Subchronic toxicity of chlorine dioxide and related compounds in drinking water in the nonhuman primate.

Subchronic toxicities of ClO2, NaClO2, NaClO3 and NH2Cl were studied in the African Green monkeys (Cercopithecus aethiops). The chemicals were administered in drinking water during 30-60 days subchronic rising dose protocols. The only unexpected and significant toxic effect was elicited by ClO2; this chemical inhibited thyroid metabolism in the animals at a dose of ca. 9.0 mg/kg/day. A statistically significant decrease of serum thyroxine occurred after the fourth week of exposure to 100 mg/l.concentration. The extent of thyroid suppression was dose dependent in each individual monkey, and was reversible after cessation of exposure. NaClO2 and NaClO3 failed to elicit similar effects in doses up to ca. 60 mg/kg/day. Also, NaClO4 or NH2Cl did not cause T-4 suppression in doses of 10 mg/kg/day. The selective thyroid effect of ClO2 was unexplained and it appeared to be paradoxical since ClO2 was rapidly reduced by the oral and gastric secretions to nonoxidizing species (presumably Cl-). No evidence of thyroid effects were detected in the serum of human volunteers who ingested approximately 1 mg/l. of ClO2 in drinking water as a result of routine use in the community water treatment process. Sodium chlorite induced dose-dependent oxidative stress on hematopoesis, causing decreased hemoglobin and red cell count and increased methemoglobin content. At the same time, serum transaminase (SGPT) levels showed significant subclinical elevation. The hematologic effects of NaClO2 rebounded during exposure indicating compensatory hemopoietic activity taking effect during oxidative stress. Sodium chlorate and chloramine did not induce detectable hematologic changes in the animals.


Introduction
Owing to its excellent microbicidal properties, chlorine dioxide (CI02), a water-soluble yellow oxidant gas, has been used in the past for drinking water disinfection. The apparent relative absence of the carcinogenic trihalomethanes (THM) in C102 treated water triggered renewed interest in this compound as a possible alternative to chlorine (1), since the latter was shown to generate THMs (reacting) with humic substances (2,3).
Concomitantly, the toxicology of CG02 and its metabolites (CGi2 and C103) have received wide attention in the recent literature. By using orally administered C102, hematologic changes and inhibition in testicular uptake of 3H-thymidine was demonstrated in rats by Abdel-Rahman (4). Effects of these chlorine oxides on the glucose-6-phosphate dehydrogenase (G6PD)-deficient mouse were reported by Moore et al. (5). C102 associated kinetics of red cell GSH depletion and intravascular hemolysis in rats and chickens was reported by Abdel-Rahman et al. (6). These workers also described the metabolism of 36CI02 (7) in rodents. In addition, the effects of C102 and metabolites, as they effect the cellular GSH system in the rat, mouse and chicken blood, were investigated by Couri et al. (8) Oxidative in vitro damage to erythrocytes by NaCI02 was reported by Heffernan et al. (9). The same authors also examined the in vivo effects of NaCIO2, demonstrating dose dependent oxidative stress in rats (10). More recently, Michael et al. (11) of this laboratory published results of a prospective epidemiology study, showing the absence of clinicopathological effects in human volunteers to CG02 treatment of their community water supply. As an adjunct to the human study, we examined the subehronic toxicity of these chlorine oxides in nonhuman primates. Additionally we incorporated monochloramine (NH2CI) since this chemical is an ubiquitous component of chlorinated drinking water, often purposefully generated in the water treatment process.

Materials and Methods
The Experimental Design Each of the chemicals examined were administered to the animal colony in exponentially rising step doses, each period lasting 30-60 days depending on the availability of scheduled resources for sampling and testing the blood samples. At each dose changeover the animals were bled, followed by bi-weekly repeated bleeding. (See Table 1 for testing schedule). Between chemicals the animals were rested for 6-9 weeks, at the end of which time base lines in clinical parameters were re-established. For purposes of statistics and evaluation, each animal served as its own control.

The Animal Model
A small stable colony of African Green Monkeys (Cercopithecus aethiops) consisting of five adult males and seven adult females were used in the study. The body weights ranged from 3.0 to 5.7 kg, and their red cell G6PD activities ranged from 9.0-11.4 IU/g Hb with a mean of 8.8 IU/g Hb. All of the animals were in our possession for the past 8 years and repeated medical and laboratory examinations ascertained that the animals were in excellent health. Their hematological normal values are listed in Table 2. Repeated tuberculin tests and chest x-rays were performed throughout the study to assure the absence of mycobacterial infections. The animals were housed individually. They were fed a diet of Purina Monkey Chow supplemented daily with fresh fruit. During periods of exposure distilled water was made available ad libitum. For purposes of restraint light anaesthesia was induced by IM injection of ketamine HCI in uniform doses of 10.0 mg/kg.

Test Solutions
A stock C102 solution of 400-500 mg/l.was prepared by purging C102 from an acidified-NaCIO2 generator through an absorbent NaClO2 solid column into 1 gallon quantities of distilled deionized water. The stock solution was then diluted to the appropriate concentrations (30, 100 and 200 mg/l) and was dispensed to the animals in dark bottles (2-3 liters/animal) equipped with ball-valve sipping tubes to minimize drippage. The solutions were changed, and consumption was measured three times weekly. Concentration and purity of the solutions were determined before administration and at the time of refilling to determine the extent of hydrolytic and photolytic degradation during residence in the bottle. Ultraviolet spectroscopy E360nm= 1.1 x i05 mole-1cm-') and titrimetry according to Palin (12) was employed for the assays. Absence of Cl2 and of OC1 was verified by the AgNO3 test for Cl-. All consumption measurements were made by weight differential.
Chlorite and chlorate solutions (25, 50, 100, 200 and 400 mg/l. as ionic equivalents to C102) were prepared from the corresponding analytical grade sodium salts. Monochloramine solutions were prepared according to Guion (personal communication). Dispensing and dosage of these solutions were as described above.

Clinical Pathology Procedures
According to the experimental schedule shown in Table 1 blood was collected from the saphenous vein under light anaesthesia. Care was exercised not to draw more than 8 ml per bleeding period to avoid excessive blood loss and anemia.
Hematology tests (Table 1) were peformed according to standard procedures described by Wintrobe (13). For the determination of cell counts, cell indices and hemoglobin content a Coulter ZBI-6 cell counter and hemoglobinometer was used. Serum chemistries and enzyme activities were determined using a Union Carbide Centrifichem-400 kinetic analyzer. Red cell glucose-6-phosphate dehydrogenase (G6PD) was determined with the Calbiochem Reagent set using the Centrifichem procedure according to Favara (personal communication). Red cell glutathione (GSH) levels were determined manually by using the Biomedix DTNB reagent kit. Serum thyroxine (T-4) was determined with the SYVA enzyme amplified immunoassay (EMIT) reagent kit using the Centrifichem procedure as per SYVA's modifications. All chemistries, enzymes and T-4 determinations were made in duplicate. Quality assurance in the laboratory procedures was according to CAP standards, and quality control was achieved by using assayed and unassayed control specimen (normal, abnormal Coulter 4C, Dade Monitrol I and II).  In Vitro Deactivation of C102 by Saliva Pooled saliva was collected by buccal scraping from anaesthetized animals. The specimen was diluted 1:5 with distilled H20, and the resultant solution was used for the subsequent experiment. Various ratios of C102 solutions were mixed with the dilute saliva in a quartz cuvet, The absorbance of the cuvet was read in a Perkin-Elmer ultraviolet spectrophotometer at 360 nm against a distilled H20 blank. At the same time another aliquot of the mixture was prepared and titrated according to Palin (12). Recovered concentrations were calculated using the molar extinctiQn coefficient of C102 of 1.1 x 105mole-lcm-, and by titer equivalent.
Reactant ratios and percent recoveries are summarized in Table 3.
In Vivo Deactivation of C102 in the Gastric Space A 5.7 kg male animal was lightly sedated after an overnight fast and immobilized in a monkey-chair. The animal's stomach was intubated via the oropharyngeal route, and an aliquot of 30 ml C102 (60 mg/l.) solution was instilled into the gastric space with a glass syringe. Immediately after discharging, a 15 ml aliquot of the solution was withdrawn, and its total oxidizing capacity (sum of C102, C102 and part of C103) was determined (within 5 min) by iodometry at pH 1.0. Of the original C102 equivalent oxidizing titer, 8% was recovered after 5 min of total contact time. Because of turbidity, spectrophotometry could not be used to quantitate unchanged C102.

Thyroid Function Test on Human Sera
Frozen serum samples from 350 human volunteers of the prospective epidemiology study previously reported by Michael et al. (11) were in our possession. Samples were selected on the basis of medical history of the volunteers. Only euthyroid   individuals were included in the study. Thyroid status of individuals with known thyroidopathies were also confirmed. The method for thyroxine determination was identical to that used for the nonhuman primate samples. Table 4 lists all clinical tests with annotations of dose responses to each chemical tested.

Results
The thyroid hormone measurement results obtained with the monkey specimen are listed in Table 5, and the human T-4 values are summarized in Table 6. The oral doses of the four chemicals tested are also summarized in Table 6. Figure 1 is a graphic representation of the thyroid inhibition regression in twelve animals versus mean doses of C102 ingested. Each point signified by an X on the graph represents the XIY coordinate of the thyroxine shift versus the mean dose ingested by a single animal during the 4-week exposure to 100 mg/l. of C102.  Table 7 lists the mean SGPT and SGOT enzyme activities in 12 monkeys during the step dose NaClO2 administration.

Discussion
In view of the extent of hematologic oxidative effects and causative concentrations ranging up to 1000 mg/l. in rats and in other animals reported by Abdel-Rahman (4) and Moore (5), the apparent absence of hematologic and clinicochemical effects in the monkeys is not surprising. At the higher concentration stages of this study (100 and 200 mg/l.) the mean daily dose remained nearly constant at ca. 9 mg/kg/day. This observation reflects on the strong irritating nature of C102 solutions. During the 200 mg/l. exposure, erythema and ulceration of the oral mucosa, mucous nasal discharge and avoidance of drinking water by the animals was observed. Throughout the C102 study liquid consumption decreased with the strength of the solu-100 ppm. Slope = 1.47; y tion. For example, at 0 mg/l. the mean water consumption was ca. 125 ml/kg/day with some seasonal fluctuations; at 100 mg/l. the consumption decreased to ca. 95 ml/kg/day, which further decreased to ca. 55 ml during the 200 mg/l. concentration. In fact, the high dose study was terminated after one week because some of the animals showed signs of dehydration and azotemia.
The most striking effect of C102 was on the thyroid gland. At the ca. 9 mg/kg/day dose this chemical appeared to be a potent inhibitor of thyroid synthesis ( Table 5). Analysis of the water consumption data at the 100 mg/l. exposure disclosed that the T-4 depression in each animal was statistically related to water consumption and C102 dose (Fig. 1). Since the animals' fluid intake was near normal, the effect cannot be explained by dehydration. Adverse influence of ketamine-nitrous oxide anaesthesia on thyroid metabolism in human patients was published by Matsuki et al. (14). The possibility of ketamine-induced thyroid inhibition, however, must be discounted, since we followed identical ketamine anaesthesia protocols during each study of the four chemicals. One may also propose that C102and/or C103-, which were shown to be metabolites of C102 (6), could inhibit iodine metabo-  This hypothesis is invalid because NaCl( failed to elicit thyroid inhibition in d mg/kg/day. Moreover, NaCI04 adminis ing water did not show antithyroid eff 20 mg/kg/day doses during a 30-day p haps the strongest argument again hypothesis arises from our in vivo re which demonstrated that ingested C reduced in the acidic stomach juices tc species (probably Cl-). We were ab only 8% of the total oxidizing capacity the C102, instilled into the animal sto min. of contact time with the gastric finding is in partial agreement with t ing studies of Abdel-Rahman et al. (6 strated approximately 80% of the ra plasma was in the Cl-form and aboi 36C1O2 was metabolized to C102in ti ilar distribution of radiochlorine in Ab study was found in the urine of the a Based on these considerations, it is the absorption of a simple chlorine caused the thyroid effects. An alterr nism may be that decrease of dietary i tion in the GI tract due to C102 ind pathology. Abdel-Rahman demonstral distribution of radiochlorine after 3 BERCZ ET AL. monkey serum stration was prominently high in the stomach and intestines 72 hr post-administration, indicating the halogen may be covalently attached to the mucosal ALT, IU/1, surface. Inhibition of iodine absorption and alter-4.1 ation of the bioavailability of dietary iodine in the 5.7 injured intestinal tract would result in progressive 6.4 iodine deficiency. This was supported by the obser- 20.0 vation that the T-4 deficiency developed slowly (4 weeks) and progressively in the animals. Another possibility is that chemical reactions between nutrients and C102 give rise to thyroid inhibitory molecules in the GI tract. Such products rn goitrogenic may interfere with iodine uptake in the thyroglobulin, Eqs. (1)-(3). or they could displace T-4 from binding sites of the carrier, thyroxine-binding globulin. Past toxicologic studies with food ingredients treated with C102 c1o2- (1) (e.g., bleached flour) disclosed no adverse effects on rabbits, monkeys (17) and rats (17), and on dogs (17,19). Moran et al. (20) identified several modified amino acids in wheat gluten treated with C102, e.g., methionine sulfone and mono-and dichlorothyrosine. + C103- (2) Other chemical effects were also seen, such as decrease of tocopherol content, etc. Considering the chronology of these studies and the unavailability of methods for thyroid assessment at the time, [ i s of I (3 effects of the CG02 modified nutrients on thyroid ism within metabolism, if they existed, would have been missed. 'oglobulin, From Abdel-Rahman's work an additional possibilus to C104. ity emerges. About 25% of the radiolabeled chlorine in the liver of rats administered 36C102 is bound to D2and NaC103 proteins in the cellular sap, indicating the radioloses up to 60 chlorine species may be bound by covalent or by tered in drinkstrong hydrogen bonding to these proteins. Such rects in 10 and findings could be explained by incorporation of chloilot trial. Per-rinated amino acids in the protein matrix, or conist the above versely involvement of chlorinated amino acids in covery study, thyroid metabolism could be a possible explanation 1102 is rapidly for the observed effect. ) nonoxidizing Our attempt to find thyroid effects in the serum ile to recover of human volunteers was unsuccessful. This group equivalent of consumed only about 1 mg/l. of C102 for every liter )mach, after 5 of drinking water, in which the C1O2and C103zontents. This content did not exceed 5 mgIl. (11). The estimated he radiolabel-human dose of C102 in this study was only about one ) that demon-thousandth (1 x 10) of that administered to the ,diochlorine in monkeys.
at 20% of the Further research is in progress to determine the ie rat. A simnature and mechanism of the thyroid inhibitory del-Rahman's effect of C102nimals.
unlikely that oxide species iative mechaiodine absorpluced mucosal ted in rats the 3CO2 admini-Chlorite Previous findings of Heffernan et al. (10) and Moore et al. (5) demonstrating oxidative stress induced methemoglobinemia and anemia in rats was verified in this study. Our results show that at most doses C102induced a self-compensating oxi-dative stress in the monkey hematopoiesis. About midway through exposure, a rebound phenomenon occurred in hemoglobin and red cell synthesis. Figures 2 and 3 show male and female hematologic parameters plotted in terms of Delta Standard Deviations (DSD) for comparability. The reticulocytotic response (Fig. 3, solid lines) to C102exposure was not accompanied by dose responsive methemoglobinemia.
During NaCl02 exposure a dose-dependent rise of ALT was detected in the monkeys. The extent of enzyme evaluation was subclinical ( Table 7) and was not corroborated by concomitant elevation of any other enzyme system or by serum bilirubin. The significance of this observation is not known, but may be associated with accelerated hepatic activity during the transient oxidative hemolytic period.

Chlorate
The effects of chlorate were similar to those seen during chlorite exposure (Fig. 4 and 5), although the rebound phenomenon was not as clearly discernible.

Chloramine
At the attainable maximum concentration (100 mg/l.) this compound appeared to have no detectable effect on the hematology of the animals, including red cell GSH content. No evidence of thyroid suppression was detected in the serum.

Conclusions
The thyroid inhibitory effects of C102 ingestion appears to be a significant health endpoint ofunknown explanation. The potential for adverse health effects during long-term chronic exposure to low levels of C102, specifically to CI02-modified nutrients, merits further research.