Toxicological Review of Barium and Compounds

Table 5–2.  Nephropathy in B6C3F1 mice chronically exposed to barium in 

drinking water 

Concentration of 

Males 

Females 

Mean 

Mean 

BaCl

2

×2H

2

O 

Dose 

Incidence of 

severity 

Dose 

Incidence of 

severity 

(mg/L) 

(mg/kg-day) 

nephropathy

a 

score 

(mg/kg-day) 

nephropathy

a 

score

 0

0 

1/59 (2%) 

1 

0 

0/60 (0%) 

0 

500 

30 

0/60 (0%) 

0 

40 

2/60 (3%) 

1 

1250 

2500 

75 

160 

2/58 (3%) 

19/60 (32%)

b 

2.5 

3.6 

90 

200 

1/60 (2%)

 37/60 (62%)

b 

2 

3.6 

a

 Incidence rates are expressed for the entire study population (15-month and 2-year); for more information 

see Section 4.2.1. 

b 

Significantly different (p

#0.01) from control group by life table analysis; statistically significant trend for 

entire dat set by Cochran-Armitage trend test (p<0.01). 

Source: NTP, 1994. 

Table 5–3.  Comparison of best-fitting models and benchmark doses for 

increased risk of nephropathy in mice 

Sex 

Best fitting model 

BMD

05

 and BMDL

05

 (mg/kg-day) 

BMD

05 

BMDL

05 

BMD

10 

BMDL

10 

Male 

Multistage 3

o 

84 

63

a 

106 

89 

Female 

Multistage 5

o 

93 

58 

119 

97 

a

 Bolded value was used in deriving the RfD. 

One advantage of BMD modeling is that any point on the dose-response curve, within or 

near the range of the observed data, can be selected as the point of departure.  There is some 

debate in the risk assessment community about the most appropriate benchmark response (BMR) 

for deriving a reference value (U.S. EPA, 2000c).  A 10% BMR (BMR

10

) has historically been 

used as a point of comparison across studies containing quantal data because this is near the limit 

of sensitivity found for most chronic animal studies (U.S. EPA, 2000b).  However, for this 

assessment it was determined that a lower BMR could be used because the critical effect was 

considered to be substantially adverse and distinctly chemical-related and because the data range 

included a response lower than 10%.  First, the lesions in the intermediate dose group (severity 

grades mild to moderate) were intermediate on a continuum leading to severe nephropathy, with 

severity between that seen in the control group (maximum severity grade minimal) and the high 

dose group (severity grades mild to marked).  Since the significantly reduced survival rate in the 

45

---

high dose group was associated with the chemical-related renal lesions (NTP, 1994), the effects 

in the intermediate dose group are considered possibly irreversible and biologically significant. 

Further, a similar pattern of effects was evident in both males and females. 

The BMD for a 5% extra risk of chemical-related nephropathy (BMD

05

) was 84 mg/kg-

day for male mice, and the lower 95% confidence limit (i.e., BMDL

05

) was 63 mg/kg-day.  For 

females the BMD

05

 was 93 mg/kg-day and the BMDL

05

 was 58 mg/kg-day.  These BMDL

05 

values are very similar, but since there is slightly less uncertainty in the estimate derived from 

the male mice (the BMD

05 

and BMDL

05

 are closer together), the male BMDL

05

 was used for 

deriving the RfD. 

5.1.3.  RfD Derivation, Including Application of Uncertainty Factors (UFs) 

Using benchmark dose modeling, the BMDL

05

 of 63 mg/kg-day for 5% extra risk of 

nephropathy in male mice exposed to barium chloride in their drinking water for 2 years (NTP, 

1994) was selected as the point of departure for the  RfD.  To calculate the RfD, a total UF of 

300 was applied to this effect level: 10 for extrapolation for interspecies differences (UF

A

: 

animal to human), 10 for consideration of intraspecies variation (UF

H

: human variability), and 10 

for deficiencies in the data base (UF

D

).  A value of 10 for both the interspecies and intraspecies 

UFs are generally used in the absence of data to indicate otherwise.  The rationale for application 

of the UFs is described below. 

A 10-fold UF was used to account for uncertainty in extrapolating from laboratory 

animals to humans (i.e., interspecies variability).  Insufficient information is available regarding 

the toxicity of chronic barium exposure to compare the dose-response relationship in animals 

with what could be expected in humans.  No information was available to quantitatively assess 

toxicokinetic or toxicodynamic differences between animals and humans.  

A 10-fold UF was used to account for variation in susceptibility among members of the 

human population (i.e., interindividual variability).  This UF was not reduced from a default of 

10 because there are insufficient data on the dose-response relationship in humans and because 

there are studies in experimental animals that suggest gastrointestinal absorption may be higher 

in children than in adults (Taylor et al., 1962; Cuddihy and Griffith, 1972). 

A 3-fold UF was used to account for uncertainty associated with deficiencies in the data 

base.  The data base of oral barium toxicity consists of two human studies that found no effect on 

hypertension (Brenniman et al., 1981; Wones et al., 1990) and several chronic and subchronic 

46

---