (1)

Survival, reproduction, growth, and development of susceptible aquatic species

Yes MOE > criterion No

value

It is presumed that populations of There is a possibility that survival,

susceptible species will not decline since reproduction, growth, and/or development of
End survival, reproduction, growth, and some individuals will be affected; however,
development of the susceptible species it is not necessarily true that the effects will
will not be affected significantly. propagate to the population level.

Assessment endpoint (2) Assessment endpoint (3)

Growth rates of local populations of 5 fish Presence and conditions of fish species
species (white-spotted char, pale chub, confirmed by field observations in
Japanese dace, barbel steed, and nekogigi) highly contaminated areas

Yes ri > 0 No Yes The presence of No

fish was confirmed
in an area.

The sustainability of There is a possibility Some fish can There is a possibility

the local populations that the population of inhabit the area at that fish cannot
of 5 fish species will 5 fish species will least for a short inhabit the area.
not be impaired. decline. period of time.

Drawing an overall conclusion regarding the sustainability of aquatic life, particularly fish

Fig.1 Framework of ecological risk assessment for BPA

In ecological risk assessment, one of the inherent difficulties is how to deal with uncertainties due
to our incomplete knowledge about an ecosystem of concern. In order to assess the risks of a chemical
substance to an ecosystem from various viewpoints that consider such uncertainties, it is preferable to
establish a series of assessment endpoints that reflect multifaceted ecological values instead of selecting a
single assessment endpoint. The overall conclusion was drawn based on the outcomes of the evaluation of
all three assessment endpoints.

8
2. Effects assessment
With regard to acute toxicity for freshwater algae, a single species of green alga
Pseudokirchneriella subcapitata was tested by two groups (Ministry of the Environment 2004d, Alexander
et al. 1988). Their results indicated that EC50 values ranged from 2,730 to 4,900 µg/L.

The acute toxicity for freshwater invertebrates was tested based on the immobilization of
Daphnia magna. The values of 48 h EC50 ranged from 10,200 to 13,000 µg/L (Ministry of the Environment
2004d, Alexander et al. 1988). LC50s for shrimp Gammarus pulex were reported as 12,800 µg/L in 24 h,
5,600 µg/L in 48 h, and 1,500 µg/L in 5–10 days (Watts et al. 2001).

The acute toxicity was also determined for rainbow trout (Oncorhynchus mykiss), Japanese
medaka (Oryzias latipes), swordtail fish (Xiphophorus helleri), and fathead minnow (Pimephales promelas),
which are freshwater fish. The toxicity values of 96 h LC50 for these fish species ranged from 3,000 to
17,930 µg/L (Reiff 1979，Kwak et al. 2001). The differences in the sensitivity of the fish were relatively
small; they were up to approximately 6-fold.

The chronic toxicity for green alga Pseudokirchneriella subcapitata was reported in terms of
NOECs of 320–1,800 µg/L by the Ministry of the Environment (2004d). Alexander et al. (1988) reported a
96 h NOEC of 1,200 µg/L for the same species based on cell count and total cell volume.

The chronic toxicity for invertebrates was reported by Caspers (1998) and the Ministry of the
Environment (2004d) based on 21-day Daphnia magna reproduction tests. NOECs ranged from 3,160 to
4,600 µg/L.

In order to determine the chronic toxicity for fish, tests were conducted on fathead minnow,
Japanese medaka, and zebrafish (Danio rerio). The smallest NOEC was reported as 16 µg/L; it was
identified in a three-generation study on fathead minnow based on the hatchability of the F2 generation
(European Commission 2003).

In addition to the conventional endpoints of ecotoxicological studies, the suspected

endocrine-disrupting effects were reviewed in this section. Furthermore, the relationship between such
effects and the assessment endpoints was discussed.

3. Exposure assessment

9
 We obtained a total of 3,956 observations of the BPA concentration in fresh surface water
covering 1,120 areas in 752 rivers in Japan. Each area was observed once or up to several tens times. The
average concentrations were 1 µg/L or less in approximately 99% of the areas (Fig.2). However, in
approximately 30% of the areas, the average concentrations were below the quantification limit. Relatively
high concentrations of BPA (1.5 µg/L or higher) were detected in the Suikawa River (Iwate Prefecture),
Edogawa River (Chiba Prefecture), Lake Teganuma (Chiba Prefecture), Ayase River (Tokyo), Tsurumi
River (Kanagawa Prefecture), Numakawa River (Shizuoka Prefecture), Itonuki River (Gifu Prefecture),
Yagou River (Mie Prefecture), Mitaki River (Mie Prefecture), Tenpaku River (Mie Prefecture), Amaike
River (Mie Prefecture), Kasama River (Nara Prefecture), and Hirokawa River (Fukuoka Prefecture).
Leachate from landfills and effluents from paper mills were suspected to be primary sources of BPA in
some of these rivers. Specific BPA sources were not identified for urban rivers such as the Edogawa River
and the Ayase River whose basin had insufficient wastewater treatment systems due to rapid population
growth and urbanization. However, in the urban rivers, high concentrations were detected only when some
specific conditions were satisfied since they were observed infrequently.

10
 99.99
99.9

99
95
90
80
70
Percentile

50
30
20
10
5
1

.1
.01
0.001 0.01 0.1 1 10 100
BPA concentration [µg/L]
Fig.2 Cumulative distribution of average BPA concentration in 1,120 areas with fresh surface water

4. Risk characterization
4.1 Assessment endpoint (1): Survival, reproduction, growth, and development of susceptible aquatic
species
A hazard quotient (HQ) is defined as

C
HQ = ,
TV / AF

where C is the BPA concentration in water [µg/L], TV is the smallest toxicity value of LC50 or NOAEL
[µg/L], and AF is the assessment factor. The value of AF is 10 when chronic NOEC values of at least three
species representing three trophic levels as well as one acute L(E)C50 value from each of the three trophic
levels of fish, zooplankton, and algae are available.

On the other hand, an MOE is defined by the following equation:

TV
MOE = .
C

In this assessment, a TV of 16 µg/L was considered adequate according to the results of the
three-generation study on fathead minnow. When the HQ is less than one, i.e., the MOE is greater than the

11
AF, it is presumed that even the populations of susceptible species will not decline because it is expected
that their mortality, growth, reproduction, and development, which govern the population dynamics, are not
affected significantly.

Based on the maximum BPA concentrations, MOEs in 19 areas of 12 rivers and a lake were
smaller than the AF of 10. Based on the average BPA concentrations, MOEs in 7 areas of 6 rivers were less
than 10. The areas in which the MOEs were less than 10 were identified as targets for further assessments
from the viewpoint of assessment endpoints (2) and (3). In the other areas of more than 1,100, it was
presumed that BPA did not decrease the populations of susceptible species.

4.2 Assessment endpoint (2): Growth rates of local populations of 5 fish species
The sustainability of the local populations of the 5 fish species—white-spotted char (Salvelinus
leucomaenis), pale chub (Zacco platypus), Japanese dace (Tribolodon hakonensis), barbel steed
(Hemibarbus barbus), and nekogigi (Pseudobagrus ichikawai)—was assessed by using their instantaneous
population growth rate (ri) values. These values were calculated as the natural logarithms of finite
population growth rates (λ). λ can be expressed as the dominant eigenvalue of a population projection
matrix (also known as a Leslie matrix), which has positive entries only in the first row (fertilities) and the
subdiagonal (survival probabilities). An instantaneous population growth rate (ri) can also be expressed as
ri = ln (Nt+1/Nt), (1)
where Nt is the number of individuals at time t and Nt+1 is the number of individuals at time t + 1.
Therefore, if ri is greater than 0, the population will increase during the period from t to t + 1. If ri is less
than 0, the population will decrease, and if ri is exactly 0, it will remain unchanged during that period. The
parameters required for calculating the values of ri were obtained from literature or estimated by a newly
developed method. In general, an instantaneous population growth rate (ri) often decrease with an increase
in the population density except when Allee effects occur. The maximum ri is referred to as the intrinsic
rate of population growth (rm).

Figure 3 shows the effect of the BPA concentration in water on the conservatively estimated ri of
the 5 fish species. The maximum BPA concentrations of the areas in which the MOEs < 10 were less than
1/8 of the concentrations at which the conservatively estimated ri became 0. The average BPA
concentrations of the areas in which the MOEs < 10 were less than 1/23 of the concentrations at which the
conservatively estimated ri became 0. Therefore, it is unlikely that the 5 fish species are at unacceptable
risks in terms of population sustainability due to the current levels of BPA in ambient water.

12
 0.2

Instantaneous population growth rate ri [1/yr]

0
white-spotted char
pale chub
Japanese dace
-0.2 barbel steed
nekogigi

-0.4

-0.6

-0.8
0 500 1000 1500
BPA concentration [μg/L]

Fig.3 Effect of BPA concentration in water on the conservatively estimated ri of 5 fish species

4.3 Assessment endpoint (3): Presence and conditions of fish species confirmed by field observations
in highly contaminated areas
Literature and our sighting survey confirmed that various fish species inhabited the areas in
which the average and/or maximum BPA concentrations were several µg/L or less. Information on the
presence of fish was not obtained for the areas in which the maximum BPA concentrations exceeded 10
µg/L; however, in the Yago River, 12 fish species including carp and Japanese dace were observed
approximately 5 km downstream in the area with average and maximum BPA concentrations of 4–7 µg/L
and 15–20 µg/L , respectively. Therefore, it was indicated that fish populations in rivers contaminated with
up to 20 µg/L of BPA did not become extinct readily.

4.4 Overall conclusion regarding the sustainability of aquatic life, particularly fish
In summary, the analysis under assessment endpoint (1) indicated that ecological risks posed by
BPA were below the level of concern in most areas. The analysis under assessment endpoint (2) confirmed
that the 5 surrogate fish species were unlikely to face unacceptable risks in terms of population
sustainability due to the current levels of BPA in ambient water. The analysis under assessment endpoint (3)
proved that fish populations in rivers contaminated with up to 20 µg/L of BPA do not become extinct
readily. The results of assessment endpoints (2) and (3) were consistent with each other. From these
considerations, the current exposure levels of BPA will not pose unacceptable risks to the local populations

13
of aquatic life, particularly fish.

Chapter VI. Economic impact analysis of risk reduction activities

In this chapter, costs and reduced BPA intakes were assessed for two risk reduction activities
conducted in Japan: the substitution of PC tableware used for school lunches and the voluntary alternation
of the method for inactivating the inner surface of drink cans. It was expected that this assessment would
present a viewpoint for considering the problem of endocrine-disrupting chemicals rather than supporting
future BPA control.

1. Substitution of PC tableware used for school lunches

According to surveys conducted by the Ministry of Education (1998b, 1999b), PC tableware was
used in 1,686 municipalities in May 1998; however, more than 300 of these municipalities substituted it
with tableware of different materials within a year. Furthermore, as of May 1999, approximately 500
municipalities planned to change their PC tableware. In order to obtain detailed information on the
substitution including time, scale, costs, etc., we distributed questionnaires to 10 municipalities randomly
sampled from those that had already substituted their PC tableware as of May 1999 and 90 municipalities
randomly sampled from those that planned to change their PC tableware as of May 1999. Sixty
municipalities responded to the questionnaires (60% recovery rate).

Among 54 municipalities that were able to confirm the past use of PC tableware, 48 substituted it
with tableware of different materials and 6 did not. Polypropylene was the most popular material for the
substitution; it was selected by 38.3% of the municipalities that substituted their PC tableware. Ceramic
(including reinforced porcelain) was the second most popular material used by 18.3% of the municipalities.
Other materials included melamine, ABS resin, polyethylene naphthalate, and stainless steel.

The reason(s) for the substitution (multiple answers allowed) were stated in the questionnaire:
96% of the municipalities substituted PC tableware because of concerns about endocrine-disrupting effects
of BPA and 16% substituted it because of respect for the food culture in Japan that traditionally used
ceramic tableware. In fact, all the municipalities that substituted their PC tableware because of respect for
the food culture selected ceramic tableware.

Collateral investments in facilities were made only when PC tableware was substituted with
ceramic tableware. Three of the 11 municipalities that introduced ceramic tableware replaced their
dishwashers. However, this was not only because they were unsuitable for fragile ceramic tableware, but

14
also because they were decrepit. Six of the above 11 municipalities replaced and added sterilization
cabinets and/or movable sinks. Three of the 6 replaced and added them because ceramic tableware required
more space than PC tableware and the existing ones were aging. The other 3 municipalities added them
only because of the increased handling space required for new tableware. Since sterilization cabinets and
movable sinks were less expensive than dishwashers, they were introduced additionally even before their
replacement time.

The net cost for the reduction in the BPA intake by the substitution of PC tableware is the
difference between the total expenses incurred only for reducing the BPA exposure and the expenses
expected if PC tableware were not substituted. The total expenses include those incurred for the first
replacement of the tableware and related facilities, their additional introduction at the time of substitution,
and subsequent replacements at the end of their lifetime. By assuming the discount rate as 3%, the
incremental national costs over 30 years from 1996 to 2025 were estimated to be 11 billion yen, which is
equivalent to 370 million yen per year or 127 yen per year per student who uses the substituted tableware.
The reduction in the daily BPA intake by this substitution was estimated to be 0.2–0.3 µg/kg/day.

2. Alternation of the method for inactivating the inner surface of drink cans
In 1998, 0.6–10 µg/L of BPA was detected from 12 out of 20 canned drinks during product
testing at the Hokkaido Consumer Affairs Center. The center then instructed the related industries to
prevent the migration of BPA (Hokkaido Consumer Affairs Center 2003). This would become a part of
consumer demands to manufacturers. Can manufacturers were asked to reduce the migration level to almost
0 by canned-drink manufacturers. Since the detection limit of BPA at that time was 5 µg/L, reduction below
5 µg/L was voluntarily set as a control standard.

Two approaches were employed to reduce the migration of BPA. One was to change the inner
surface inactivation of cans from the EXR coating to the PET film lamination and the other was to use EXR
paint from which a small amount of BPA migrates. Since industrial equipment for manufacturing
film-laminated cans had already been developed and such cans were economically preferred, no facility
investment was made only for reducing the migration of BPA.

Due to the alternation of the method for inactivating the inner surface of drink cans, 0.1–0.2
µg/kg/day of the BPA intake was reduced for average-exposure individuals and 0.2–0.6 µg/kg/day was
reduced for high-exposure individuals (i.e., 95th percentiles).

15
 Chapter VII. Conclusion

This chapter summarizes Chapters I–VI and discusses future issues.

Since both the human health risk assessment (Chapter IV) and ecological risk assessment
(Chapter V) concluded that the risks posed by BPA were below the levels of concern, it will be unnecessary
to prohibit or restrict the use of BPA at this time.

The following points were identified as topics for future research and recommendations against
the current risk levels of BPA.
(1) A wide range of stakeholders including consumers, industry, and public sector regulators
should continue to share information on the risks posed by BPA in order to achieve a broad
consensus on the management of these risks. It would be necessary to create an environment
that encourages the disclosure of the voluntary management activities of industries without
suffering unfair disadvantages from the disclosure, particularly in industries that manufacture
food articles and products for infants and children.
(2) There exist conflicting observations in low-dose developmental toxicity studies in mice.
Further research will be required to understand the factors, such as housing, diet, species,
strain and substrain, and study design, that influence the outcome and variabilities in
developmental toxicity studies.
(3) Leachate from landfills and effluents from paper mills were suspected to be the sources that
discharged BPA into rivers wherein relatively high BPA concentrations were observed.
Although the observed BPA levels were unlikely to pose any unacceptable risks, it would be
useful to continue monitoring the BPA concentrations around the suspected sources in order
to obtain valuable information for managing the aquatic ecosystems. In addition, it would be
preferable to continue monitoring the BPA levels in selected areas in terms of conservation
priorities, locations of potential sources, and the possibility of pollution.
(4) In ecological risk assessments, it is sometimes one of the most difficult tasks to adequately
treat uncertainties due to luck of knowledge. To reduce such uncertainties, it is essential to
develop sophisticated approaches such as quantitative risk analysis that effectively
incorporates systematic field surveys and ecological theories.

16