Priority Substances List Assessment Report for Chloroform

3.0 Summary of Critical Information and Assessment of "Toxic" under CEPA 1999

3.1 CEPA 1999 64(a): Environment

The environmental risk assessment of a PSL substance is based on the procedures outlined in Environment Canada (1997a). Environmental assessment endpoints (e.g., adverse reproductive effects on sensitive fish species in a community) are selected based on analysis of exposure pathways and subsequent identification of sensitive receptors. For each endpoint, a conservative Estimated Exposure Value (EEV) is selected and an Estimated No-Effects Value (ENEV) is determined by dividing a Critical Toxicity Value (CTV) by an application factor. A conservative (or hyperconservative) quotient (EEV/ENEV) is calculated for each of the assessment endpoints in order to determine whether there is potential ecological risk in Canada. If these quotients are less than one, it can be concluded that the substance poses no significant risk to the environment, and the risk assessment is completed. If, however, the quotient is greater than one for a particular assessment endpoint, then the risk assessment for that endpoint proceeds to an analysis based on more realistic assumptions, and the probability and magnitude of effects are considered. This latter approach involves a more thorough consideration of sources of variability and uncertainty in the risk analysis.

3.1.1 Assessment endpoints

In Canada, nearly all chloroform is released to air, but there are also some direct releases to surface water. Chloroform is also present in groundwater, particularly in the vicinity of landfills. Therefore, assessment endpoints for the environmental assessment of chloroform relate to populations of terrestrial animals living near industrial sources, freshwater pelagic organisms and groundwater-dwelling organisms.

3.1.2 Environmental risk characterization

3.1.2.1 Terrestrial organisms

Since chloroform does not bioaccumulate, biota are exposed via the atmosphere; given that the highest concentrations occur in air in cities, urban wildlife has the greatest potential for exposure to chloroform. Small mammals such as deer mice are likely to have the highest exposure because of their rapid respiration rate and high metabolism. Although no data have been identified for wild animals, data on effects are available for surrogates such as laboratory mammals.

For terrestrial wildlife exposed to chloroform via inhalation, the hyperconservative EEV is 110 µg/m³, the highest atmospheric concentration of chloroform reported in the United States. This value is very conservative, because it is much higher than atmospheric concentrations reported for Canada. Chloroform in the atmosphere can be transported over long distances, but concentrations in Canada from this source would be much less than the EEV because of environmental transformation and dispersion.

The CTV is 9.8 x 10³ µg/m³, the lowest concentration of chloroform reported to cause adverse effects in inhalation toxicity tests with laboratory animals. Dividing this CTV by an application factor of 10 (to account for the extrapolation from laboratory to field conditions and interspecies and intraspecies variations in sensitivity) results in an ENEV of 9.8 x 10² µg/m³.

Table 10 Summary of risk quotients for chloroform for CEPA 1999 64(a)

Environmental compartment	Estimated Exposure Value EEV	Critical Toxicity Value CTV	Application factor AF	Estimated No-Effects Value ENEV	Risk quotient EEV/ENEV
Terrestrial wildlife	110 µg/m3	9.8 x 103 µg/m3	10	9.8 x 102 µg/m3	0.11
Freshwater pelagic biota	44 µg/L	65.7 µg/L	10	6.57 µg/L	6.7
Groundwater biota	13.8 µg/L	500 µg/L	10	50 µg/L	0.28

The hyperconservative quotient is calculated as follows:

Because the hyperconservative quotient is less than one, it is unlikely that chloroform emissions will cause adverse effects on terrestrial wildlife in Canada.

A summary of the values used for the assessment of potential effects of chloroform on terrestrial wildlife is presented in Table 10.

3.1.2.2 Aquatic organisms

3.1.2.2.1 Freshwater pelagic biota

The highest levels observed in Canadian surface waters have in the past been near pulp and paper mills using chlorine bleaching. The maximum concentrations in the Fraser River below the Northwood Pulp and Timber outfall in 1989 and below the Canadian Pacific Forest Products Kraft Mill in Thunder Bay in 1986 were 83 µg/L and 200 µg/L, respectively. Chloroform concentrations in Canadian surface water samples collected after 1989 have been much lower. The maximum reported concentration of chloroform in 984 water samples collected from British Columbia, Alberta, Ontario and Quebec from 1990 to 1996 was 44 µg/L. This value will be used as the EEV.

Based on the available effects data, the most sensitive freshwater pelagic biota are early life stages of spring peepers. The 4-day post-hatching LC₅₀ for the spring peeper was 0.27 mg/L, or 270 µg/L. Environment Canada (1997a) recommends estimating an EC₂₅ or LC₂₅ for the CTV and dividing by a factor of 10 to account for uncertainties arising from laboratory to field extrapolations and interspecies and intraspecies variations in sensitivity. Using an EC₂₅ or LC₂₅ ensures that the toxicity estimates are not model dependent, as is often the case with levels of effect below 5% (e.g., LC₁) (Moore and Caux, 1997). The 4-day post-hatch LC₂₅ for spring peepers was 65.7 µg/L (95% CI = 36.6-106 µg/L). Dividing this value by 10 produces an ENEV of 6.57 µg/L.

The conservative quotient is calculated as follows:

In order to determine the likelihood of chloroform causing harm to populations of freshwater pelagic organisms, it is necessary to examine the exposure and effects data more closely. From the 984 water samples collected from British Columbia, Alberta, Ontario and Quebec from 1990 to 1996, the 99th- and 95th-percentile chloroform concentration values were 2.94 µg/L and <1 µg/L, respectively. The median value was <0.2 µg/L. Only five of the samples contained chloroform concentrations above the ENEV value of 6.57 µg/L: three samples (44, 31.6 and 13 µg/L) were from Quebec, one sample (18 µg/L) was from British Columbia and one sample (7 µg/L) was from Alberta. Chloroform concentrations in Canadian surface water are therefore only rarely above the ENEV.

Table 11 Summary of risk quotients for freshwater pelagic biota

EEV (µg/L)	Descriptor	CTV (µg/L)	Application factor	ENEV (µg/L)	Quotient (EEV/ENEV)
44	Maximum reported conc., post-1990-1996	65.7 (4-day hatch LC25 - spring peepers)	10	6.57	6.7
2.94	99th percentile - all data, 1990-1996	65.7 (4-day post-hatch LC25 - spring peepers)	10	6.57	0.45
<1	95th percentile - all data, 1990-1996	65.7 (4-day post-hatch LC25 - spring peepers)	10	6.57	<0.15
<0.2	Median - all data, 1990-1996	65.7 (4-day post-hatch LC25 - spring peepers)	10	6.57	<0.03
44	Maximum reported conc., 1990-1996	17.7 (4-day post-hatch LC10 - spring peepers)	1	17.7	2.5
2.94	99th percentile - all data, 1990-1996	17.7 (4-day post-hatch LC10 - spring peepers)	1	17.7	0.17
<1	95th percentile - all data, 1990-1996	17.7 (4-day post-hatch LC10 - spring peepers)	1	17.7	<0.06
	Median - all data, 1990-1996	17.7 (4-day post-hatch LC10 - spring peepers)	1	17.7	<0.01

In the toxicity study with spring peepers, the LC₅₀, LC₂₅, LC₁₀ and LC₁ were 270 µg/L, 65.7 µg/L, 17.7 µg/L and 1.9 µg/L, respectively. The LC₁₀ can be used as a good representation of threshold mortality, given that acute toxicity test protocols allow 10% mortality in control treatments. Only 2 of the 984 water samples contained concentrations substantially above the LC₁₀ value, and 1 sample contained chloroform at a concentration almost identical to the LC₁₀ value.

Other amphibians tested along with spring peepers were less sensitive. The LC₁₀ for the second most sensitive amphibian (the leopard frog, Rana pipiens) was 383 µg/L. Other types of aquatic organisms (microorganisms, invertebrates and fish) were less sensitive still.

Based on the available information, concentrations of chloroform in Canadian surface waters are rarely above estimated toxicity thresholds for sensitive aquatic organisms. Chloroform therefore does not appear to pose significant risks to pelagic biota in Canada. A summary of the risk quotients for freshwater pelagic biota is presented in Table 11.

3.1.2.2.2 Groundwater-dwelling biota

No toxicity data were available for groundwater-dwelling biota. The only available toxicity data that could reasonably be extrapolated to effects on groundwater-dwelling biota are from studies on microbial populations used in wastewater treatment. Under anaerobic conditions, however, Yang and Speece (1986) observed inhibition of unacclimated cultures at 500 µg/L. Taking this value, 500 µg/L, as the CTV and dividing by an application factor of 10 produces a hyperconservative ENEV of 50 µg/L. This ENEV is uncertain, because no data were available to estimate effect levels for groundwater-dwelling invertebrates and because of the need to extrapolate from wastewater microbial populations to groundwater-dwelling populations. There are very few data available on the concentration of chloroform in groundwater not associated with the specialized conditions at a landfill site. In what may be regarded as typical of the groundwater conditions independent of the contamination found at landfill sites, 31 groundwater samples collected in British Columbia in 1987 and 1989 were all below the 1 µg/L detection limit (B.C. MOE, 1996).

Furthermore, Carmichael (1996) reported a maximum concentration of 13.8 µg chloroform/L in 16 samples of B.C. groundwater collected in 1992 and 1993. Using 13.8 µg/L as the EEV, a conservative quotient would be calculated as follows:

Therefore, it appears that chloroform poses little risk to groundwater-dwelling biota in Canada at locations that are not in the immediate vicinity of contaminated landfills.

Not surprisingly, the situation at some landfills in Canada is very different from the conditions existing in groundwater in general. These areas have been recognized as contaminated sites and are typically managed or have undergone remediation. They are atypical of the overall conditions that prevail and are therefore not suitable for use in assessing the impact of chloroform or other substances on the environment in general. For example, the maximum chloroform concentration first observed in groundwater at a landfill site in the Ottawa, Ontario, area in 1981 was 53 200 µg/L (Jackson et al., 1985). This site has since undergone extensive remediation, and, in 1988, the highest concentration of chloroform in groundwater from the same sampling site was 97.1 µg/L, while the concentration of chloroform at a sampling site approximately 50 m away was 5.8 µg/L (Moralejo, 1999). The highest concentrations reported at the other two contaminated sites mentioned in Section 2.3.2.5, 950 µg/L in leachates from a chemical company landfill near Sarnia, Ontario (King and Sherbin, 1986), and 916 µg/L in the groundwater at Ville Mercier, Quebec (Pakdel et al., 1992), were the primary figures used to determine the applicability for site remediation. Deriving quotients for these sites would not provide any further help in defining the risk that chloroform poses to the Canadian environment.

A summary of the values used for the assessment of potential effects of chloroform on groundwater-dwelling biota is presented in Table 10.

3.1.2.3 Discussion of uncertainty

There are a number of potential sources of uncertainty in this environmental risk assessment. Direct releases of chloroform from its use by industry are fairly well characterized. The quantity of chloroform released to the Canadian environment from wastewater treatment plants that chlorinate for disinfection is not known. Chloroform releases are highly variable, depending on the flow rate handled at the treatment plants and on the chemical conditions at the plants. Chloroform can be produced in the environment through reactions of chlorine with organic chemicals, and the quantity released from these sources is unknown.

High concentrations of chloroform were reported for surface waters in the vicinity of pulp and paper mills in the 1980s. Since that time, new government regulations have discouraged the use of elemental chlorine by these facilities, and it is believed that the release of chlorinated substances has dropped very significantly. For example, the total discharge of dioxins and furans from pulp and paper mills has fallen by approximately 99%. Concentrations of chloroform in water in the vicinity of pulp and paper mills have also likely decreased considerably, but there are few monitoring data available. According to Environment Canada's EEM database, chloroform was monitored in surface water in the vicinity of four mills in British Columbia. The concentration of chloroform was below the 1 µg/L detection limit in each of the 85 water samples (Environment Canada, 1999b).

Chloroform has been reported at quite high concentrations in leachate from landfills. These leachates could have the potential to contaminate groundwater and/or surface waters in the vicinity, but, again, data are lacking. Remediation work has been undertaken at some of these landfills, and the threat of pollution of groundwater and surface waters has been lowered considerably. Uncertainty also arises from the need to extrapolate effects data from wastewater microbial populations to groundwater-dwelling populations. However, it was shown that wastewater microbial organisms acclimated readily to chloroform and subsequently were able to tolerate concentrations of chloroform up to 15 mg/L.

Few studies have determined the toxicity of chloroform to terrestrial invertebrates, and the studies that do exist are not directly relevant for estimating potentially harmful concentrations in the soil. No information was found on the toxicity of chloroform to birds or wild mammals, but there are data on experimental animals.

3.2 CEPA 1999 64(b): Environment upon which life depends

The net chlorine loadings to the stratosphere from chloroform and its degradation products are small; therefore, chloroform is not considered to be an effective agent of stratospheric ozone depletion. The potential of chloroform to contribute to climate change and ground-level ozone formation is considered to be negligible. The magnitude of these effects would depend on the concentration of chloroform in the atmosphere, and, in Canada, concentrations of chloroform in air are low, usually less than 1 µg/m³.

3.3 CEPA 1999 64(c): Human health

3.3.1 Estimated population exposure

As a basis principally to assess the relative contribution of various media and routes of exposure of the general population in Canada to chloroform, deterministic estimates of exposure were developed for six age groups. These were based on data on concentrations of chloroform in outdoor and indoor air acquired in national surveys in Canada and on estimates of the concentrations in foods in Canada and the United States, assuming age group-specific daily average rates of intake of these media (EHD, 1998). Although national surveys are available, estimates of intake in drinking water were based on monitoring data from the provinces and territories, which included much larger numbers of samples over an extended time frame. These data were also more representative of the water supplies of a larger proportion of the population and lead to more conservative estimates of intake, although they were collected and analysed by less consistent, less reliable and less comparable methodology than for the national surveys. Estimates of the average daily intake of chloroform by inhalation and dermal absorption during showering were also derived for teenagers, adults and seniors (Health Canada, 1999). Probabilistic estimates of exposure for various age groups of the general population were also developed based on distributions of the concentrations of chloroform in outdoor air, indoor air and drinking water in Canada from the same sources that served as the basis for the deterministic estimates. Age group-specific lognormal distributions of daily intake rates for these media were also assumed (EHD, 1998). Data were considered insufficient to develop probabilistic estimates of exposure from ingestion of foods or from showering (Health Canada, 1999).

3.3.1.1 Deterministic estimates of exposure to chloroform for the general population

Point estimates of the average daily intake (per kilogram body weight), based on these data (Section 2.3.2) and on reference values for body weight, inhalation volume and amounts of food and drinking water consumed daily, are presented for six age groups in Table 12. On this basis, average intake was estimated to range from 0.6 to 10.3 µg/kg-bw per day. The upper value in the range of estimated intakes (i.e., 10.3 µg/kg-bw per day) is for infants in the age group of 0-6 months and is based on the assumption that infants are exclusively formula fed during this period, with powdered infant formula reconstituted with tap water containing the maximum annual mean concentration of chloroform (i.e., 89.4 µg/L) as determined from provincial/territorial data. If it is assumed instead that infants are fed table-ready foods containing the same concentrations of chloroform as assumed for the remaining five age groups, the estimated average daily intakes for infants are much lower, ranging from 0.2 to 1.1 µg/kg-bw per day; for the six age groups, the average daily intakes then range from 0.2 to 6.9 µg/kg-bw per day, as indicated in Table 12.

water in Canada and on the maximum reported concentrations in foods in Canada and/or the United States, have also been developed. These are also based on the reference values for body weight, inhalation volume and amounts of food and drinking water consumed daily (EHD, 1998) and are presented for six age groups in Table 13. On this basis, upper bounding estimates of daily intake range from 40 to 95 µg/kg-bw per day. It is assumed that infants are fed table-ready foods only and that their average intake of total tap water is 0.3 L/day (EHD, 1998). If it is assumed instead that infants are exclusively formula fed and that powdered infant formula is prepared with tap water containing the maximum reported concentration in Canada (i.e., 1224 µg/L), the upper bounding estimate of total daily intake for infants is more than twice as high (i.e., 147.6 µg/kg-bw per day, with 130.6 µg/kg-bw per day resulting from ingestion of total tap water).

The contribution of outdoor air to the estimates of average total daily intakes (i.e., Table 12) is considerably less than the contributions from indoor air, food and water, which are approximately similar in magnitude. The contributions of outdoor air and food to the upper bounding estimates of total daily intake (i.e., Table 13) are considerably less than the contributions from indoor air and tap water. On the basis of these deterministic estimates, the main pathways of exposure to chloroform for the general population in Canada are inhalation of indoor air and ingestion of tap water. It is also apparent from these deterministic estimates that the average daily intake from a single daily 10-minute shower can exceed the intake from all other exposure pathways.

3.3.1.2 Probabilistic estimates of exposure to chloroform for the general population

Probabilistic estimates of daily intake of chloroform by six age groups of the general population of Canada were generated in an Excel^TM (Microsoft Corporation, 1997) spreadsheet using Crystal Ball^TM (Decisioneering, Inc., 1996). Age group-specific body weights and rates of intake of air and tap water were assumed to be lognormally distributed and are characterized by their geometric means and standard deviations (EHD, 1998). A normal distribution of hours per day spent outdoors is assumed, characterized by an arithmetic mean and standard deviation of 3.0 ± 2.0 hours (EHD, 1998) and truncated at 0 and 9 hours. The same distribution is assumed for each of the age groups (Health Canada, 1999).

Table 12 Deterministic estimates of average daily intake of chloroform by the general population

Medium of exposure	Average intake (µg/kg-bw per day) by various age groups in the general population
	0-6 months 1	7 months-4 years 2	5-11 years 3	12-19 years 4	20-59 years 5	60+ years 6
Outdoor air 7, 8	0.002-0.034	0.004-0.072	0.003-0.056	0.002-0.032	0.001-0.027	0.001-0.024
Indoor air 9	0.559-0.744	1.197-1.596	0.933-1.244	0.531-0.708	0.456-0.608	0.396-0.528
Food 10	- 11	0.150-1.145	0.105-0.899	0.060-0.612	0.043-0.478	0.028-0.349
Water 12	1.003-9.536	0.424-4.037	0.334-3.172	0.190-1.806	0.199-1.891	0.209-1.987
Subtotal	1.56-10.31	1.78-6.85	1.38-5.37	0.78-3.16	0.70-3.00	0.63-2.89
Inhalation and dermal intake from daily showering 13				0.43-4.06	0.36-3.40	0.35-3.35

Table 13 Upper bounding estimates of daily intake of chloroform by the general population

Medium of exposure	Upper bounding estimates of intake (µg/kg-bw per day) by various age groups in the general population
	0-6 months 1	7 months-4 years 2	5-11 years 3	12-19 years 4	20-59 years 5	60+ years 6
Outdoor air 7, 8	0.21	0.45	0.35	0.20	0.17	0.15
Indoor air 9	16.81	36.02	28.08	15.97	13.72	11.92
Food 10	- 11	2.87	2.36	1.58	1.25	0.89
Water 12	130.6 11	55.28	43.43	24.73	25.90	27.20
Subtotal	147.6	94.62	74.22	42.48	41.04	40.16
Inhalation and dermal intake from daily showering 13				55.64	46.61	45.90

1. ¹Assumed to weigh 7.5 kg, to breathe 2.1 m³ of air per day and to consume 0.3 L of total tap water per day when eating table-ready foods (EHD, 1998).
2. ² Assumed to weigh 15.5 kg, to breathe 9.3 m³ of air per day and to consume 0.7 L of total tap water per day (EHD, 1998).
3. ³Assumed to weigh 31.0 kg, to breathe 14.5 m³ of air per day and to consume 1.1 L of total tap water per day (EHD, 1998).
4. ⁴Assumed to weigh 59.4 kg, to breathe 15.8 m³ of air per day and to consume 1.2 L of total tap water per day (EHD, 1998).
5. ⁵Assumed to weigh 70.9 kg, to breathe 16.2 m³ of air per day and to consume 1.5 L of total tap water per day (EHD, 1998).
6. ⁶Assumed to weigh 72.0 kg, to breathe 14.3 m³ of air per day and to consume 1.6 L of total tap water per day (EHD, 1998).
7. ⁷Based on the assumption that all age groups spend 3 hours outdoors per 24-hour day (EHD, 1998).
8. ⁸Based on the maximum 24-hour average concentration measured in the NAPS data set (i.e., 5.99 µg/m³ at NAPS site 100127 in 1992). These data are from Dann (1998).
9. ⁹Based on the maximum 24-hour average concentration of chloroform in indoor air (i.e., 68.6 µg/m³) from among 754 Canadian homes in nine provinces (Concord Environmental Corporation, 1992).
10. ¹⁰Estimates of intakes of chloroform from ingestion of foods are based on per capita arithmetic mean daily consumption rates (grams/day) from EHD (1998) and the maximum concentrations of chloroform in 131 food items measured in Canada and/or the United States. In the remaining food items, the concentration of chloroform is assumed to be zero. Infants are assumed to eat table-ready foods only, containing the maximum concentration of chloroform, as for the remaining six age groups.
11. ¹¹It is assumed that infants are exclusively formula fed; the estimate of the total daily intake by infants from ingestion of food and water is 130.6 µg/kg-bw per day. If infants are assumed to be fed table-ready foods, the estimate of the average daily intake of total tap water is 0.3 L/day. Based on this assumption, the total daily intake by infants from ingestion of food and water is 51.2 µg/kg-bw per day. If infants are assumed to be exclusively breastfed a maximum of 1.033 L/day (EHD, 1998) of breast milk containing 65 µg chloroform/L (Erickson et al., 1980), the estimate of total daily intake by infants from ingestion of food and water is 8.95 µg/kg-bw per day.
12. ¹²Estimates of intakes of chloroform from ingestion of drinking water are based on average daily consumption rates of "total tap water" from EHD (1998) and on the maximum concentration of chloroform in drinking water in Canada (i.e., 1224 µg/L, in Alberta) among provincial/territorial data. Total tap water includes water used in the preparation of beverages.
13. ¹³Based on the assumption from Benoit et al. (1998) that the combined intake of chloroform from the inhalation and dermal routes during a 10-minute shower is equivalent on an annual average to the intake from ingestion of 2.7 L of cold tap water from the same source. It is assumed that this tap water contains the maximum reported concentration of chloroform in drinking water in Canada (i.e., 1224 µg/L, as in footnote 12).

Two scenarios were developed for estimating daily intakes from exposure to chloroform in outdoor and indoor air and tap water. In a scenario for general population exposure, the following distributions of concentrations were assumed. For outdoor air, this was based on the distribution of chloroform in the air of 8807 samples collected during the 1990s in the NAPS program (Dann, 1998). For indoor air, it was based on the estimated geometric mean and standard deviation of an assumed lognormal distribution of chloroform in the indoor air of 754 Canadian homes (Concord Environmental Corporation, 1992; Health Canada, 1999). For tap water, the distribution of chloroform in the treated drinking water of 6607 samples, based on provincial/territorial data, was assumed.

In an RWC exposure scenario, the following distributions of concentrations were assumed. For outdoor air, the distribution of chloroform in air was that in 800 samples collected during the 1990s from four sites adjacent to major roadways in the NAPS program (Dann, 1998). For indoor air, this was again based on the estimated geometric mean and standard deviation of an assumed lognormal distribution of the concentrations of chloroform in the indoor air of 754 Canadian homes (Concord Environmental Corporation, 1992; Health Canada, 1999), since these data were inadequate as a basis to define a subset of concentrations for use in the RWC scenario. For tap water, the distribution of chloroform in the treated drinking water of 2597 samples, based on data from Manitoba and Alberta only where reported concentrations were highest, was assumed.

Simulations of 10 000 trials were run five times each using two sampling methods (i.e., Monte Carlo random and Latin Hypercube) to gauge the reproducibility of the parameters estimated. For the average population scenario, the 95th percentiles of the distribution of intakes from inhalation and ingestion of drinking water for five age groups of the general population (i.e., 0.5 years to 60+ years of age) range from 4.9 to 12.9 µg/kg-bw per day (Health Canada, 1999). Similar estimates were obtained from each of the two sampling methods. The relative standard deviations (for n = 5 simulations of 10 000 trials each) of the upper-percentile estimates of intake did not exceed 5%, indicating a high degree of reproducibility.

inhalation and ingestion of drinking water for five age groups of the general population (i.e., 0.5 years to 60+ years of age) range from 7.0 to 19.1 µg/kg-bw per day (Health Canada, 1999). Similar estimates were obtained from each of the two sampling methods. The relative standard deviations (for n = 5 simulations of 10 000 trials each) of the upper-percentile estimates of intake did not exceed 7%, indicating a high degree of reproducibility.

For both the population exposure and RWC scenario, due to limitations of the data concerning the daily intake rate of total tap water by infants (EHD, 1998), probabilistic estimates could not be developed for the sixth age group (i.e., birth to 0.5 years).

3.3.2 Hazard characterization

As indicated in Section 2.4.4, available data on the toxicity of chloroform to humans contribute to our understanding of its toxicity to the extent that the target organ in populations exposed occupationally to high concentrations is similar to that in experimental animals (i.e., the liver).

The weight of evidence indicates that chloroform may be carcinogenic only at concentrations that induce the obligatory precursor lesions of sustained cytotoxicity and persistent proliferative regenerative response. For consistency with other assessments and for ease of presentation, cancer and non-cancer effects are considered separately here, although it is recognized that, based on consideration of mode of action, they are inextricably linked.

3.3.2.1 Cancer

As described in Section 2.4.3.7, there is considerable information available concerning the potential mode of induction of liver and kidney tumours by chloroform. This includes a range of metabolic studies. In addition, while there have been no cancer bioassays in which a range of intermediate endpoints has been investigated, proliferative response in target organs has been examined in numerous subsequent investigations For the RWC scenario, the 95th percentiles of the distribution of intakes from following exposure via regimens similar to those in the long-term studies. The histopathology in the target organ for one of the more critical studies has also been reexamined (Hard et al., in press). These data have been generated to investigate primarily the hypothesized mode of action for tumour induction in rodents that requisite precursor steps to cancer are 1) metabolism of chloroform by the target cell population, 2) induction of sustained cytotoxicity by metabolites and 3) subsequent persistent regenerative cell proliferation.

Metabolism to phosgene, resulting from the oxidative pathway that predominates at low exposures, is believed to be the principal determinant of sustained toxicity and resulting persistent proliferation that is hypothesized to lead to a higher probability of spontaneous cell mutation and subsequent cancer. Measures of cytotoxicity include histopathological effects and release of hepatic enzymes and labelling indices as surrogates for regenerative cell proliferation.

Chloroform causes liver and kidney tumours in mice and kidney tumours in rats. Although the hypothesized modes of induction of these tumours are similar, the weight of evidence varies considerably, and, as a result, they are addressed separately here.

Liver tumours are observed in B6C3F1 mice following administration of bolus doses by gavage in corn oil (NCI, 1976), but not following administration of the same daily doses in drinking water (Jorgenson et al., 1985). That dose rate is a critical determinant of tissue damage (e.g., being greater following bolus dosing by gavage compared with continuous administration) is consistent with the proposed mode of induction of tumours. Doses at which tumours have been observed following administration in corn oil in the cancer bioassay are associated in shorter-term studies with sustained proliferative response in the liver of the same strain exposed similarly (Larson et al., 1994c; Pereira, 1994; Melnick et al., 1998). Sustained increases in proliferative response have not been observed following ingestion in drinking water of concentrations that did not induce increases in hepatic tumour incidence in the long-term bioassay (Larson et al., 1994a).

The incidence and severity of hepatic necrosis in the mouse liver have been related to the degree of covalent binding of chloroform metabolites to tissue proteins. The linking of metabolism to toxicity is underscored by localization of covalent binding to the necrotic lesions and the predictable variations in toxic response produced by pretreatment with inducers or inhibitors of cytochrome P450-mediated metabolism, specifically CYP2E1. There is strong recent evidence that it is the oxidative metabolites specifically that predominate at low concentration and cause cytotoxicity in the mouse liver. This includes observation of a direct correlation between binding to the polar heads of phospholipid molecules (caused by oxidative metabolites) and protein binding in the liver of the strain of mice in which tumours have been observed (Ade et al., 1994). Particularly strong evidence of the role of CYP2E1 in the induction of mouse liver tumours is also provided by recent studies in CYP2E1 null mice. There was no cytotoxicity or cell proliferation in the liver of two strains of CYP2E1 null mice (Sv/129 and B6C3F1 strains) at a concentration that caused severe hepatic lesions in the wild type of either strain (Constan et al., 1999). There is a consistent association between CYP2E1 distribution, chloroform metabolism, pattern of covalent tissue binding and toxic injury to hepatocytes in mice.

Evidence of concordance between metabolism to reactive intermediates, cytotoxicity, regenerative proliferation and tumour development in the mouse liver is, therefore, very strong. Indeed, there is a wealth of information that indicates a relationship between sustained enhanced proliferative response and induction of liver neoplasia in the strain in which tumours have been observed (B6C3F1 mice).

Chloroform also induces renal tumours in BDF1 mice following inhalation (Yamamoto, 1996) and in ICI mice exposed by gavage in toothpaste (Roe et al., 1979), although at lower rates than liver tumours. The response is strain and sex specific, occurring only in males.

Evidence of concordance between metabolism to reactive intermediates, cytotoxicity, regenerative proliferation and tumour development in the mouse kidney, although strong, is not as robust as for the mouse liver, due primarily to the more limited data available on sustained enhanced proliferative response in the strains in which tumours have been observed. Indeed, this is limited to a single study in BDF1 mice, in which there was an increase in labelling index in the kidneys of males but not females at concentrations that induced renal tumours in this strain in the long-term inhalation bioassay (Templin et al., 1996c; Yamamoto, 1996). The available data concerning the relationship between sustained cellular proliferation and induction of renal tumours in another strain (B6C3F1) of mice indicate that sustained proliferative response is not always associated with tumours. In this strain, in shorter-term studies, there were sustained proliferative responses at doses at which kidney tumours were not observed in the relevant cancer bioassays following exposure by both gavage in corn oil and drinking water (NCI, 1976; Jorgenson et al., 1985; Larson et al., 1994a,c).

In mice, covalent binding of chloroform to renal proteins and microsomes is correlated with the degree of renal tubular necrosis, with strain and sex differences in sensitivity to nephrotoxicity being correlated with the ability of the kidney to metabolize chloroform. Similar to the liver, there is strong recent evidence that it is the oxidative metabolites specifically that predominate at low concentration and cause cytotoxicity in the mouse kidney. This includes observation of a direct correlation between binding to the polar heads of phospholipid molecules (caused by oxidative metabolites) and protein binding in the kidney of DBA/2J mice (Ade et al., 1994). Particularly strong evidence of the role of CYP2E1 in the induction of mouse renal tumours is also provided by recent studies in CYP2E1 null mice. There was no cytotoxicity or cell proliferation in the kidney of two strains of CYP2E1 null mice (Sv/129 and B6C3F1 strains) at a concentration that caused severe hepatic lesions in the wild type of either strain (Constan et al., 1999).

The weight of evidence for the hypothesized mode of induction of tumours in the rat kidney is considerably less than that for the mouse liver and kidney due primarily to limited data on intermediate endpoints in the only strain (Osborne-Mendel) in which increases in kidney tumours have been observed. These increases have been reported following exposure via both gavage in corn oil and drinking water (NCI, 1976; Jorgenson et al., 1985). There are also few identified data on the relationship between the metabolism of chloroform and induction of renal lesions in rats. In the F344 rat, there were sustained increases in proliferative response in shorter-term studies following administration of doses similar to those that induced tumours in Osborne-Mendel rats following administration by gavage in corn oil but not following ingestion in drinking water (Larson et al., 1995a,b). However, there are no bioassays in this strain following ingestion for direct comparison with these results. Sustained increases in labelling index were observed in the proximal tubules of F344 rats exposed to daily doses of 30 ppm (147 mg/m³) and greater and at 90 ppm (441 mg/m³) and greater at 5 days per week (Templin et al., 1996b). However, increases in kidney tumour incidence were not observed in this strain exposed to up to 90 ppm (441 mg/m³) for 5 days per week in the only inhalation cancer bioassay (Yamamoto, 1996).

Based on studies conducted primarily in F344 rats in which tumours have not been observed, a mode of action for carcinogenicity in the kidney observed in the carcinogenesis bioassay in Osborne-Mendel rats based on cytotoxicity and tubular cell regeneration is, therefore, plausible. For Osborne-Mendel rats, the results of reanalyses of the original renal tissues (Hard and Wolf, 1999; Hard et al., in press), from both the drinking water bioassay (Jorgenson et al., 1985) and the gavage study (NCI, 1976), have been critical. They provide strong support for the contention that the mode of induction of these tumours is consistent with the hypothesis that sustained proximal tubular cell damage is a requisite precursor lesion for chloroform-induced tumours.

In all cases where examined, therefore, sustained cytotoxicity and cellular proliferation were observed in the liver and kidney of the same strain of mice and rats exposed in a similar manner in short-term studies to concentrations or doses that induced tumours in these organs in cancer bioassays. However, the converse is not always true. Tumours have sometimes not been observed in cases where there have been sustained increases in damage and resulting proliferation in the same strain exposed to similar concentrations in the same manner in shorter-term studies, namely kidney lesions in B6C3F1 mice and F344 rats. These results are consistent with the hypothesis that, where chloroform causes tumours, toxicity and reparative hyperplasia are obligatory precursor steps. Tumours would not necessarily be expected whenever there is an increase in cell replication. The multiple susceptibility factors that produce tumours following cytotoxicity will depend on tissue-specific factors and will likely vary between species and strains. For example, in spite of the overt toxicity and sustained increased cell proliferation in the epithelial tissue of the nose in both rats and mice, no tumours have been noted in this tissue in any chronic studies, including the inhalation bioassay in which nasal tissues were carefully evaluated (Yamamoto, 1996).

The organs in which chloroform-induced cytotoxicity and proliferative lesions are observed (liver, kidney and nasal passages) correlate well with the distribution of CYP2E1 both across and within species (Löfberg and Tjälve, 1986). This consistent pattern of response to chloroform across species and organs supports a conclusion that chloroform-induced neoplasia is dependent on cytotoxicity coupled with regenerative cell proliferation. This is further supported by the considerable weight of evidence indicating that chloroform is not genotoxic, with unconvincing evidence for direct DNA reactivity. Due principally to limitations of the available data, though, weak genotoxicity in the rat cannot be precluded, which detracts somewhat from the weight of evidence in this species, although it is unknown whether this might be a result of secondary effects on DNA.

The hypothesized mode of carcinogenesis for chloroform is in keeping with the growing body of evidence supporting the biological plausibility that prolonged regenerative cell proliferation can be a causal mechanism in chemical carcinogenesis. This has been addressed in numerous articles, including Ames and Gold (1990, 1996), Cohen and Ellwein (1990, 1991, 1996), Preston-Martin et al. (1990), Ames et al. (1993), Tomatis (1993), Cohen (1995), Cunningham and Matthews (1995), Butterworth (1996), Farber (1996) and Stemmermann et al. (1996). Enhanced cell proliferation can lead to an increased frequency of spontaneous genetic damage either through errors that result from the infidelity of DNA replication or through the increased conversion of endogenous DNA changes into heritable genetic changes (Cohen and Ellwein, 1990, 1991, 1996; Ames et al., 1993; Cohen, 1995). Additionally, during periods of cell replication, heritable non-mutagenic modifications of the genome may occur that may lead to changes in gene expression, contributing to carcinogenesis (U.S. EPA, 1996b). This view that cell proliferation is a risk factor for carcinogenesis is not universally accepted, because strict correspondence between increased cell turnover and carcinogenic response is not always demonstrable (Melnick, 1992; Farber, 1996). However, as indicated above, in view of the complex interplay of factors involved in the carcinogenesis process, it is not surprising that acute measures of cell proliferation do not always indicate a one-to-one correlation. Among the factors to be considered are the kinetics of DNA adduct formation and repair, the balance between cell proliferation, differentiation and death, proliferation in the target cell compartments compared with that of non-target cells and the consequences of overt tissue toxicity. While the evidence is fairly convincing that chloroform is active principally through cytotoxic effects of phosgene and other products of oxidation, several other possibilities in which mutagenicity might play a role were also considered. One is that the effects of chloroform are a composite of those of metabolites from both oxidative and reductive pathways contributing to toxicity and carcinogenicity. However, several observations strongly support the predominant role of oxidative pathways in chloroform toxicity and make any significant role of reductive metabolism highly unlikely. Firstly, the macromolecular binding following administration of chloroform represents only a very small portion of the delivered dose. Secondly, the mechanisms of action related to the nature of the necrotic lesion, the time course of injury after single doses and the differences in cumulative damage on multiple exposures are very different for chloroform and carbon tetrachloride, the latter a compound for which the free radical (reductive) pathway is causative for toxicity. In addition, carbon tetrachloride, which is largely metabolized to a free radical, is not itself mutagenic. Based on these considerations, it was concluded that free radicals do not play a significant role in the toxicity or carcinogenicity of chloroform.

Another possibility is that minor pathways, associated with glutathione conjugation, produce mutagenic metabolites, as is believed to be the case for dichloromethane. However, there is little evidence for a significant direct conjugation pathway for chloroform. In studies with Salmonella tester strains with glutathione transferase T1-1 inserted into the bacterial genome and expressed during testing, a small increase in mutagenic activity (less than a factor of 2) was noted for chloroform at very high doses, even though positive controls with methylene chloride and bromochloromethane produced much larger responses (Pegram et al., 1997). Neither of these other two potential modes of action is believed to play a significant role in the observed toxicity and carcinogenicity of chloroform, although further investigation of weak genotoxicity in the rat is desirable.

In summary, then, chloroform has induced liver tumours in mice and renal tumours in mice and rats. The weight of evidence of genotoxicity, sex and strain specificity and concordance of cytotoxicity, regenerative proliferation and tumours are consistent with the hypothesis that marked cytotoxicity concomitant with a period of sustained cell proliferation likely represent a secondary mechanism for the induction of tumours following exposure to chloroform. This is consistent with a non-linear dose-response relationship for induction of tumours. This cytotoxicity is primarily related to rates of oxidation of chloroform to reactive intermediates, principally phosgene and hydrochloric acid. The weight of evidence for this mode of action is strongest for hepatic and renal tumours in mice and more limited for renal tumours in rats.

3.3.2.2 Non-neoplastic effects

Effects observed most consistently at lowest concentrations or doses following repeated exposures to chloroform in rats and mice are cytotoxicity and regenerative proliferation. As discussed in relation to cancer, target organs are the liver (centrilobular region) and kidney (cortical region). In addition, chloroform has induced nasal lesions in rats and mice exposed by both inhalation and ingestion at lowest concentrations or doses.

Effects on the hematological, neurological and immunological systems have been reported less consistently and only at concentrations higher than those reported to induce effects on the liver, kidney and nose. Teratogenic effects have not been reported. Developmental/reproductive effects have been restricted to those observed most often at dose levels that caused other manifestations of systemic toxicity in the same studies, primarily hepatic effects. Such effects have also only been observed at doses greater than the lowest values reported in other studies to induce effects on the liver, kidney or nose.

3.3.3 Exposure-response analyses

While the target organ in populations exposed occupationally to high concentrations of chloroform is similar to that in experimental animals (i.e., the liver), the levels at which effects occur (i.e., dysfunction and necrosis) are not well documented and are inadequate as a basis to meaningfully characterize exposure-response.

3.3.3.1 Cancer

Available data are consistent with a mode of action for the carcinogenicity of chloroform that is a secondary consequence of cytotoxicity and associated reparative cell proliferation induced by oxidative metabolites. Hence, where chloroform causes tumours, oxidative metabolism, sustained cytotoxicity and persistent reparative hyperplasia are considered obligatory precursor steps. Based on this mode of action, the optimum approach to quantitation of exposure-response for comparison with estimated human exposure to draw conclusions concerning "toxic" as defined in Paragraph 64(c) of CEPA 1999 would be as follows. Data on non-cancer precursor events (cytotoxicity and regenerative proliferation) from interim kills in the critical cancer bioassay could be analysed on the basis of rates or amounts of oxidative metabolites produced per volume of tissue in the critical organ.

Liver tumours in mice (males, females) have been induced only by administration of bolus doses in corn oil (NCI, 1976) or at lethal concentrations (male and female mice) following inhalation (Yamamoto, 1996). Kidney tumours have been reported in male mice following ingestion in a toothpaste vehicle (Roe et al., 1979) or inhalation (Yamamoto, 1996), but at concentrations in the latter that cause severe kidney necrosis and acute lethality. Renal tumours in rats have been observed, however, in an adequate and relevant study in which the route and pattern of exposure were similar to those of humans (i.e., continuously in drinking water) (Jorgenson et al., 1985).

The critical carcinogenesis bioassay for quantitation of exposure-response for this assessment is, therefore, that of Jorgenson et al. (1985). Unfortunately, there were no data collected in this bioassay that might serve as the basis for quantitation of exposure-response for precursor lesions such as cytotoxicity or regenerative hyperplasia. A proportion of the slides from several dose groups was recently reexamined, however (Hard and Wolf, 1999; Hard et al., in press) (Section 2.4.3.4). While this reexamination confirmed histopathological changes consistent with the hypothesis that sustained tubular cytotoxicity and regenerative hyperplasia led to renal tubular tumour induction, data amenable for quantitation of exposure-response in this investigation were limited but permit at least crude comparison. ⁵

There have been numerous subsequent shorter-term investigations of the proliferative response in the liver and kidney of various strains of mice and rats exposed to doses and concentrations of chloroform similar to those administered in the cancer bioassays in which tumours have been observed (Section 2.4.3.7.3). However, for renal tumours, most of these investigations have been conducted in the F344 rather than the Osborne-Mendel rat in which increases in renal tumours have been observed.

Limited available data indicate that the proliferative response in the F344 rat is not an appropriate surrogate for characterization of exposure-response for an intermediate endpoint for renal tumours in the Osborne-Mendel rat. For example, there is no indication of sex-specific variation in the proliferative response in the kidney of F344 rats (Larson et al., 1995a,b), although the increase in renal tumours in Osborne-Mendel rats is sex specific (i.e., restricted to males). In addition, in metabolic studies in F344 rats, intrarenal activation by cytochrome P450 was not implicated as a determinant of nephrotoxicity (Smith et al., 1985). Available data are also inadequate as a basis of characterization of the relative sensitivity of the two strains to cytotoxicity. In the single study in which proliferative response was examined in Osborne-Mendel rats (Templin et al., 1996a), it was concluded that they were about as susceptible as F344 rats to chloroform-induced renal injury, based on comparison 2 days following a single gavage administration.

However, a statistically significant increase in labelling index was observed at a much lower dose (10 mg/kg-bw) in the Osborne-Mendel than in the F344 rat (90 mg/kg-bw). This latter observation may have been a function of the low value in controls for the Osborne-Mendel rats, attesting to the fact that these data are inadequate in themselves to characterize variations in sensitivity of the two strains. Rather, the results of this study contribute inasmuch as they are not inconsistent with a mode of action of induction of tumours involving tubular cell regeneration in Osborne-Mendel rats.

Since quantitative data on the incidence of precursor lesions for cancer in the strain of interest are inadequate to meaningfully characterize exposure-response, a tumorigenic concentration has been developed for this purpose, based on the incidence of tubular cell adenomas and adenocarcinomas in the bioassay of Jorgenson et al. (1985) (Section 2.4.3.4).

In view of the weight of evidence for the role of oxidative metabolites in induction of requisite damage and resulting tumours, dose-response for cancer for chloroform is optimally expressed in terms of amounts or rates of formation of reactive metabolites in the target tissue. These rates have been estimated pharmacokinetically based on models that include specific parameters related to metabolic rates, enzyme affinities and enzyme tissue distribution (Section 2.4.3.7.2).

Characterization of exposure-response for cancer associated with exposure to chloroform in the context of rates of formation of reactive metabolites in the target tissue is considered appropriate in view of the sufficiency of the evidence to support the following assumptions inherent in the PBPK modelling:

1. In both experimental animals and humans, metabolism of chloroform by CYP2E1 is responsible for production of the critical reactive metabolite, phosgene.
2. The ability to generate phosgene and phosgene hydrolysis products determines which tissue regions in the liver and kidney are sensitive to the cytotoxicity of chloroform.
3. This dose-effect relationship is consistent within a tissue, across gender and across route of administration, and it may also be consistent across species.

Although several PBPK models in animals have been developed previously for chloroform, a human component was not available. Results presented here are from the "hybrid" animal model of the ILSI Expert Panel (ILSI, 1997), which was revised for this assessment and developed to permit its extension to humans (Section 2.4.3.7.2) (ICF Kaiser, 1999).

Various dose metrics have been considered in exposure-response analyses for chloroform. ILSI (1997) investigated four dose metrics in their "hybrid" animal model (Section 2.4.3.7.2) in relation to the labelling indices (assumed to be representative of response for cytotoxicity, the intermediate endpoint in induction of cancer) in the liver and kidney of exposed F344 rats. As would be expected based on the hypothesized mode of action, the fit for two of these - namely, the total amount of phosgene produced and the maximum concentration of chloroform reached in each experimental dosing interval with proliferative response - was poor. Of the other two, the mean and maximum rates of phosgene production during each experimental dosing interval, the fit with the labelling indices was best for maximum rate (VRAMCOR) (ILSI, 1997). For the current assessment, maximum rate of metabolism per unit kidney cortex volume (VRAMCOR) and mean rate of metabolism per unit kidney cortex volume during each dose interval (VMRATEK) were considered.

Although similar, the fit of the data on tumour incidence for VRAMCOR (p = 0.97) was slightly better than that for VMRATEK (p = 0.84). However, human equivalent concentrations for the former could be developed only for the lower 95% confidence limit of the Tumorigenic Concentration₀₁ (TC₀₁), since the maximum rate of human metabolism in the kidney is less than that in the rat. The maximum rate of metabolism that can be achieved in the human kidney, based on metabolic parameters included in the model (approximately 8.1 mg/L per hour), was between the animal dose metrics associated with the Benchmark Concentration₀₁ (BMC₀₁) and the lower 95% confidence limit of the BMC₀₅.

The results of the exposure-response assessment presented here are, therefore, those for the combined incidence of renal adenomas and adenocarcinomas in Jorgenson et al. (1985) versus VMRATEK,⁶ fit to the following model (Howe, 1995):

where d is dose, k is the number of dose groups in the study, P(d) is the probability of the animal developing the effect at dose d and qi > 0, i = 1,..., k is a parameter to be estimated. The model was fit to the incidence data using THRESH (Howe, 1995), and the Benchmark Dose₀₅s (BMD₀₅s) were calculated as the concentration D that satisfies:

Figure 1 Tumorigenic tissue dose (humans) for combined incidence of renal adenomas and adenocarcinomas in Osborne-Mendel rats (Jorgenson et al., 1985)

Results of the model fitting are presented in Figure 1. The relevant measure of exposure-response, i.e., the mean rate of metabolism (VMRATEK) in humans associated with a 5% increase in tumour risk (TC₀₅) estimated on the basis of the PBPK model, is 3.9 mg/L per hour (95% lower confidence limit, 2.5, chi-square = 0.04, degrees of freedom = 1, P-value = 0.84). This dose rate would result from continuous lifetime exposure to 3247 mg/L in water or 30 ppm (147 mg/m³) chloroform in air. Respective lower 95% confidence limits for these values are 2363 mg/L and 15 ppm (74 mg/m³).

Although data on dose-response were less robust than those for the cancer bioassay, for comparison, a benchmark dose was developed for histological lesions in the kidney in the reanalysis of a subset of the slides from the Jorgenson et al. (1985) bioassay. Results of the model fitting are presented in Figure 1. The mean rate of metabolism (VMRATEK) in humans associated with a 5% increase in histological lesions characteristic of cytotoxicity is 1.7 mg/L per hour (95% lower confidence limit, 1.4, chi-square = 3.9, degrees of freedom = 2, P-value = 0.14). This dose rate would result from continuous lifetime exposure to 1477 mg/L in water or 6.8 ppm (33.3 mg/m³) in air. These values are approximately 2-fold less than those presented above, based on the more robust data on tumour incidence.

3.3.3.2 Non-neoplastic effects

The results of repeated-dose toxicity studies in which effects were observed at the lowest concentrations are summarized by manner of administration in Tables 14, 15 and 16 for bolus dosing by gavage, continuous administration in drinking water and inhalation, respectively. For ease of comparison, in addition to expression as concentrations in the administered medium (for continuous administration by drinking water and inhalation), effect levels have also been converted to mg/kg-bw, based on assumed volumes for inhalation and ingestion of drinking water and body weights (Health Canada, 1994), with the exception of those studies in which effects were observed at site of contact (i.e., nasal lesions following inhalation).

Following exposure by inhalation, effects at the site of contact are limiting, with proliferation in the nasal passages being reported at concentrations as low as 2 ppm (9.8 mg/m³) in both rats and mice for 6 or 7 hours a day for periods ranging from 4 to 7 days (Larson et al., 1996; Templin et al., 1996b). At 5 ppm (25 mg/m³), ossification of the nasal septum was observed in BDF1 mice exposed for 5 days per week for 2 years (Yamamoto, 1996). At 10 ppm (49 mg/m³), cell proliferation and histopathological lesions were reported in the nasal passages of rats exposed for 6 hours per day for 1-3 days and mice exposed for 6 hours per day for 4-7 days (Mery et al., 1994; Templin et al., 1996b); ossification of the nasal turbinates was reported in rats exposed to this concentration for 5 days per week for 2 years (Yamamoto, 1996). In one study (Larson et al., 1994c), moderate hepatic changes were observed in mice exposed to 10 ppm (49 mg/m³) for 6 hours per day for 7 days. At concentrations of 25-30 ppm (123-147 mg/m³), effects on the kidney and liver in rats and mice, including increases in organ weights, histopathological lesions and increases in proliferation, are observed following exposure for periods ranging from 4 days to 6 months (Table 16).

Following administration in drinking water, renal effects were reported at the lowest doses in rats and mice, with hepatic effects observed at higher doses. Regenerative proliferation was observed following 3 weeks' exposure to 17 and 40 mg/kg-bw per day in rats and mice, respectively (200 mg/L in drinking water) (Larson et al., 1994a, 1995b). Histological alterations in the liver of F344 rats were reported at 58 mg/kg-bw per day after 4 days' exposure (Larson et al., 1995b) (Table 15).

In protocols with bolus administration, the weight of the liver was affected in rats at the lowest dose following gavage in corn oil for 4 days (10 mg/kg-bw per day), while at higher doses (34 mg/kg-bw per day), there were histological changes in the liver (Larson et al., 1995a,b). At 15 mg/kg-bw per day, fatty cysts in the liver were observed in dogs exposed to chloroform in toothpaste base in gelatin capsules 6 days per week for 7.5 years (Heywood et al., 1979). At 34 mg/kg-bw per day, effects upon kidney and liver were reported in mice (Larson et al., 1994c); proliferation and lesions in the olfactory epithelium were observed at this dose in rats (Table 14).

In summary, then, short-term exposure by inhalation resulted in cellular proliferation in nasal passages in rats and mice at concentrations as low as 2 ppm (9.8 mg/m³), with ossification being observed at slightly higher concentrations following long-term exposure. In short-term studies, moderate hepatic changes were observed in mice at 10 ppm (49 mg/m³); following both short- and long-term exposure to 25-30 ppm (123-147 mg/m³), there were multiple adverse effects in the kidney and liver in both rats and mice in several studies. Following ingestion in drinking water, regenerative proliferation following short-term exposure of mice to doses as low as 17 mg/kg-bw has been observed. Following bolus dosing, increases in proliferation in the liver of rats have been observed following short-term exposure of rats at 10 mg/kg-bw per day and fatty cysts in the liver of dogs at 15 mg/kg-bw per day.

For oral exposure, therefore, lowest reported effect levels in various species for different endpoints are similar and occur following bolus dosing. One of the lowest dose levels at which effects on liver and kidney have been observed is that in dogs reported by Heywood et al. (1979). As a result, a PBPK model in dogs was developed for this assessment, since characterization of exposure-response for ingestion on the basis of this study is likely to be protective, although it should be considered in the context of an example, in view of the fact that effects on the liver of rodents have also been observed in a similar dose range.

Two dose metrics were investigated in exposure-response: the mean rate of metabolism per unit centrilobular region of the liver (VMRATEL) and the average concentration of chloroform in the non-metabolizing centrilobular region of the liver (AVCL2). The two dose metrics were selected in order to evaluate the possibility of the fatty cyst formation in the dogs being the result of the solvent effects of chloroform or effects of a reactive metabolite.

Figure 2 Benchmark tissue dose (humans) for incidence of hepatic fatty cysts in dogs (Heywood et al., 1979)

The incidence of fatty cysts in this study (Table 4) versus VMRATEL and AVCL2 was fit to the model in the manner described for the assessment of exposure-response for cancer described above. Results of the model fitting are presented in Figure 2. The fit of the data on the incidence of fatty cysts was better for VMRATEL (p = 1) than for AVCL2 (p = 0.45). Hence, fit supported the assumption that a metabolite rather than chloroform itself was responsible for the observed effects. The mean rate of metabolism per unit centrilobular region of the liver (VMRATEL) in humans associated with a 5% increase in fatty cysts estimated on the basis of the PBPK model is 3.8 mg/L per hour (95% lower confidence limit = 1.3, chi-square = 0.00, degrees of freedom = 1, P-value = 1.00). This dose rate would result from continuous lifetime exposure to 37 mg/L in water or 2 ppm (9.8 mg/m³) in air. Respective lower 95% confidence limits for these values were 12 mg/L and 0.7 ppm (3.4 mg/m³).

3.3.4 Human health risk characterization

The exposure of Canadians was compared with the tissue dose measures described above through modelling of tissue doses resulting from a 24-hour exposure scenario. This scenario included inhalation, ingestion and dermal absorption from one 10-minute shower, a brief washing-up period before retiring, discrete periods of food and water consumption and inhalation of chloroform at various concentrations. The scenarios were based on midpoint and 95th percentiles of concentrations in outdoor air (background and commuting), indoor air, air in the shower compartment, air in the bathroom after showering, tap water and food (Table 17). The greatest single contributor to chloroform exposure within the 24-hour period results from inhalation during showering, which also includes dermal absorption. The human model was run with concentrations and durations in the multimedia scenario presented in Tables 18 and 19. This resulted in an estimated tissue dose that was 1794 (lower 95% confidence limit, 570) times less than that associated with the TC for cancer. For non-cancer, the comparable margin for the BMD05 was 591 (lower 95% confidence limit, 165).

Table 14 Effect levels in laboratory animals exposed to chloroform by bolus administration (presentation limited to those studies in which the lowest effect levels were reported)

Effect level	Endpoint	Species/sex	Protocol	Reference
10 mg/kg-bw per day, LOAEL	Significant increase in liver weight	Male F344 rat	Gavage, corn oil, 4 days	Larson et al. (1995b)
3 mg/kg-bw per day, NOEL	At higher doses: histopathological changes in liver at 34 mg/kg-bw per day; increase in hepatic labelling index at 90 mg/kgbw per day; degenerative changes in kidney at 34 mg/kg-bw per day; increase in renal labelling index at 180 mg/kg-bw per day
15 mg/kg-bw per day, LOEL (lowest dose tested)	Fatty cyst, liver; male and female	Beagle dog	Gelatin capsules; 6 days per week for 7.5 years	Heywood et al. (1979)
34 mg/kg-bw per day, LOAEL (lowest dose tested)	Increase in hepatic labelling index; histopathological changes (mild centrilobular hepatocyte swelling and pale eosinophilic staining); increase in renal labelling index; necrosis	Male B6C3F1 mouse	Gavage, corn oil, 4 days	Larson et al. (1994c)
34 mg/kg-bw per day, LOAEL (lowest dose tested)	Proliferation in olfactory epithelium; nasal lesions At higher doses: Increase in hepatic labelling index and slight centrilobular vacuolar change at 100 mg/kg-bw per day; increase in renal labelling index at 200 mg/kg-bw per day and necrosis	Female F344 rat	Gavage, corn oil, 4 days	Larson et al. (1995a)
34 mg/kg-bw per day, LOAEL 10 mg/kg-bw per day, NOEL	Histopathological alterations in liver (pale cytoplasmic eosinophilia of centrilobular hepatocytes and mild vacuolation of centrilobular and midzonal hepatocytes); alanine aminotransferase and sorbitol dehydrogenase significantly increased At higher doses: Hepatocellular proliferation significantly increased at next higher dose, 90 mg/kgbw per day	Female B6C3F1 mouse	Protocol Gavage, corn oil, 3 weeks	Larson et al. (1994a)
34 mg/kg-bw per day, LOAEL (lowest dose tested)	Minimal changes in olfactory epithelium	Female F344 rat	Gavage, corn oil, 5 days	Dorman et al. (1997)
37 mg/kg-bw per day, LOAEL (lowest dose tested)	Histopathological changes in kidney and liver	Male CD-1 mouse	Gavage, corn oil, 14 consecutive days	Condie et al. (1983)
41 mg/kg-bw per day, LOAEL	Increased liver weight and hepatocellular degeneration in F1 females and increased epididymal weight with ductal epithelial degeneration in males	Swiss CD-1 mouse	Continuous breeding protocol, gavage in corn oil	EHRT (1988)
50 mg/kg-bw per day, LOAEL (lowest dose tested)	Decreased humoral immunity in both sexes and increased relative liver weight in females	Male and female CD-1 mice	Gavage in Emulphor for 14 days	Munson et al. (1982)
50 mg/kg-bw per day, LOAEL (lowest dose tested)	Significant dose-related increase in liver weight and hepatic microsomal activity	Female CD-1 mouse	90-day gavage in Emulphor	Munson et al. (1982)
50 mg/kg-bw per day, LOEL	Maternal: decreased body weight gain	Female Sprague- Dawley rat	Gavage, corn oil, days 6-15 of gestation	Thompson et al. (1974)
50 mg/kg-bw per day, LOEL	Maternal: decreased body weight gain	Female Dutchbelted rabbit	Stomach tube, corn oil, days 6-18 of gestation	Thompson et al. (1974)
5 mg/kg-bw per day, LOAEL (lowest dose tested)	Increase in activity of serum alanine aminotransferase and serum sorbitol dehydrogenase; mild hepatocyte hydropic degeneration At higher doses: Increase in hepatocyte labelling index at next higher dose, 110 mg/kg-bw per day	Female B6C3F1 mouse	Gavage, corn oil; 3 weeks, 5 days per week	Melnick et al. (1998)
60 mg/kg-bw, LOEL	Significant decrease in liver weight, increase in enzyme activity	Male F344 rat	Gavage, aqueous vehicle, single administration; 24-hour sacrifice	Keegan et al. (1998)
60 mg/kg-bw per day, LOAEL	Dose-related increase in both absolute and relative weights of liver, both sexes	Male and female B6C3F1 mice	90-day gavage study, corn oil	Bull et al. (1986)
60 mg/kg-bw per day, LOAEL	Dose-related increase in both absolute and relative weights of liver, females only	Male and female B6C3F1 mice	Emulphor vehicle	Bull et al. (1986)

Since the tumorigenic and benchmark doses for cancer and non-cancer, respectively, are based on metabolized dose, they adjust for kinetic differences between animals and humans. An appropriate uncertainty factor for derivation of a Tolerable Intake for both cancer and non-cancer effects would therefore be in the range of 25, i.e., 10 (for intraspecies variation in toxicokinetics and toxicodynamics) x 2.5 (for interspecies variation in toxicodynamics) (Health Canada, 1994). Hence, the margins between estimated exposure and tumorigenic and benchmark doses for cancer and non-cancer, respectively, for chloroform are considerably greater than that considered as appropriate as a basis for development of Tolerable Intakes. As a result, exposure of the general population is considerably less than the level to which it is believed a person may be exposed daily over a lifetime without deleterious effect.

Table 15 Effect levels in laboratory animals exposed to chloroform by drinking water (presentation limited to those studies in which the lowest effect levels were reported)

Effect level 1	Concentra- tion in drinking water (ppm)	Endpoint	Species/sex	Protocol	Reference
17 mg/kg-bw per day, LOEL 6 mg/kg-bw per day, NOEL [intakes reported by authors]	200 60	Foci of regenerating renal proximal tubular epitheliumAt higher levels: At highest dose, mild histopathological changes in liver) (1800 ppm, 106 mg/kg-bw per day)	Male F344 rat	3 weeks	Larson et al. (1995b)
38 mg/kg-bw per day, LOAEL 19 mg/kg-bw per day, NOAEL [intakes reported by authors]	400 200	Histopathological alterations in kidney (Hard and Wolf, 1999; Hard et al., in press)	Male Osborne- Mendel rat	2 years	Jorgenson et al. (1985)
40 mg/kg-bw per day, LOAEL [intake when exposed to 200 ppm (Health Canada, 1994)] 12 mg/kg-bw per day, NOEL [intake when exposed to 60 ppm (Health Canada (1994)]	200 60	Significant increase in labelling index in kidney (medulla only)	Female B6C3F1 mouse	3 weeks	Larson et al. (1994a)
58 mg/kg-bw per day, LOAEL 33 mg/kg-bw per day, NOAEL [intakes reported by authors]	1800 400	Histopathological alterations in liver; decrease in labelling index in kidney	Male F344 rat	4 days	Larson et al. (1995b)

Table 16 Effect levels in laboratory animals exposed to chloroform by inhalation (presentation limited to those studies in which the lowest effect levels were reported)

Enlarge table

Table 17 Recommended concentrations in media for midpoint and upper-percentile exposure scenarios for PBPK modelling

Medium	Midpoint estimate		Upper-percentile estimate
	Conc.	Developed from	Conc.	Developed from
outdoor air (background)	0.14 µg/m3 (29 ppt)	arithmetic mean from NAPS data (n = 5463) for 1993-1996 1	0.31 µg/m3 (63 ppt)	95th percentile from NAPS data (n = 5463) for 1993-1996 1
outdoor air (commuting)	0.27 µg/m3 (55 ppt)	arithmetic mean from NAPS data (n = 800) for 4 "road" sites for 1989-19961	0.66 µg/m3 (135 ppt)	95th percentile from NAPS data (n = 800) for 4 "road" sites for 1989-1996 1
indoor air (all)	2.28 µg/m3 (465 ppt)	arithmetic mean from Concord Environmental Corporation (1992) data (n = 754) following lognormal imputation 2	8.0 µg/m3 (1630 ppt)	95th percentile from Concord Environmental Corporation (1992) data (n = 754) following lognormal imputation 2
95th percentile from Concord Environmental Corporation (1992) data (n = 754) following lognormal imputation 2	833 µg/m3 (170 000 ppt)	experimental data assessing the transfer efficiency of chloroform from tap water to shower air, assuming an average concentration 3	1950 µg/m3 (398 000 ppt)	experimental data assessing the transfer efficiency of chloroform from tap water to shower, assuming the 95th percentile of the distribution of concentrations 4
air in bathroom after showering	5 µg/m3 (1020 ppt)	estimated with the onecompartment model of Blancato and Chiu (1994), 5 assuming a bathroom volume of 13 m3 and air exchange rate of 2.2 air changes per hour (ACH) from Wilkes et al. (1992) 6 and an average concentration of chloroform in tap water	18 µg/m3 (3670 ppt)	estimated with the onecompartment model of Blancato and Chiu (1994), 5 assuming a bathroom volume of 13 m3 and air exchange rate of 2.2 ACH from Wilkes et al. (1992) 6 and the 95th percentile of the distribution of concentrations of chloroform in tap water
tap water (cold)	47.3 µg/L	arithmetic mean from provincial/territorial data (n = 6607) for 1990-1997 7	166 µg/L	95th percentile from provincial/territorial data (n = 6607) for 1990-1997 7
food (all)	0.0035 µg/g	Canadian data for 24 food items 8	0.0298 µg/g	Canadian and U.S. data for 131 food items 9

1. ¹NAPS data from Dann (1998). Arithmetic mean concentrations were calculated for samples of 24-hour duration. See Health Canada (1999) for further information.
2. ²These data are from Concord Environmental Corporation (1992). Twenty-four samples of indoor air were collected using passive sampling devices from 754 homes in nine provinces during 1991 and 1992. At a limit of detection of 3.5 µg/m3, chloroform was detected in only 10.7% of these indoor air samples. The distribution of concentrations was assumed to be lognormal. Arithmetic mean (i.e., 2.28 µg/m³) and geometric mean (i.e., 0.72 µg/m³) concentrations were estimated by lognormal imputation, as described in Health Canada (1999). A 95th-percentile concentration (i.e., 8.0 µg/m³) was also estimated.
3. ³Estimates of the average concentrations of chloroform in the air of a shower compartment during a 10-minute shower were developed in Health Canada (1999) for typical conditions of water temperature (i.e., approximately 40°C) and flow rates (i.e., 5 and 10 L/min), using the arithmetic mean and 95th percentiles of the distribution of concentrations of chloroform in tap water in Canada. A midpoint estimate of the average concentration was developed as follows. At an assumed concentration in water of 50 µg/L (compared with an arithmetic mean concentration of 46.4 µg/L; see Health Canada, 1999) and assuming minimum air exchange between the shower compartment and the adjacent (bathroom) area, estimates of the average concentration of chloroform in the air of the shower compartment during showering ranged from 300 to 1333 µg/m³. An average concentration of 833 µg/m³ was selected as the midpoint estimate, based on the assumptions of a water flow rate of 10 L/min and a transfer efficiency of 0.5 (i.e., 50% of the chloroform in the water passing through the shower head is assumed to be volatilized into the air of the shower compartment before the water passes through the shower drain).
4. ⁴An upper-end estimate of the average concentration in the air of a shower compartment during a 10-minute shower was developed in a similar manner. At an assumed concentration in water of 117 µg/L (the 95th percentile of the distribution of concentrations in tap water in Canada; see Health Canada, 1999) and assuming minimum air exchange between the shower compartment and the adjacent (bathroom) area, estimates of the average concentration of chloroform in the air of the shower compartment during showering ranged from 702 to 3120 µg/m³. An average concentration of 1950 µg/m³ was selected, also based on the assumptions of a water flow rate of 10 L/min and a transfer efficiency of 0.5.
5. ⁵Blancato and Chiu (1994) indicate that the equilibrium relation of the concentration of chloroform in air to the concentration in tap water can be described according to C_a = (f x R_w x C_w) ÷ (V_b x R_b), where: C_a is the resulting average concentration (mg/m³) of chloroform in the indoor air; f is the transfer efficiency (i.e., 0.5 assumed; see Health Canada, 1999); R_w is the rate of water use, expressed as L/shower, assuming a flow rate of 10 L/min and a duration of 15 minutes; C_w is the concentration (mg/L) of chloroform in the tap water (i.e., 0.0464 mg/L for the midpoint estimate, and 0.117 mg/L for the upper-percentile estimate); V_b is the volume (m³) of the bathroom (a volume of 13 m³ was assumed, based on Wilkes et al., 1992); and R_b is the bathroom ventilation rate (air exchanges/day).
6. ⁶Wilkes et al. (1992) estimated a range of air exchange rates: 0.8 ACH (19.2 per day) when the bathroom door is closed; 2.2 ACH (52.8 per day) when the bathroom door is open; and 7.4 ACH (178 per day) when the bathroom door is closed and an exhaust fan is operating. A bathroom ventilation rate of 2.2 ACH was assumed for both the midpoint and upper-percentile exposure scenarios.
7. ⁷Data concerning the distribution of concentrations of chloroform in treated tap water in Canada in the 1990s are summarized in Health Canada (1999).
8. ⁸TRanges of average intakes of chloroform from ingestion of foods for six age groups among the population were developed in Health Canada (1999) using average daily consumption rates (g/day) for 181 food items (EHD, 1998). The minimum intakes in the ranges were based on midpoint estimates of the concentrations of chloroform in 24 specific food items using data originating in Canada only.
9. ⁹The maximum intakes in the ranges were based on midpoint estimates of the concentrations of chloroform in 131 specific food items using data originating in Canada or the United States. For the adult age group, the range of intakes (assuming an average body weight of 70.9 kg) was from 0.084 to 0.71 µg/kg-bw per day. Equivalent intakes in µg/day are 5.96-50.3. The total average daily consumption of the 181 food items by adults is 2353 g (EHD, 1998). Among these, there are two specific food items that are generally prepared using tap water. These are tea (at 317 g/day for adults) and coffee (at 348 g/day for adults). No intake estimates were developed for tea or coffee, as data indicating concentrations of chloroform were not available. The total average daily consumption of 179 food items (i.e., excluding tea and coffee) by adults is (2353 - 665) 1688 g. This amount was divided by the minimum (5.96 µg/day) and maximum (50.3 µg/day) of the range of daily intakes to estimate average concentrations of chloroform in the food consumed. The resulting range of average concentrations is 0.0035 µg/g (i.e., midpoint estimate) to 0.0298 µg/g (i.e., upper-percentile estimate).

Table 18 Values of the input parameters representing the midpoint estimates of chloroform concentrations for use in the multimedia exposure scenario

Enlarge table

The lowest concentrations reported to induce cellular proliferation in the nasal cavities of rats and mice in short-term studies (i.e., 2 ppm [9.8 mg/m³]) were compared directly with the midpoint and 95th-percentile estimates of concentrations of chloroform in indoor air in Canada. These values were the same as those selected to run the human models for the kidney and liver. The midpoint and 95th-percentile estimates are 4298 and 1225 times less than the lowest value reported to induce a proliferative response in rats and mice (midpoint for indoor air = 2.28 µg/m³, 95th percentile = 8.0 µg/m³). Comparisons with midpoint and 95th-percentile estimates of concentrations during showering were considered unwarranted, since such exposures are intermittent and last for very limited periods of time during the day. Based on considerations similar to those mentioned above for cancer and non-cancer effects associated with ingestion of chloroform, these margins are considerably greater than that considered appropriate as a basis for development of a Tolerable Concentration.

Table 19 Values of the input parameters representing the 95th percentile of chloroform concentrations for use in the multimedia exposure scenario

Enlarge table

3.3.5 Uncertainties and degree of confidence in human health risk characterization

For the principal source of exposure of at least the older age groups of the general population to chloroform (i.e., showering), uncertainty is introduced by the assumption that concentrations in the water at the shower head are similar to those in the incoming cold tap water. Based on limited data, the average concentrations in the warm water may be twice as high as that in the incoming cold water during the summer months and up to 4 times as high as that in the colder incoming water during winter months (Benoit et al., 1997). Additional uncertainty is introduced by the assumption that the concentrations measured in the water treatment plants and distribution systems are representative of the concentrations at the consumers' taps, to which the general population is exposed. Available data indicate that average concentrations may be 50% higher at the most remote locations than at the water treatment plants, depending on the specific treatment processes used and other factors.

For indoor air, confidence in characterization of concentrations is less than that for other media due primarily to the limited number of homes sampled and lack of sensitivity of analysis in the available survey (Concord Environmental Corporation, 1992).

Concentrations measured were less than the limit of detection in approximately 90% of the samples from 754 homes, although the approach adopted for estimation of levels in these samples for characterization of exposure is not considered to be unrealistic or overly conservative.

There is a moderate degree of confidence in the quantitative estimates of the average intake of chloroform in drinking water for the general population. As indicated in relation to the estimates of exposure during showering, some uncertainty is introduced by the assumption that the concentrations measured in the water treatment plants and distribution systems are representative of the concentrations at the consumers' taps, to which the general population is exposed. This database included over 10 000 samples analysed between 1985 and 1997. Although analyses were performed by a number of different laboratories, sampling and analytical methods were similar. Although similar dechlorinating preservatives were utilized, the pH of the preserved samples was not adjusted concomitantly, and hence there may have been some alteration in concentrations of chloroform during storage (Lebel and Williams, 1995). Uncertainty in the quantitative estimates of daily intakes of chloroform is also introduced by assuming daily rates of intake of total tap water, which includes tap water used to prepare beverages. Concentrations of chloroform in hot beverages (e.g., tea and coffee) are unlikely to be as high as the concentrations in the cold tap water used for their preparation, as chloroform rapidly volatilizes from tap water during heating and boiling.

Although it contributes minimally to total exposure, there is a moderate degree of confidence in the characterization of the concentrations of chloroform in ambient air in Canada, due to the magnitude and sensitivity of the monitoring data. This was based on a large data set of 24-hour average concentrations, measured across the country, throughout the 1990s (Dann, 1998). Samples were collected by a standardized protocol, on a cyclical basis, at a fixed network of atmospheric monitoring sites, for analysis by a single, specialized laboratory. Confidence in the data is increased by the observation that chloroform was detected with similar frequencies and at similar ranges of concentrations in the ambient air from rural areas situated in widely separated geographical locations of Canada. The observation that the frequencies of detection and concentrations of chloroform were higher at suburban and urban locations than in these rural areas is also consistent with what might be expected on the basis of proximity to sources. Some uncertainty is introduced by the locations of the monitors, which are not strictly representative of personal exposure.

Uncertainty is introduced into the estimates of the average daily intake of chloroform from ingestion of foods by the assumption that the limited Canadian data available for a specific food item are representative of the concentrations generally encountered by the general population when ingesting that food item. Additional uncertainty is introduced by the assumption that concentrations of chloroform measured in specific food items in the United States are similar to the concentrations in those food items in Canada. Also, the concentration of chloroform was assumed to be zero in all food items for which data are not available. Nevertheless, there is a high degree of certainty that chloroform is not highly concentrated in foods in Canada, since chloroform is only moderately lipophilic and does not significantly biomagnify in food chains.

Confidence in the quantitative estimates of daily intakes of chloroform by ingestion for infants is low. As no data were available concerning the presence or concentrations of chloroform in human breast milk in Canada, estimates of intake for exclusively breast-fed infants could not be developed. Uncertainty is introduced by the assumption that infants are exclusively formula fed, since data concerning the presence or concentrations of chloroform in concentrated (i.e., powdered or liquid) infant formula were not available. As a result, the concentration of chloroform in the reconstituted infant formula was assumed to be identical to the concentrations in the domestic water supply. Similar uncertainty is introduced when it is assumed that infants are fed table-ready foods, due to the limitations identified previously regarding the concentrations of chloroform in the majority of food items consumed daily in Canada.

With respect to the toxicity of chloroform, the degree of confidence that critical effects in animal species are well characterized in the available database is high. Indeed, in numerous investigations in experimental animals by various routes of exposure, effects on the kidney, liver and nose have been consistently observed at lowest doses. The nature of the effects has been similar and generally consistent with a mode of action that involves cellular degeneration and death and regenerative proliferation induced by oxidative metabolites.

The degree of confidence in the database that supports an obligatory role of sustained cytotoxicity in the carcinogenicity of chloroform is also high, although there are some uncertainties. Indeed, there are few compounds for which the supporting database in this regard is as complete, consistent and cohesive as it is for chloroform. The weight of evidence in this regard is strongest for hepatic and renal tumours in mice. The evidence is more limited for renal tumours in rats, primarily due to the relative paucity of data, in strains where tumours have been observed, on metabolism and intermediate endpoints and the relationship between them. Uncertainty could be reduced, therefore, by acquisition of additional information on metabolism, cytotoxicity and proliferative response in the strain in which tumours were observed (i.e., Osborne-Mendel rats) following long-term exposure to chloroform. Additional data on metabolism and chronic (e.g., 2-year) cytotoxicity/proliferative response in the kidneys of F344 rats might also have contributed to greater confidence in the hypothesized mode of action.

While the overall weight of evidence for the genotoxicity of chloroform is negative, on the basis of available data, weak genotoxicity in the rat cannot be precluded. It would be desirable, therefore, to investigate further the possible nature of the interaction of chloroform with DNA in rats. Another area that could be clarified by further work is whether any of the metabolites of chloroform are DNA reactive.

For the PBPK model, among those parameters considered in the sensitivity analysis to have most impact on output, uncertainty was greatest for the metabolic parameters particularly in the kidney and for humans. Additional in vitro data on the metabolism of chloroform in the human kidney and liver would be useful not only to reduce uncertainty in these values, but, if performed on tissues from a number of individuals, potentially to address the issue of variability across the human population. In particular, it would be desirable to clarify whether the same pathways of metabolism contribute to the potential for cytotoxicity in rodents and humans, specifically with respect to CYP2E1 and other P450 isozymes. Determination of the kinetic constants for the CYP2E1 and CYP2B1 isoforms in vivo is also desirable and could be addressed through comparative kinetic analysis of gas uptake curves in phenobarbital-induced CYP2E1 knockout and normal mice. For the PBPK model for dogs, the blood/air partition coefficient in this species was considered to be similar to that for the rat, although for low-molecular-weight chlorocarbons, these are normally higher in smaller species, possibly due to variations in binding to hemoglobin. Similarly, for this model, local rates of metabolism in the dog were based on hepatic and renal distribution of CYP2E1 in rats.

Characterization of exposure-response for both non-cancer and cancer is based on increased incidence of the relevant endpoints (both fatty cysts in dogs and renal tumours in rats) for a small number of doses. However, the dose at which non-cancer effects were observed in this study is similar to the lowest reported effect levels for proliferative response in target organs of other species.

3.4 Conclusions

CEPA 1999 64(a): Based on available data, it has been concluded that chloroform is not entering the environment in a quantity or concentration or under conditions that have or may have an immediate or long-term harmful effect on the environment or its biological diversity. Therefore, chloroform is not considered to be "toxic" as defined under Paragraph 64(a) of CEPA 1999.

CEPA 1999 64(b): Based on available data, it has been concluded that chloroform is not entering the environment in a quantity or concentration or under conditions that constitute or may constitute a danger to the environment on which life depends. Therefore, chloroform is not considered to be "toxic" as defined under Paragraph 64(b) of CEPA 1999.

CEPA 1999 64(c): Based on available data, it has been concluded that chloroform is not entering the environment in a quantity or concentration or under conditions that constitute or may constitute a danger in Canada to human life or health. Therefore, chloroform is not considered to be "toxic" as defined under Paragraph 64(c) of CEPA 1999.

Overall conclusion:

Based on the information available, chloroform is not considered to be "toxic" as defined in Section 64 of CEPA 1999.

3.5 Considerations for follow-up (further action)

Since chloform is not considered "toxic" as defined in Section 64 of CEPA 1999, investigation of options to reduce exposure under CEPA 1999 is not considered a priority at this time. However, this is based upon current use patterns; thus, future releases of this compound should continue to be monitored to ensure that exposure does not increase to any significant extent.

In view of the fact that showering is estimated to be the single greatest contributor to total daily intake of chloroform from drinking water, measures to reduce uptake from this source will be most effective in minimizing exposure of the general public.