Chronic toxicity of chloroform to Japanese medaka fish.

Japanese medaka (Oryzias latipes) were continually exposed in a flow-through diluter system for 9 months to measured chloroform concentrations of 0.017, 0.151, or 1.463 mg/L. Parameters evaluated were hepatocarcinogenicity, hepatocellular proliferation, hematology, and intrahepatic chloroform concentration. Histopathology was evaluated at 6 and 9 months. Chloroform was not hepatocarcinogenic to the medaka at the concentrations tested. Chronic toxicity was evidenced at these time points by statistically significant ([alpha] = 0.05) levels of gallbladder lesions and bile duct abnormalities in medaka treated with 1.463 mg/L chloroform. We assessed hepatocellular proliferation by exposing test fish to 5-bromo-2'-deoxyuridine in the aquarium water for 72 hr after 4 and 20 days of chloroform exposure; we then quantified area-labeling indices of the livers using computer-assisted image analysis. We observed no treatment-related increases in cellular proliferation. We analyzed cells in circulating blood in medaka after 6 months of chloroform exposure. Hematocrit, leukocrit, cell viability, and cell counts of treated fish were not significantly different from those of control fish. Using gas chromatography (GC), we evaluated intrahepatic concentrations of chloroform in fish after 9 months of exposure. Livers from the 0.151 and 1.463 mg/L chloroform-treated fish had detectable amounts of chloroform, but these levels were always lower than the aquaria concentrations of chloroform. Thus, it appeared that chloroform did not bioaccumulate in the liver. Unidentified presumptive metabolite peaks were found in the GC tracings of these fish livers.

Chloroform is a common drinking water disinfection by-product. Recently, present U.S. Environmental Protection Agency (U.S. EPA) drinking water standards for chloroform have been scrutinized (1,2): "Are today's standards based on rodent megadose studies relevant to real world exposures?" Carcinogenic results in these rodent studies were found at concentrations several orders of magnitude higher than the chloroform maximum contaminant level goal (MCLG) of 0 mg/L and the total trihalomethane maximum contaminant level (MCL) of 0.08 mg/L (3).
Nontraditional models such as fish can be used to study chloroform-induced toxicity at exposures closer to concentration levels found in drinking water. Fish have been shown to be sensitive to trace levels of contaminants in aquatic media (4)(5)(6)(7)(8). As waterdwelling organisms, fish receive dermal exposure through whole-body immersion in the exposure solution. Intake of test material also occurs through feeding and respiration.
Japanese medaka have been studied extensively both in the United States and in Japan (9)(10)(11)(12)(13). Their hardiness, small size, ease of culturing, and relatively short timeto-tumor response make the medaka an attractive test model. Sections of the entire animal will fit on one microscope slide so that examination of every tissue is possible in its anatomical context. The low rate of spontaneous neoplasms in medaka aids in the interpretation of bioassay results (14).
Our purpose was to determine the effect of chloroform on Japanese medaka after 9 months of continuous exposure. Specific end points evaluated were severity and prevalence of neoplasms, hepatocellular proliferation, hematology, intrahepatic chloroform concentration, fish growth, and fish survival. Test design. We chose the amount of chloroform routinely found in dechlorinated tap water as the lowest concentration tested. In 96-hr range-finding studies with juvenile medaka, a length NOEL (no-observed-effect level) of 21-25 mg/L and a weight NOEL of 22-26 mg/L were established for chloroform (28). We chose a log scale for the three concentrations of chloroform, which ensured that all concentrations tested were below the NOELs established in acute testing. Four test aquaria were randomly assigned to each nominal concentration of 0, 0.015, 0.15, and 1.5 mg/L chloroform.

Materials and Methods
The test began when 14-day-old fry (± 1 day) were randomized to each 5-gallon test aquarium, with 80 fry and approximately 15 L test solution per sealed aquarium. The proportional diluter was set to deliver 300 ± 15 mL to each aquarium every 3 min ± 15 sec, yielding 9-10 tank volumes per day.
During the first month of chloroform exposure, five fish per aquarium for each of four time points were used for the hepatocellular proliferation assays. After 6 months of chloroform exposure, we evaluated 20 fish per aquarium by histopathology, and an additional five fish per aquarium by hematology. At 9 months of exposure, five fish per aquarium were used for the chloroform intrahepatic concentration analysis. We euthanized all remaining fish to assess the following tissues by histopathology: bone (vertebra), brain, chromaffin tissue, corpuscle of Stannius, esophagus, eye, gallbladder, gill, heart, hematopoietic tissue, interrenal tissue, intestine, kidney, liver, nares, ovary, pancreas, peripheral nerve, pineal organ, pituitary gland, pseudobranch, skeletal muscle, skin, spinal cord, spleen, stato-acoustic organ, swim bladder, testis, thymus, thyroid tissue, urinary bladder, and gross lesions.
Statistical analyses. We analyzed lengths, weights, and histopathology data statistically for the chronic carcinogenicity test. For these and all other test end points, we applied normalizing transformations where required, and selected the best-fitting model for each test end point. We used analysis of variance (ANOVA) to test for effects, and regression analysis for point estimation. We used SAS PROC GLM computer software (SAS Institute, Inc., Cary, NC) for these analyses (29).
Chloroform chemical analyses. All exposure tanks were sampled weekly and analyzed on the day of collection. Daily stock solutions were collected and stored at 5°C until the weekly analysis of the exposure tanks. All samples were collected in 40 mL borosilicate glass U.S. EPA water-sampling vials with Teflon-lined silicone rubber septa in the cap.
To analyze samples we used a Hewlett Packard 6890 gas chromatograph equipped with an electron capture detector and interfaced to Hewlett Packard model 7694 headspace sampler (Agilent Technologies, Palo Alto, CA). We used a Hewlett Packard ChemStation for instrument control and data acquisition. The capillary column used throughout the study was a Hewlett Packard 25-m HP-1 (cross-linked methyl polysiloxane), 0.2 mm i.d. and 0.33 µm film thickness. The oven temperature of the gas chromatograph held isothermally at 40°C, the inlet was set at 250°C, and the electron capture detector was maintained at 300°C.
We chose headspace analysis for this study for its sensitivity, simplicity, and rapid throughput. Headspace analysis has been used elsewhere to analyze rat liver, urine, and blood (30,31) and was applied here to fish tissue and water samples. For the weekly analysis of exposure tanks, three 5 mL aliquots of each sample were placed in 10 mL headspace vials and sealed with Teflonlined crimp caps. The vials were then placed in the headspace sampler with the sampler oven set at 60°C, the 1 mL sample loop at 65°C, and the transfer line to the gas chromatograph at 70°C. The sample vial equilibration time was 30 min with agitation set on high. External calibration standards were used for this method and prepared fresh on the day of analysis. The technique required no further sample preparation and there were no interfering contaminants from the sample matrix. The detection limit was 0.003 mg/L and the recovery was 95.2%.
BrdU chemical analyses. We collected samples for chemical analysis at the initiation and termination of each exposure. Spiked samples showed an average recovery of 102%. We used a Hewlett Packard 1050 series high performance liquid chromatograph equipped with a variable wavelength detector, autosampler, and Hewlett Packard Chemstation (Agilent Technologies) to analyze BrdU samples. Samples were filtered through a 0.45 µm membrane before analysis. The solvent delivery system was programmed to deliver 15% methanol/85% water at a flow rate of 1.5 mL/min. The UV detector was set at 277 nm to monitor the eluate. A Supelco TM LC-18 column (25 cm × 0.46 i.d., 5 µm particle size; Supelco, Bellefonte, PA) was used for the separation. The injection volume was 5 µL.
Hepatocellular proliferation assays. Five fish per aquarium were removed from each of the 16 test aquaria on test days 2, 4, 7, and 20, exposed to 75 mg/L BrdU for 72 ± 4 hr, and euthanized with an overdose of tricaine methane sulfonate (MS-222) on test days 5, 7, 10, and 23, respectively. BrdU exposure, fixation, sectioning, staining, and counting methods have been described previously (26). We used a Bioquant/True Color Image Analysis System (Bioquant-R&M Biometrics, Inc., Nashville, TN) to evaluate BrdU-labeled slides, with five BrdU-stained sections from each fish. A count labeling index (CLI) was done to validate the area labeling index (ALI) method and to ensure that hepatic cell size did not change due to chloroform treatment. R 2 values comparing the CLI and ALI for the 4-day and 20-day sacrifice points were 0.934 and 0.921, respectively.
Chronic carcinogenicity. At 6 and 9 months of chloroform exposure, fish were euthanized with an overdose of MS-222. Fish necropsy procedures, fixation, sectioning, and staining were done as described previously (32). Fish tissues, listed in "Test Design," were evaluated by histopathology.
Fish hematology. During the fish chronic test, five fish per aquarium were removed after 6 months of continuous exposure to chloroform. Fish were euthanized by an overdose of MS-222, and then weighed and measured. A capillary tube (10-20 µL) of blood was removed from each fish, and each fish's anterior kidneys, a major site of hematopoeisis, were excised. Hematocrit and leukocrit were measured for each blood sample. Cell counts were performed using a hemacytometer. Cell viability was assessed (trypan blue exclusion) from kidney cell suspensions.
Fish intrahepatic chloroform concentration. At 9 months of chloroform exposure, five fish per aquarium were removed from the test for analysis of chloroform intrahepatic concentration. Fish were euthanized with an overdose of MS-222, weighed, and measured. The livers were excised, weighed, flash frozen in individual cryovials in liquid nitrogen, and then stored at -70°C. On the day of analysis, livers were thawed and transferred to 10 mL glass head space vials containing 5 mL of 2% (w/v) sodium dodecyl sulfate (SDS). Two percent SDS was added to denature liver enzymes immediately, thus terminating metabolism. The vials were capped and heated at 60°C for 2 hr before headspace analysis by capillary gas chromatography (as described in "Chloroform Chemical Analyses"). The detection limit for chloroform in the SDS solution was 0.5 µg/L, which was converted to milligrams of chloroform per gram of liver tissue. Recovery was 95.2%.

Results
Water quality. Water quality parameters monitored were temperature, pH, dissolved oxygen, conductivity, alkalinity, hardness, and un-ionized ammonia. All water quality parameters were within acceptable limits Fish growth and survival. Mortality through the 9-month exposure period was < 4% in all control and treated groups. Growth measurements are summarized in Table 1. At 6 months, there was a suggestion of growth reduction, but these results were not statistically significant. Comparison of estimated means and their 95% confidence limits revealed a reduction in fish length (f = 7.66, p = 0.0059) with the highest test-concentration fish smaller than control fish, 23.8 mm (23.43, 24.18) and 24.8 mm (24.23, 25.47), respectively. There was also a reduction in fish weight (f = 5.41, p = 0.025) overall. At 9 months, no reduction in growth was found for length (f = 3.22, p = 0.0732) or weight (f = 1.58, p = 0.2090).
Fish histopathology. At 6 months of chloroform exposure, the only significant (f = 10.74, p = 0.0055) finding for males at the 1.464 mg/L concentration was proliferation, or hyperplasia, of bile ducts of the liver. In contrast, female medaka at the 1.464 mg/L concentration exhibited nine significant findings in the bile ducts of the liver and the gallbladder. As in males, females had an increase (f = 23.46, p = 0.0003) in bile duct hyperplasia. Additionally, females had bile duct epithelium hyperplasia (f = 7.84, p = 0.0142), dilatation of the bile ducts (f = 18.93, p = 0.0007), and concretions in the lumen (f = 42.98, p = 0.0001). Granulomatous pericholangitis-an inflammation around bile ducts characterized mainly by macrophages with variable numbers of lymphocytes-significantly (f = 31.62, p = 0.0001) occurred in females, presumably as a result of bile leakage into the liver parenchyma. Concretions also significantly (f = 35.39, p = 0.0001) occurred in the lumen of the gallbladder. The epithelium of the cystic duct as well as the gallbladder itself exhibited significant hyperplasia (f = 36.17, p = 0.0001 and f = 19.94, p = 0.0005, respectively). Dilatation of the cystic duct was significant (f = 10.60, p = 0.0057) in female medaka.
After 9 months of exposure to various concentrations of chloroform, there were significant differences overall between males and females in their responses to chloroform for gallbladder concretions (f = 5.76; p = 0.0231), cystic duct hyperplasia (f = 4.94; p = 0.0341), and bile duct epithelium hyperplasia (f = 6.43, p = 0.0169). These changes occurred with greater frequency among females than among males.
At 9 months and with increased measured chloroform concentrations, males demonstrated a significantly (f = 9.57; p = 0.0079) higher incidence of dilatation of the cystic duct of the gallbladder and a tendency toward a significantly (f = 4.12; p = 0.0617) higher incidence of hyperplasia of the epithelium of the gallbladder. Females responded to increased measured chloroform concentrations with a higher incidence of hyperplasia of the cystic duct of the gallbladder (f = 25.73, p = 0.0002), hyperplasia of the gallbladder epithelium (f = 5.94, p = 0.0287), concretions in the lumen of the gallbladder (f = 32.51, p = 0.0001), and granulomatous inflammation in the wall of the gallbladder (granulomatous cholecystitis) (f = 8.30, p = 0.0121) characterized by the presence of macrophages and lymphocytes.
Few liver neoplasms were observed for control and treated fish at 6 and 9 months. One hepatocellular adenoma, a benign neoplasm of hepatocytes, was observed in a male exposed to 0.151 mg/L chloroform for 6 months, while no malignant hepatocellular neoplasms occurred in any 6month-exposed fish. At 9 months, one hepatocellular carcinoma (a malignant neoplasm of hepatocytes) was observed in a control female. Two hepatocellular adenomas occurred in medaka examined at 9 months: one in a male treated with 0.017 mg/L chloroform and one in a female treated with 0.151 mg/L chloroform.

Articles • Chloroform chronic toxicity to fish
Environmental Health Perspectives • VOLUME 109 | NUMBER 1 | January 2001 A tabular summary of non-neoplastic tissue alterations in and around the liver in males is shown in Table 2, females in Table  3. Trends seen in the occurrence of gallbladder lesions and bile duct abnormalities at 6 months were confirmed at 9 months. Representative photomicrographs of tissue alterations occurring at the 1.463 mg/L level of chloroform are shown in Figures 1 and 2. Table 4 shows relevant findings at 6 and 9 months where responses were observed at both time points. The overall trend of the data was to have findings of no significance or marginal significance at 6 months and to have a significant finding at 9 months, except for cystic duct concretions in 6-month females and bile duct hyperplasia in 9-month males, which showed the opposite trend.
Fish hematology test. The hematology analysis was conducted after 6 months of chloroform exposure. ANOVA performed on data from chloroform-treated fish for hematocrit, leukocrit, cell viability, and cell count demonstrated that none of these parameters were significantly different from those of controls (p = 0.05).
Fish intrahepatic concentration. We detected no chloroform in any of the 20 fish livers from the 0 or the 0.017 mg/L aquaria concentration groups. Of the 20 fish analyzed in the 0.151 mg/L aquaria concentration group, 2 fish livers had measured concentrations of chloroform (33 and 133 mg chloroform/g of fish liver). At the 1.463 mg/L aquaria chloroform concentration, 9 fish livers out of the 20 sampled had detectable amounts of chloroform (23,26,35,41,128,144,159,194, and 219 mg/g).
One or two other unidentifiable peaks were seen frequently on the chloroform chromatographs, as illustrated by a representative chromatograph in Figure 3. One of these peaks was in size equal to or greater than the chloroform peak. Repeated attempts to identify these other peaks through related experiments have not yet been successful.

Discussion
In medaka exposed to chloroform concentrations ranging from 0.017 to 1.463 mg/L, induction of liver neoplasms due to chloroform exposure was not significantly different (α = 0.05) in treated fish when compared to control fish, after 6 or 9 months of exposure. Gallbladder abnormalities and bile duct abnormalities were observed in treated fish at significantly increased frequencies at the 1.463 mg/L chloroform level. Chloroform was not hepatocarcinogenic to the medaka after 6 or 9 months of exposure.
Numerous pathology findings were dissimilar between chloroform-exposed rodents and the current fish study. In rats (477 mg/kg chloroform/day, 48-hr duration, single gavage dose) and mice (30 ppm chloroform, 90-day duration, inhalation; and 30 ppm, 14 days of dosing with 2-year duration, inhalation), males had a greater sensitivity to chloroform presumably through testosterone receptor mechanisms in the proximal convoluted tubular cells (33)(34)(35)(36). In medaka, females demonstrated a greater sensitivity to chloroform at both 6 and 9 months. Chloroform target organs in rodents were kidney, liver, and nasal passages (33)(34)(35), while in medaka only the gallbladder and bile ducts showed tissue abnormalities. In an inhalation study, rats that were exposed to 300 mg/L chloroform developed intestinal crypt-like ducts with periductular fibrosis from nonbiliary cells in their livers (37). No such corresponding abnormality was observed during fish pathology in the current study. Interestingly, the rodent studies revealed an association    between hepatocyte labeling indices and prevalence of aberrant ducts (23,(33)(34)(35). In the fish, we saw no concurrent increase in hepatocyte labeling and tissue abnormalities. Some of these differences can be attributed to route of exposure and applied concentration levels, while others are undoubtedly related to choice of animal model. The reason for the occurrence of concretions in the gallbladder and the bile ducts of the medaka is unknown. Concretions are rare in animals (38,39). Previously, biliary concretions have been identified in monkeys, cattle, and pigs (39). The principal constituents of these concretions were cholesterol, precipitated bilirubin, and calcium carbonate, respectively. When concretions occur, it is not unusual for them to be composites of these three materials (39). The cause of these concretions is usually an infection. In sheep choleliths (concretions), Pseudomonas aeruginosa has been identified as the infectious agent (39). The mechanism for formation is usually solid particles of dead organisms serving as a nidus for crystallization. Additionally, disturbances in the resorptive activities in the gallbladder may promote the development of concretions (38). Further investigations are necessary to determine the origin and makeup of these concretions in chloroform-treated medaka.
Hepatocellular proliferation in chloroform-treated fish livers was not significantly different from proliferation rates observed in control fish livers at 4 and 20 days of chloroform exposure. The exact cause of fish death in the other cell proliferation time points was not determined, but we suspect that the high mortality resulted from fungal and bacterial contamination in the stagnant processed well water delivery line used to make the BrdU exposure solutions.
Rodent studies have shown a positive association between increased cell proliferation and the hepatocarcinogenicity of a compound (2,40,41). In a previous study (26), this relationship between carcinogenicity and cell proliferation was also seen in the medaka after a 48-hr exposure to the liver carcinogen, diethylnitrosamine.
Intrahepatic chloroform concentration was less than external aquaria concentrations, with a higher number of fish livers containing chloroform at the 1.463 mg/L concentration. Phosgene is a known mammalian metabolite of chloroform (42), but we did not confirm its presence in fish livers exposed to chloroform. Chloroform does not appear to have bioconcentrated in fish liver; the chloroform intrahepatic concentration was always lower than the external aquarium concentration.
Chloroform was not acutely toxic to fish at concentrations two or three orders of magnitude above median drinking water levels. Chronic toxicity effects of chloroform were demonstrated by statistically significant findings in the gallbladder and bile ducts of fish treated with 1.463 mg/L chloroform.
Given the significant biliary findings observed at the high concentrations without a similar early induction of cell proliferation, it is intriguing to hypothesize that in initiated populations of cells (43), chronic exposure to the high concentration of chloroform may promote biliary carcinogenesis. It is also significant that although we observed no evidence of early hepatocellular necrosis and compensatory hyperplasia, the highest concentration of chloroform appeared to cause a chronic hyperplasia in animals sacrificed after 9 months of exposure. Further studies are warranted with this nonmammalian vertebrate model to add to the weight of evidence in public health decisions about balancing potential disinfection by-product toxicity and disease risk from microbial contamination.  Chromatogram of a fish liver sample used to determine intrahepatic chloroform concentration in medaka fish exposed to 1.463 mg/L chloroform for 9 months. Peaks include the injection peak, two unknown peaks, and the chloroform peak, respectively. Instrumental detection limits of chloroform were 0.001 mg/L. Time (min)