reduced calcium intake
and hypertension, and because hypertension has not been reported in
animals receiving the recommended dietary concentration of calcium, the data from Perry et al.
(1989, 1985) were not considered further in the derivation of the RfD.
Acute hypertension has been observed in humans after accidental or intentional ingestion
of soluble barium salts (CDC, 2003; Downs et al., 1995). Two human studies have investigated
the effects of longer-term barium ingestion on blood pressure (Wones et al., 1990; Brenniman et
al., 1981). Both investigations found no hypertensive effect with their highest exposure
concentrations. Brenniman and Levy (1984) found no effect on hypertension between two
communities with a 70-fold difference in the barium concentrations of their drinking water.
Wones et al. (1990) found no hypertensive effect in a before and after comparison of 11 subjects
that were exposed to two concentrations of barium in their drinking water over the course of 10
weeks. Coincidently, the same NOAEL of 0.21 mg/kg-day was identified for both studies.
These NOAELs were estimated by EPA using standard estimates for drinking water intake
(2 L/day) and average body weight (70 kg).
Neither Brenniman et al. (1981) nor Wones et al. (1990) provided sufficient data to
support or refute the hypothesis that chronic barium exposure causes hypertension. Hypertension
is a complex multifactorial condition, and it is very possible that the effect of chronic barium
exposure on blood pressure is relatively small compared to other determinates, such as diet and
exercise. Wones et al. (1990) attempted to control for the effect of diet by providing a standard
diet to all of the study participants. Unfortunately, the power of this study was limited by the
very small number of participants (n=11). They also used short exposure durations (4 weeks for
each exposure concentration), which may not have been sufficient to observe a chronic effect.
Brenniman et al. (1981) also examined a relatively small number of subjects (n=85) in the
subpopulation that was controlled for key risk factors. Other limitations of Brenniman et al.
(1981) were that they collected replicate blood pressure measurements from individuals during a
single 20-minute period, they used community-wide exposure estimates, and they didn’t control
for a number of important risk factors for hypertension, including diet and exercise. In the
absence of dose-response data for barium-induced hypertension, the RfD was not based on this
effect.
The effect of barium on reproductive functions was evaluated in rats and mice by Dietz et
al. (1992). A significant reduction in litter size was observed in mice receiving a barium dose of
approximately 100 mg/kg-day, but a dose of approximately 200 mg/kg-day did not produce that
effect. Birth weight in rat pups was significantly reduced in the 200 mg/kg-day treatment group,
42
but no effect was observed at postnatal day 5. The
observed effects, decreased birth weight and
decreased litter size, were either transient or not dose-dependent. These data suggest that any
potential reproductive effect of barium is likely to occur at a dose higher than that found to
produce nephropathy in mice.
In consideration of the available data on the adverse effects of chronic and subchronic
barium ingestion in humans and animals, the increased incidence of chemical-related
nephropathy in mice provides the best evidence of a dose-response relationship. For this reason,
the chronic mouse study conducted by NTP (1994) was selected as the principal study and
nephropathy was identified as the critical effect for deriving the RfD.
43
Table 5–1. Effects of subchronic and chronic oral barium exposure on rodents
Species
Duration
Sex
Estimated barium
doses (mg/kg-day)
Incidence of
nephropathy
Effect on kidney weight
Rat
13 weeks
0, 10, 30, 65, 110,
Control: 0/10
Increased relative wt.
M
200
High dose: 3/10
(200 mg/kg-day)
0, 10, 35, 65, 115,
Control: 0/10
High dose: 3/10
Increased relative wt.
(
$65 mg/kg-day); Increased
relative wt. & absolute wt.
F
180
(
$115 mg/kg-day)
2 years
M
0, 15, 30, 60
Control: 46/47
High dose: 47/49
Decreased absolute wt.
(
$30 mg/kg-day)
Control: 43/48
Increased relative wt.
F
0, 15, 45, 75
High dose: 48/50
(
$45 mg/kg-day)
Mouse
13 weeks
0, 15, 55, 100, 205,
Control: 0/10
Decreased absolute wt.
M
450
High dose: 10/10
(450 mg/kg-day)
0, 15, 60, 110, 200,
Control: 0/10
Increased relative wt.
F
495
High dose: 9/10
(495 mg/kg-day)
15
Control: 1/59
months/
Inter. dose: 2/58
No effect
2 years
a
M
0, 30, 75, 160
High dose: 19/60
(160 mg/kg-day)
Control: 0/60
Inter. dose: 1/60
No effect
F
0, 40, 90, 200
High dose: 37/60
(200 mg/kg-day)
a
Animals from both the 15-month and 2-year evaluations were considered in this evaluation because of the
reduced life expectancy of mice in the high dose group.
Source: NTP, 1994.
5.1.2. Methods of Analysis
The incidence of nephropathy in mice chronically exposed to barium in drinking water
was modeled using EPA’s Benchmark Dose Modeling Software Version 1.3.2 (U.S. EPA,
BMDS). All of the available models for dichotomous endpoints were fitted to the incidence data
shown in Table 5–2. Details of the modeling and the model output for the best fitting model are
provided in Appendix B. Best fit was determined using the criteria in the draft Benchmark Dose
Technical Guidance Document (U.S. EPA, 2000c): the lowest Akaike Information Criterion
(AIC) among the models with adequate fits (p>0.1). Third degree and fifth degree multistage
models provided the best fit for the male and female data, respectively; these models are
summarized in Table 5–3. These best-fitting models also had the lowest benchmark doses
(BMDs) and BMDLs (95% lower bound on benchmark dose) for each data set.
44