http://informahealthcare.com/iht ISSN: 0895-8378 (print), 1091-7691 (electronic) Inhal Toxicol, 2014; 26(3): 175–184 ! 2014 Informa Healthcare USA, Inc. DOI: 10.3109/08958378.2013.872212

RESEARCH ARTICLE

Comparison of sarin and cyclosarin toxicity by subcutaneous, intravenous and inhalation exposure in Gottingen minipigs Stanley W. Hulet, Douglas R. Sommerville, Dennis B. Miller, Jacqueline A. Scotto, William T. Muse, and David C. Burnett US Army Edgewood Chemical and Biological Center (ECBC), Aberdeen Proving Ground, Aberdeen, MD, USA

Abstract

Keywords

Sexually mature male and female Gottingen minipigs were exposed to various concentrations of GB and GF vapor via whole-body inhalation exposures or to liquid GB or GF via intravenous or subcutaneous injections. Vapor inhalation exposures were for 10, 60 or 180 min. Maximum likelihood estimation was used to calculate the median effect levels for severe effects (ECT50 and ED50) and lethality (LCT50 and LD50). Ordinal regression was used to model the concentration  time profile of the agent toxicity. Contrary to that predicted by Haber’s rule, LCT50 values increased as the duration of the exposures increased for both nerve agents. The toxic load exponents (n) were calculated to be 1.38 and 1.28 for GB and GF vapor exposures, respectively. LCT50 values for 10-, 60- and 180-min exposures to vapor GB in male minipigs were 73, 106 and 182 mg min/m3, respectively. LCT50 values for 10-, 60 - and 180-min exposures to vapor GB in female minipigs were 87, 127 and 174 mg min/m3, respectively. LCT50 values for 10-, 60- and 180-min exposures to vapor GF in male minipigs were 218, 287 and 403 mg min/m3, respectively. LCT50 values for 10-, 60- and 180-min exposures in female minipigs were 183, 282 and 365 mg min/m3, respectively. For GB vapor exposures, there was a tenuous gender difference which did not exist for vapor GF exposures. Surprisingly, GF was 2–3 times less potent than GB via the inhalation route of exposure regardless of exposure duration. Additionally GF was found to be less potent than GB by intravenous and subcutaneous routes.

Cyclosarin, intravenous, LCT50, LD50, nerve agents, sarin, subcutaneous

Recent events in Syria have highlighted the importance of continuing to investigate the toxic effects of chemical weapons. Organophosphorous nerve agents, such as sarin (GB) and cyclosarin (GF), are potent cholinesterase inhibitors and therefore extremely toxic. Additionally they are both highly volatile, making them ideal for dispersion as vapors. Traditionally, the military and other organizations dealing with inhalation toxicology have accepted Haber’s (1924) principle of dosage, the product of concentration (C) and time (T), as constant over time when assessing the impact of nerve agent vapor exposures. Thus, Haber’s rule was used to extrapolate dose–response data (based upon relatively short exposure times) to predict response probabilities involving longer exposure times. However, this concept is now considered inadequate for assessing the biological effects of exposure to many acutely toxic gases and aerosols (ten Berge et al., 1986). For even a clear toxicological endpoint as lethality, historical assumptions used to extend the prediction of exposures over time have been shown to be overly conservative for GB, which is the best-studied agent. There is

Address for correspondence: Stanley W. Hulet, PhD, DABT, US Army ECBC, Building E3150, Aberdeen Proving Ground, Aberdeen, MD 21010, USA. Tel: +410 436 2685. E-mail: [email protected]

Received 9 October 2013 Revised 20 November 2013 Accepted 2 December 2013 Published online 31 January 2014

even less available data on the relative effects of concentration and duration of exposure on toxicity for GF. As a consequence, more recent studies (Hulet et al., 2006a; Mioduszewski et al., 2002a; Whalley et al., 2004) have focused on generating data sets that include low concentration exposures over long durations. These data are best described with a toxic-load model (ten Berge et al., 1986), rather than the traditional Haber’s model. In the toxic-load model, the median effective dosage (ECT50) increases with exposure time in a non-linear relationship and the data can be fit to the toxic load equation, CnT ¼ k. In this equation, C is the concentration, T is exposure duration and n is the toxic load exponent, which depends on the particular vapor or exposure scenario. In rats, others have shown that the lethality of vapor GB (Mioduszewski et al., 2002b) and GF (Anthony et al., 2004) exposures is best modeled by toxic load. To develop models for predicting the probability of toxicity from low-level nerve agent exposures for different concentrations and durations of exposure, additional data from a non-rodent species are needed. Pigs have been found to be similar in anatomy and physiology to humans (Smith, 2000) leading to the increased use of minipigs as a non-rodent model alternative. In 2006, we (Hulet et al., 2006a) demonstrated that the biological endpoint of pupil constriction in minipigs following vapor exposure to GB or GF is best modeled with the toxic load model. The intent of the current

20 14

Introduction

History

176

S. W. Hulet et al.

studies is to estimate the lethal concentrations of GB and GF as a function of exposure-duration in the Gottingen minipig. Additionally, in order to compare the relative potencies of the agents across species and routes of exposure it is necessary to generate the basic toxicological endpoints by the other routes. Subsequently, up-and-down studies were performed in order to calculate LD50 data for intravenous and subcutaneous dosing. All experiments described in this manuscript were approved by an investigational Review Board.

Materials and methods Animals Sexually mature male (3- to 4-month old) and female (4- to 5month old) Gottingen minipigs (Sus scrofa) were obtained from Marshall Farms USA (North Rose, NY). Each shipment of pigs contained 10 animals of a single sex, which were utilized before arrival of the next batch of pigs. Male and female pigs were never present at the facility simultaneously. Upon arrival to the facility the minipigs underwent an initial health examination by the attending veterinary staff. The pigs were then quarantined for at least 3 days. After this time the involved research personnel familiarized the pigs to various procedures that included daily handling, change of location within the animal facilities and adaptation to a sling apparatus. Surgeries Surgeries to implant catheters into the external jugular vein were performed on all pigs used in these studies in order to assess the inhibition of whole blood acetylcholinesterase (AChE) and butylcholinesterase (BuChE) activity during the course of the vapor exposures. A complete description of the surgical procedures can be found in previous work (Hulet et al., 2006b). Additionally, a second catheter was inserted into the saphenous vein of the right rear leg only in the pigs for studies used in determining the LD50 for intravenous dosing of GB or GF. At least 3 days were allowed for recovery from the surgical implantation of the catheters before pigs were used for exposures to nerve agent vapor. During that time, the vascular access ports were flushed with heparinized saline as needed. During the exposures, the catheter was maintained by a continuous slow i.v. infusion of lactated Ringers solution and blood samples were withdrawn periodically. The total volume of blood drawn did not exceed 1% of the animal’s body weight over a 1-week span. Sling restraint A sling was used to restrain each minipig during the exposure to the nerve agent vapors. The frame of the sling was constructed of airtight stainless steel pipe and SwagelokÔ fittings. The slings were custom designed (Lomir Biomedical, Inc., Malone, NY) to fit the build and size of the minipigs. The slings were constructed of canvas and fitted to accept the animal through four leg holes. The pig was maintained in the sling by two straps that secured over the pig’s shoulders and hips. A muzzle harness was placed over the animal’s

Inhal Toxicol, 2014; 26(3): 175–184

snout and secured both laterally and ventrally to the stainlesssteel framing in order to prevent the animals from freely moving their head. The harness was fitted so that it did not interfere with the animal’s ability to open its mouth to breath. Chemicals Isopropyl methyl phosphonofluoridate (GB) or Cyclohexyl methylphosphonofluoridate (GF or cyclosarin) was used for all vapor exposures in this study. The chemical agent standard analytical reagent material (CASARM)-grade GB 2035 (lot # GB-U-6814-CTF-N) was verified (usually 98.3 + 0.48 wt.% pure as determined by quantitative 31P-NMR) and stored in sealed ampoules containing nitrogen. Analysis for agent impurities was conducted using acid–base titration as well as Gas Chromatography/Mass spectrometry (GC-MS) and 1H NMR. Acid–base titration identified the following impurities for GB: Compound Methylphosphonofluoridic acid (fluor acid) Diisopropyl methylphosphonate (DIMP) Methylphosphonic difluoride (DF)

Mole %

Calculated wt.%

0.3 0.2 0.2

0.2 0.3 0.2

Additionally, GC-MS positively identified DIMP, Diisopropyl phosphonofluoridate, Tributylamine and Isopropyl ethylphosphonofluoridate, but did not quantify the amounts. Tributylamine was also confirmed using 1H NMR with a concentration of 5 0.1 wt.% of GB. The munitions grade GF (lot # GF-93-0034-147.2) was verified as 98.16 ± 0.36 wt.% pure as determined by quantitative 31P-NMR and stored in sealed ampoules containing nitrogen. Ampoules were opened as needed to prepare external standards or to be used as neat agent for vapor generation. All external standards for GB or GF vapor quantification were prepared on a daily basis. Triethylphosphate (99.9% purity), obtained from Aldrich Chemicals (Milwaukee, WI) was used as the internal standard for the GB and GF purity assays. The following impurities were identified in GF by acid–base titrations: Compound GF acid

Calculated wt.% 1.51 ± 0.13

Vapor generation The vapor generation system is located at the chamber inlet and is contained within a stainless steel glove box maintained under negative pressure. A gas-tight syringe, containing the test material, is secured into a variable rate, pulse-free syringe drive with the material delivered into a spray atomizer. To provide a smaller orifice in the sprayer, the atomizer was modified by retrofitting a syringe needle (SS, 25 gauge, 300 ) into the top of the sprayer. Typically, the syringe was loaded with 2–4 mL of liquid nerve agent. The entire apparatus was contained within a generator box that was mounted at the top of the inhalation chamber. This setup was capable of precisely generating GB or GF vapor over a concentration range of 0.001–2.0 mg/m3.

DOI: 10.3109/08958378.2013.872212

Inhalation chamber Whole body exposures were conducted in a 1000 l dynamic airflow inhalation chamber. The Rochester style chamber is constructed of stainless steel with Glass or Plexiglas windows on each of its six sides. The interior of the exposure chamber was maintained under negative pressure (0.0.25–0.5000 H2O), which was monitored with a calibrated magnehelix (Dwyer, Michigan City, IN). A thermoanemometer (Model 8565, Alnor, Skokie, IL) was used to monitor chamber airflow at the chamber outlet. Physical parameters (chamber airflow, chamber room temperature and relative humidity) were monitored during exposure and recorded approximately every 10 min. Two sampling methods were used to monitor and analyze the GB or GF vapor concentrations in the exposure chamber. The first method was a quantitative technique using a solid sorbent tube system (Tenax/Haysep) to trap GB or GF (Hulet et al., 2006a), followed by thermal desorption and gas chromatographic (GC) analysis (HP Model 6890, Agilent Technology, Baltimore, MD). All samples were drawn from the same area (middle) of the chamber after it had attained equilibrium (t99). Sample flow rates were controlled with calibrated mass flow controllers (Matheson Gas Products, Montgomeryville, PA) and verified before and after sampling by temporarily connecting a calibrated flow meter (DryCalÕ , Bios International, Pompton Plains, NJ) in-line to the sample stream. The solid sorbent tube sampling system was calibrated by direct injection of external standards (GB or GF mg/ml) into the heated sample line of the Dynatherm. In this way, injected nerve agent standards were put through the same sampling and analysis stream as the chamber samples. A linear regression fit (r2 ¼ 0.999) of the standard data was used to compute the GB or GF concentration of each chamber sample. The second method was a continuous monitoring technique using a phosphorus monitor (HYFED, Model PA260 or PH262, Columbia Scientific, Austin, TX). Output from the HYFED provided a continuous strip chart record of the rise, equilibrium and decay of the chamber vapor concentration during an exposure. Cholinesterase analysis Indwelling jugular catheters were implanted into the pigs in order to draw ‘‘real time’’ blood samples to assess cholinesterase inhibition. Blood draws were taken prior to the start of the exposure and at periodic intervals throughout: approximately every 2 min during the 10-min exposure, every 15 min during the 60-min exposure and every 20 min during the 180-min exposure. The total volume of blood drawn did not exceed 1% of the animal’s body weight over a 1-week span. An equivalent volume of Lactated Ringers replaced drawn sample volumes. Assays for AChE and BChE activity were performed on whole blood. About 10 mL of whole blood was added to a disposable borosilicate glass tube (Chase Scientific Glass, Rockwood, TN) containing 2000 mL of distilled water. About 200 mL of 0.69 mM phosphate buffer at pH 7.4 (EQM Research, Cincinnati, OH) was then added to each tube. The tubes were vortexed and allowed to sit at room

GB and GF inhalation toxicology in minipigs

177

temperature for 20 min. About 200 mL of the sample solution from each tube was transferred to two adjacent wells on a 96-well plate. About 25 mL of 30 mM 5,5-dithiobis-2nitrobenzoic acid (DTNB) was added to each well. The plate was covered, and incubated at 37  C for 10 min. For the determination of AChE activity, 25 mL of a solution containing 10 mM acetylthiocholine and 200 mM 10-(adiethylaminopropionyl)-phenothiazine, a specific inhibitor of butyrylcholinesterase (EQM Research, Cincinnati, OH), was added to the appropriate wells of the 96-well plate. For determination of BChE activity, 25 mL of a solution containing 20 mM butyrylthiocholine (EQM Research, Cincinnati, OH) was added to the appropriate wells of the 96-well plate. The plate was shaken briefly to ensure mixing of the reagents and read at 450 nm and 37  C using a SpectraMax Plus384 microplate spectrophotometer (Molecular Devices Corp., Sunnyvale, CA) for 10 min, and analyzed using SoftMax Pro LS version 4.3 software. AChE and BChE activity values were expressed as units of activity per mL of whole blood (U/mL). Intravenous and subcutaneous exposures Prior to beginning the experiments the pigs were weighed and the dosages of GB or GF were calculated on a per body weight basis. Neat GB or GF was diluted in saline to the concentration desired in order to deliver the correct dose of nerve agent. Intravenous injections were delivered, through a second catheter placed in a leg vein, as a single bolus in a total volume of 1 ml. The catheter was then flushed with 3 ml of saline to clear the residual agent from the line. The subcutaneous injections were delivered as a single bolus, in a total volume of 0.5 ml under the loose skin behind the ear. Design and data analysis To determine the progression of experimental exposure concentrations, we used the up-and-down method (Bruce, 1985) with an assumed probit slope of 10 for vapor inhalation exposures and of 15 for intravenous and subcutaneous injections. The binary response used for executing the upand-down method was whether the animal survived for 24 h after the exposure. Additionally, the signs of nerve agent exposure were classified as moderate, severe or lethal. Criteria for classifying an animal as having severe signs were if the animal had any of the following: gasping, prostration, collapse or convulsions. Muscle tremors, salivation, lacrimation or miosis constituted a moderate exposure. The method of maximum likelihood estimation (MLE) (Fox, 1997) was used on the resulting quantal-response data to calculate LCT50 values (and associated asymptotic 95% confidence intervals) and ECT50 values for severe signs for each of the six gender-exposure duration groups. Up-and-down experiments normally use 6–10 animals, which is not enough animals to permit reliable estimation of the probit slope. Data from several up-and-down experiments can be combined to obtain enough animals (at least 30) to estimate the probit slope. The resulting data set can then be analyzed via traditional probit analysis (Finney, 1971) or ordinal logistic regression (Agresti, 1990; Sommerville, 2003; Whalley et al., 2004) to obtain a probit slope estimate.

178

S. W. Hulet et al.

Inhal Toxicol, 2014; 26(3): 175–184

For modeling the response distribution, the following general relationships were used: YN ¼ ðYP  5Þ ¼ k0 þ kC ðlog10 CÞ þ kT ðlog10 TÞ þ ks ðGenderÞ þ kTS ðlog10 TÞðGenderÞ

ð1Þ

YN ¼ ðYP  5Þ ¼ k0 þ kC ðlog10 CÞ þ kT ðTimeÞ þ ks ðGenderÞ þ kTS ðTimeÞðGenderÞ ð2Þ where YN is a normit, YP is a probit, the k’s are fitted coefficients, C is vapor concentration, and both T and Time represent the exposure duration. In Equation (1), exposure duration is treated as a covariate (T), whereas in Equation (2), exposure duration is treated as a three-level factor (Time). The constant, kTS, has six values, one for each time–gender combination. The constants kC and kT are the probit slopes for concentration and time, respectively. The toxic load exponent, n, is the ratio kC/kT. If this ratio is not different (with statistical significance) from one, then Haber’s rule is appropriate for modeling the toxicity. Otherwise, the toxic load model (C nT) is the proper approach, assuming that there is no significant curvature in the data used to fit the model. Should significant curvature exist, then the toxic load model is not appropriate, but it is still superior to Haber’s rule in modeling the data. The present protocol has exposure durations of 10, 60 and 180 min. For each of the exposure durations, six or seven pigs of each gender were used. Statistical analysis routines contained within MinitabÕ versions 13 and 14 (Minitab, Inc., State College, PA), as well as an in-house developed spreadsheet program, were used for the analysis of the data.

Results Animals Thirty-eight pigs (19 male and 19 female) were exposed to concentrations of GB vapor to estimate LCT50 and ECT50 (severe) values. An additional male pig was used as an air control. At the time of the surgeries, the 20 males (19 experimental plus 1 control), weighed an average of 10.68 ± 0.26 (SEM) kg and the 19 females, weighed an average of 10.62 ± 0.21 (SEM) kg. Forty-two pigs (19 males and 23 females) were exposed to concentrations of GF vapor to estimate LCT50 and ECT50 (severe) values. At the time of the pigs’ surgeries, the 19 males, on average, weighed 10.28 ± 0.25 (SEM) kg and the 23 females, on average, weighed 10.10 ± 0.25 (SEM) kg. Additionally 13 male pigs were used in up-and-down studies to calculate LD50 values for GF exposures via intravenous and subcutaneous exposures. At the time of the pigs’ surgeries the 13 males, on average, weighed 8.08 ± 0.20 (SEM) kg. About 10 male pigs were used in up-and-down studies to calculate LD50 values for GB exposures via intravenous and subcutaneous exposures. At the time of the pigs’ surgeries the 10 males, on average, weighed 9.63 ± 0.22 (SEM) kg. Inhibition of cholinesterase activity Depression of cholinesterase (AChE and BChE) activity was assessed during the GB and GF vapor exposures by collecting

blood specimens through the jugular catheter. Of the 38 pigs that were exposed to GB (19 males and 19 females) blood samples could not be collected from three males and two females due to the loss of patency in the catheter between the day of surgery and the day of the experiment. For male pig no. 58, samples could only be collected after exposure had been concluded and the pig had been removed from the exposure chamber. AChE values were decreased to 8% or less of the baseline values before the conclusion of the exposures in 31 out of 32 pigs. There was very little subsequent depression in AChE activity after the conclusion of the exposures. In fact, AChE depression in 22 of 32 pigs had reached the absolute lowest value during the course of the exposure. Depression of BChE was variable but dropped below 50% of baseline in only one of the pigs. Of the 42 pigs that were exposed to vapor GF (19 males and 23 females) we were unable to collect blood samples during the exposure on 5 males and 2 females due to the catheters losing patency between the day of surgery and the day of the experiment. Whole blood acetylcholinesterase values dropped quickly in all GF exposed pigs regardless of exposure concentration or durations. Absolute values of whole blood AChE had dropped to less than 9% of baseline values in every pig exposed by the end of the GF exposure duration. Whole blood BChE depression varied, but no pig’s BChE dropped below 65% of baseline values. Median effective (severe) and median lethal dosages for inhalation exposures The results of the exposures were classified as moderate, severe or lethal (see ‘‘Methods’’ section for a description of criteria). The method of maximum likelihood estimation (MLE) was used to calculate ECT50 (severe) and LCT50 values (and associated asymptotic 95% confidence intervals) for each of the six gender-exposure duration groups. LC50 and LCT50 values (with their respective 95% confidence intervals) for both GB and GF vapor exposures can be found in Table 1 and are plotted in Figures 1 and 2. The LCT50 values are not constant over time. After calculating the MLE for lethality it was determined that pig #63 was statistically considered to be an outlier; pig #63 was exposed for 60 min to a total CT of 89.4 mg min/m3 and died from the exposure. The calculated LCT50 MLE for 60-min exposures was 127.1 mg min/m3 with 95% Wald limits of 98.5–163.9. Therefore the CT at which pig 63 died (89.4 mg min/m3) was statistically outside the 95% limits. Subsequently the toxic load model fits were run both with and without pig #63 included in the data set. For GF exposures, 16 out of 42 exposures resulted in fatalities. Only one animal that showed signs of severe effects of exposure survived for 24 h. Because of the lack of animals classified as severe, we were unable to calculate ECT50 (severe) values for GF. The EC50 and ECT50 (severe) values for GB vapor exposures, with their respective 95% confidence intervals, can be found in Table 1C. The ECT50 values, like LCT50 values, are not constant over time. The ratio of ECT50 (severe) values to LCT50 values for GB exposures are shown in Table 2. This ratio was statistically higher in female pigs (99% confidence) as compared to male pigs.

GB and GF inhalation toxicology in minipigs

DOI: 10.3109/08958378.2013.872212

Table 1. MLE for median lethal and median effective concentrations and dosages (with 95% confidence intervals on the dosages) for whole body vapor GB and GF exposures. We were unable to calculate severe effects for the GF exposures due to a lack of surviving animals that exhibited severe signs. Males Exposure-duration (min)

Females

LC50

LCT50

95% limits

LC50

LCT50

95% limits

(A) Lethality (GB) 10 60 180

7.25 1.76 1.01

72.5 105.7 182.3

55.1–95.2 83.7–133.5 140.6–236.3

8.69 2.12 0.97

86.9 127.1 174.2

67.3–112.3 98.5–163.9 129.4–234.7

(B) Lethality (GF) 10 60 180

21.8 4.78 2.24

218 286.8 403.2

163–292 224–366 306–529

18.3 4.70 2.03

183 282 365.4

133–251 214–371 261–511

EC50

ECT50

95% limits

EC50

ECT50

95% limits

5.15 1.38 0.74

51.5 83.0 134.0

36.9–71.9 62.6–110.0 97.6–182.3

7.74 1.88 0.81

77.4 112.5 145.9

60.5–99.0 86.6–146.0 108.4–196.5

(C) Severe effects (GB)

Figure 1. Toxic load (TL) model fits of MLE LCT50 estimates for whole body vapor GB exposures in male and female minipigs.

250 Dosage (Concentration-Time) (mg-min/m3)

10 60 180

Female MLE (Lethal) Male MLE (Lethal) TL Fit (L5) Lethal (Male) TL Fit (L5) Lethal (Female)

200 150

100 90 80 70 60 50 40 35 30

10

100 Exposure Time (minutes)

600 Dosage (Concentration-Time) (mg-min/m3)

Figure 2. Toxic load (TL) model fit of MLE LCT50 estimate for whole body vapor GF exposures in male and female minipigs.

Female MLE (Lethal) Male MLE (Lethal) TL Fit Lethal

500 400 350 300 250 200

150

100

10

100 Exposure Time (minutes)

179

180

S. W. Hulet et al.

Inhal Toxicol, 2014; 26(3): 175–184

Table 2. Ratios of ECT50 (severe) values to LCT50 values for GB vapor exposures. The ratio of severe to lethal concentrations was statistically higher in female pigs (99% confidence). Duration (min) 10 60 180

Gender

Severe/lethal

Gender

Severe/lethal

Male Male Male

0.711 0.785 0.735

Female Female Female

0.891 0.885 0.837

Table 3. Comparison of the potencies of LCT50 values for GB and GF whole body vapor exposures in the minipig. GF/GB potency Duration (min) 10 60 180

Males

Females

3.03 2.70 2.22

2.13 2.70 2.08

Gender differences The models were tested for possible gender effects and sex was tentatively found to be a significant term for lethal GB vapor exposures. However, this significant difference was fully dependent on the exclusion of the one outlying pig from the data set. When pig 63 was excluded from the analysis, male minipigs were statistically more sensitive (p ¼ 0.013) than females. Gender was not a statistically significant term (p ¼ 0.067) when pig 63 was included in the analysis. The gender term was not statistically significant for any of the models where exposure-duration was treated as a covariate. The interaction of Sex with Time or Sex with logT was not statistically significant, regardless of the inclusion of pig 63. The failure to find statistically significant differences between the interactions of gender and exposure-duration may have been due to the low sample size (n ¼ 6–7). In contrast, when gender was considered, regardless of exposure-duration, the sample size was much larger (n ¼ 38). There were no statistically significant differences observed between the genders in the pigs exposed to GF in the current studies. Statistical models for the probability of lethality In order to model the probability of lethality as a function of exposure concentration, exposure duration and gender, several models were fit to the generated quantal data and ordinal regression was used to fit various response models. The number of pigs used per gender-exposure duration group was not large enough to estimate the response distribution per each group. Instead, the response distribution was estimated using Equations (1) or (2) with the data for all 38 pigs exposed to GB and all 42 pigs exposed to GF. GB exposures For GB vapor exposures, Pig 63 was statistically considered to be an outlier and therefore all of the toxic load model fits were tested with and without pig 63. The recommended best toxic load model fit did not include pig 63: Yn ¼ constant þ 12:4 log10 ðC Þ þ 9:0 log10 ðT Þ  0:605 Sex where the constant depends on the effect (severe or lethal) and Sex is coded as 1 for male and 1 for female. The value of the toxic load exponent (n ¼ kC/kT) was essentially independent of the model used. The toxic load exponent was 1.38 with a 95% confidence interval of 1.24–1.52. Because this interval did not overlap one, Haber’s rule was not considered an appropriate time dependence model for this data set. Potential curvature in the data was evaluated by inserting a (logT)2 term into the model. This term was found to be statistically insignificant.

For executing the up and down method in this study, the probit slope on concentration, kC, was assumed to be 10. The probit slope of the best model fit was 12.4 with a 95% range of 6.2–18.6. GF exposures For the GF vapor exposures no gender term was included in the model fits because gender was found to be statistically insignificant. For GF vapor exposures the recommended best model fit was: Yn ¼ constant þ 10:9 log10 ðC Þ þ 8:5 log10 ðT Þ where the constant depends on the effect (severe effect or lethality). The value of the toxic load exponent (n ¼ kC/kT) was essentially independent of the model used. The toxic load exponent was 1.28 with a 95% confidence interval of 1.12– 1.44. Because this interval does not overlap one, Haber’s rule is not an appropriate time dependence model for this data set. Potential curvature in the data was evaluated by inserting a (logT)2 term into the model and this term was found to be statistically insignificant. The probit slope of the best model fit was 11 with a SE of 3.4. GF/GB potency ratios GF was less potent than GB by the inhalation route regardless of exposure duration or gender (Table 3). The ratio of GF to GB LCT50 values ranged from 2.08 to 3.03 depending on duration of exposure. GF was less potent than GB by both the intravenous and subcutaneous routes of exposure. The GF to GB potency ratio was found to equal 1.57 (with 95% confidence limits of 1.25–1.97) when the total data set [data from both agents and routes of exposure (IV and SC), and using two toxicological endpoints] is considered. Median lethal dosages for intravenous and subcutaneous exposures LD50 values for intravenous and subcutaneous GB and GF exposures are shown in Table 4. In contrast to the GF inhalation studies, we were able to calculate ED50 (severe) values for both the GF intravenous and subcutaneous exposures. For GB, the LD50 for intravenous injections was 16.1 mg/kg with Wald limits of 11.2–23.1 mg/kg, respectively. The ED50 (severe) for GB exposures after intravenous injections was 10.0 mg/kg with Wald limits of 7.5–13.5 mg/kg. For GF, the LD50 for intravenous injections was 21.9 mg/kg with Wald limits of 17.9–26.7 mg/kg, respectively. The ED50 (severe) for GF exposures after intravenous injections was 18.4 mg/kg with

GB and GF inhalation toxicology in minipigs

DOI: 10.3109/08958378.2013.872212

Table 4. Maximum likelihood estimates (with 95% confidence limits) of ED50 (severe) and LD50 values for intravenous and subcutaneous injections of GF in minipigs.

Agent GB GF

Route Intravenous Subcutaneous Intravenous Subcutaneous

LD50 95% limits ED50 (severe) 95% limits (mg/kg) (mg/kg) (mg/kg) (mg/kg) 16.1 36.8 21.9 43.6

11.2–23.1 26.2–51.6 17.9–26.7 36.1–52.6

10.0 18.9 18.4 31.4

7.5–13.5 8.8–40.6 15.6–21.7 25.1–39.3

Wald limits of 15.6–21.7 mg/kg. The ED50 to LD50 ratio for the intravenous exposures to GF was 0.84. The LD50 for subcutaneous injection of GB was 36.8 mg/kg with Wald limits of 26.2–51.6 mg/kg, respectively. The ED50 for severe exposures was 18.9 mg/kg with Wald limits of 8.8–40.6 mg/kg. The ED50 to LD50 ratio for the subcutaneous exposures to GB was 0.51. The LD50 for subcutaneous injection of GF was 43.6 mg/kg with Wald limits of 36.1–52.6 mg/kg, respectively. The ED50 for severe exposures was 31.4 mg/kg with Wald limits of 25.1–39.3 mg/kg. The ED50 to LD50 ratio for the subcutaneous exposures to GF was 0.72.

Discussion Cholinesterase inhibition Whole blood AChE activity dropped to 8% or less of baseline value by the end of the exposure duration in 97% (31 of 32) of pigs that had patent catheters during the vapor GB exposures. Despite the low levels of AChE activity, 16 of the 32 pigs survived. Similarly, whole blood AChE had dropped to 59% of baseline values in every pig (n ¼ 35) that had patent catheters during the vapor GF exposures and 22 of 35 pigs survived. Grob & Harvey (1958) identified as early as 1958 that red blood cell cholinesterase activity could be depressed in humans to near zero (with multiple low dose exposures) without resultant death. The current study supports previous ones that there is a poor correlation between absolute values of AChE activity and predictability of lethality. LCT50s The calculated GB and GF LCT50 values for pigs in the current studies are consistent with the notion that larger mammals (pigs, dogs, cats, monkeys) have lower threshold values than smaller animals (mice, rats, rabbits). Larger mammals are also more reflective of estimated LCT50 values in humans. For instance, the 10-min LCT50 values for vapor GB exposures in mice, male rats and rabbits are 380, 231 and 115 mgmin/m3, respectively (McGrath & Fuhr, 1948; Mioduszewski et al., 2002b). Among mammals, the calculated LCT50s for 10-min GB exposures in monkeys and male cats are 74 mg min/m3 and 79 mg min/m3, respectively (Cresthull et al., 1957; McGrath & Oberst, 1952). Crook et al. (1952) calculated the LCT50 for a 10-min GB vapor exposure in pigs to be 34 mg min/m3. The LCT50s for 10-min GB exposures in male and female pigs in the current study were 72.5 and 86.9 mg min/m3, respectively.

181

The 10-min LCT50 values for GF in mice (McGrath et al., 1953) and guinea pigs (Muir et al., 1949) are 280 and 165 mg min/m3, respectively. Muir et al. (1949) reported a GF LCT50 value of 249 mg min/m3 in rats. More recently, Anthony et al. (2004) reported lethality values of 253 and 371 mg min/m3, respectively, for female and male rats. In comparison, the calculated LCT50 for 10-min GF exposures in monkeys (Cresthull et al., 1957) was 130 mg min/m3. The LCT50’s for 10-min GF exposures in male and female pigs in the current study were 218 and 183 mg min/m3, respectively. ECT50 (severe) The LCT50 reported for a 10-min GB vapor exposure in monkeys was 74 (62–87 F.I.) mg min/m3 (Cresthull et al., 1957). In the study, monkeys were categorized with the toxicological signs of collapse and/or convulsions as ‘‘incapacitated’’ and then ICT50 (incapacitation) values for the 10-min GB and GF exposures were calculated. For GB the ratio of the ICT50 to LCT50 was 0.89. For GF the ICT50 to LCT50 ratio was 0.77. In the current studies, we utilized the toxicological signs of collapse and/or convulsions (along with prostration and gasping) to characterize a pig as having ‘‘severe’’ signs of exposure. When GB was the nerve agent the ratio of ECT50 (severe) to LCT50 values in male pigs for 10-, 60- and 180-min exposures were 0.71, 0.79 and 0.74, respectively. The ratio of ECT50 (severe) to LCT50 for female pigs for 10-, 60- and 180-min exposures were 0.89, 0.89 and 0.84, respectively. Statistically, the ratios of severe to lethal concentrations for vapor GB exposures were higher in female pigs (99% ANOVA confidence) than in male pigs. The data suggest that there is less difference between severely toxic and lethal dosages for vapor GB exposures in female as compared to male pigs. While the Cresthull study used both male and female monkeys, the breakdown of sexes only stated that ‘‘most of them were females’’ limiting the ability to compare the two studies in order to distinguish whether statistical gender differences exist in the severe to lethal ratios of other species. The current study is most likely the first to identify gender differences between the ratios of severe to lethal effects for GB vapor exposures. Unfortunately, we were unable to calculate ECT50 (severe) values for the vapor GF exposures due to only one surviving pig with severe signs. Crook et al. (1952) determined that 87% of the pigs that died from 10-min GB vapor exposures did so either during exposure or within the first 10 min after the conclusion of the exposure. In the current study, 83% of the pigs that died from the 10-min exposure to GB did so within the first 10 min after exposure. Additionally, 72% of the pigs that died, regardless of the GB exposure duration, did so either during the exposure or within the first 10 min after the exposure. These data support that the toxic actions of GB occur because of the inhalation of GB vapor rather than the delayed absorption of GB through the skin. Indeed it has been demonstrated in open air testing that GB has minimal effectiveness in humans as a nerve agent via the percutaneous route of exposure (Marzulli & Willliams, 1953). Similarly, 88% (14/16) of pigs that died from GF exposures did so during the exposure or within the

182

S. W. Hulet et al.

first 10 min after the exposures. The GF exposures to the other two pigs proved fatal within 1 h from the end of the exposure. Toxic load models In the current study, the toxic load exponents for GB and GF lethal vapor exposures were 1.38 (with a 95% confidence interval of 1.24–1.52) and 1.28 (with a 95% confidence interval of 1.12–1.44), respectively. Bide et al. (2004) suggested an estimate for a 10-min GB vapor exposure of 57 mg min/m3 in humans. This estimate was extrapolated based on data taken from 38 historical animal studies involving 7 species (none being swine), regardless of gender. The overview took into account the minute volume (MV) to body weight (BW) ratio for each of the species. The calculated MV/BW ratio for humans was 0.223. The next closest MV/BW ratio of species used in the study was 0.328 for dogs. The MV/BW ratio of pigs (Denac et al., 1977) is 0.225. The calculated toxic load exponent for his large data set was 1.38. Interestingly, the calculated toxic load exponent for GB exposures in current study was also 1.38. There is no significant statistical difference between the toxic load exponents of GB and GF suggesting that regardless of the nerve agent used (GB or GF) the effect of concentration on toxicity over time is the not significantly different. Additionally, the confidence intervals for both nerve agents do not overlap one supporting that Haber’s rule is not an appropriate time dependence model for these data sets. Intravenous and subcutaneous LD50s For GB, the LD50 for intravenous injections was 16.1 mg/kg with Wald limits of 11.2–23.1 mg/kg, respectively. This value is consistent with the previously published value of 15 mg/kg in the pig (U.S. Department of the Army, 1974). Intravenously, GF has been reported to be approximately equipotent to GB in rabbits (Marrazzi et al., 1951; Marzulli et al., 1955) and mice (Manthei et al., 1996). In the current study, GF was less potent than GB with a GF to GB potency ratio of 1.39. By the subcutaneous route GF has been identified as less potent than GB in a number of species. The GF to GB potency ratios for lethality in rats (Shih & McDonough, 1999), mice (Clement, 1992, 1994) and guinea pigs (Shih & McDonough, 1999) were 1.68, 1.42 and 1.36, respectively. In the current study, the GF to GB potency ratio for lethality (LD50) by subcutaneous injection in minipigs was 1.20. In contrast to the subcutaneous and intravenous routes, GF has traditionally been considered to be equipotent or more potent than GB by the vapor inhalation route. Indeed, the GF to GB potency ratios in early inhalation studies in rodents (mice, guinea pigs, rats) ranged from 0.74 to 0.92 (Callaway & Blackburn, 1954; McGrath et al., 1953; Muir et al., 1949). More recent studies in rodents (Anthony et al., 2004) lend further support to this notion. In Anthony’s study, the potency ratio drastically changes depending on the duration of exposure; with the two agents being approximately equipotent over a 10-min exposure but when the duration of exposure increases to 240-min GF becomes approximately twice as potent as GB. In contrast, in the current study GF was 2–3

Inhal Toxicol, 2014; 26(3): 175–184

times less potent than GB via the inhalation route of exposure regardless of exposure duration. The only other animal species in which GF has been identified as being less potent than GB is in monkeys (Cresthull et al., 1957). Considering the higher order of these species in comparison to rodents one could logically assume that GF may be less potent than GB via the inhalation route in humans. Yet if hazard estimates rely heavily on data generated from lower order species (rodents) or routes of exposure other than inhalation there is a risk of overestimating (or underestimating) the toxicity posed. Indeed, even the most recent Army toxicity estimates (Army Field Manual 3-11.9, 2005) suggest that GF is equipotent or more potent than GB. These findings highlight the importance of the choice of animal model and exposure route in investigating nerve agent toxicity and the difficulties associated with extrapolating data obtained from one species and exposure route to data obtained by another species and exposure route. Gender differences In 1998, a recommended change to airborne occupational exposure limits suggested that the Immediately Dangerous to Life or Health (IDLH) exposure guidelines for inhaled GB be lowered (Mioduszewski et al., 1998). This suggestion was made to correct the failure of the existing guidelines to take into account that there may be differences in sensitivity to nerve agents based on gender. The existing value at the time for a 30-min exposure to GB of 0.2 mg/m3 was lowered to 0.1 mg/m3. This suggestion was made based on work done by Callaway & Blackburn (1954) in which female rats were found to be as much as twice as sensitive to the lethal effects of inhaled GB than male rats. The significantly greater sensitivity to inhaled GB in female rats has subsequently been shown to occur over longer (240 min) durations of exposure as well (Mioduszewski et al., 2002a,b). The female hamster has also been identified as being more susceptible to GB vapor exposure than its male counterpart (McPhail, 1953). In contrast, male mice have been identified as being significantly more sensitive than female mice to GB vapor via inhalation (McPhail, 1953) and intravenous administration (DeCandole & McPhail, 1957). Given that there are no relevant human data available and there is a surprising lack of literature investigating gender differences in sensitivity to inhaled GB in higher species (cat, dog, pig, monkey), the best possible course of action is to base human estimates on available data, the majority of which are derived from rodents. The current study has identified that male pigs are significantly more sensitive to inhaled GB than female pigs. However, this finding is tentative at best because of its dependence on the exclusion of one animal from the data set. Regardless of the inclusion of this animal it is clear that in minipigs, unlike in rodents, females are not more sensitive to GB vapor than are males. There were also no significant gender differences found for vapor GF exposures. While these findings are not, by themselves, enough to suggest that current human estimates be revised, gender differences in a species that more closely reflects human toxicity estimates warrants consideration, if not perhaps priority, in deriving such estimates.

DOI: 10.3109/08958378.2013.872212

Conclusions The current studies were conducted with the intent of estimating lethal concentrations of the nerve agents sarin (GB) and cyclosarin (GF) as a function of exposure duration in the Gottingen minipig. Ordinal regression was used to fit various response models to the data. LCT50 values were calculated in male and female pigs exposed to GB or GF vapor for 10, 60 and 180 min. ECT50 (severe) were calculated in male and female pigs exposed to GB vapor for the same durations of exposure. The ECT50 values (severe) were 71–79% of the LCT50 values in male pigs and 84–89% of the LCT50 values in female pigs. The ratios of severe to lethal concentrations were higher in female minipigs (99% ANOVA confidence) indicating that there is less difference between severely toxic and lethal dosages in the female as compared to male pigs. We were unable to calculate ECT50 values (severe) for the GF inhalation studies due to only one animal with severe signs of exposure surviving The values of the toxic load exponents were essentially independent of the model used for both nerve agents. The toxic load exponent of the best-fit model for GB exposures was 1.38 (with a 95% confidence interval of 1.24–1.52). The toxic load exponent of the best-fit model for GF exposures was 1.28 (with a 95% confidence interval of 1.12–1.44). Because neither of these intervals overlaps one, Haber’s rule is not an appropriate time dependence model for this data set. Potential curvature in the data was evaluated by inserting a (logT)2 term into the model and this term was found to be statistically insignificant for both nerve agents. For GB exposures the probit slope of the best model fit was 12.4 with a 95% range of 6.2–18.6. For GF exposures, the probit slope of the best model fit was 11 with a SE of 3.4. The models were tested for possible gender effects and Sex was found to be a significant term (p ¼ 0.013) for GB vapor exposures, with males being significantly more sensitive than females. Gender was not found to be a significant term for GF vapor exposures. For GB exposures, the LD50 values for intravenous and subcutaneous injections in male pigs were 16.1 mg/kg (11.2–23.1 mg/kg) and 36.8 mg/kg (26.2–51.6 mg/kg), respectively. The ED50 values (severe) for the intravenous and subcutaneous injections were 62% and 51% of the LD50 values in male pigs. For GF exposures the LD50 values for intravenous and subcutaneous injections in male pigs were 21.9 mg/kg (17.9–26.7 mg/kg) and 43.6 mg/kg (36.1–52.6 mg/ kg), respectively. The ED50 values (severe) for the intravenous and subcutaneous injections were 84% and 72% of the LD50 values in male pigs.

Acknowledgements The authors thank Drs. Christopher Whalley, Robert Mioduszewski and Edward Jakubowski for their help in editorial review of the manuscript.

Declaration of interest The authors would like to thank the Dr. Cherrie ReutterChristy for her role in coordinated funding for the low level

GB and GF inhalation toxicology in minipigs

183

toxicology program and the Defense Threat Reduction Agency for providing funding for the studies.

References Agresti A. (1990). Categorical data analysis. New York, NY: John Wiley & Sons. Anthony JS, Haley M, Manthei J, et al. (2004). Inhalation toxicity of cyclosarin (GF) vapor in rats as a function of exposure concentration and duration: potency comparison to sarin (GB). Inhal Toxicol 16: 103–11. Army Field Manual 3-11.9. January 2005. Potential military chemical/ biological agents and compounds. Army Chemical School. Fort Leonard Wood, MO. DISTRIBUTION UNLIMITED. Bide RW, Armour SJ, Yee E. (2004) Estimation of human toxicity from animal toxicity data: GB toxicity reassessed using newer techniques for estimation of human toxicity. DRDC Suffield Technical Report, 167. DISTRIBUTION UNLIMITED. Bruce RD. (1985). An up-and-down procedure for acute toxicity testing. Fundam Appl Toxicol 5:151–7. Callaway S, Blackburn JW. (1954) A comparative assessment of the vapour toxicities of GB, GD, GF, T.2317 and T.2146 to male and female rats; Porton Technical Paper No. 404. DISTRIBUTION LIMITED. Clement JG. (1992). Efficacy of various oximes against GF poisoning in mice. Arch Toxicol 66:143–4. Clement JG. (1994). Toxicity of the combined nerve agents GB/GF in mice: efficacy of atropine and various oximes as antidotes. Arch Toxicol 68:64–6. Cresthull P, Koon WS, McGrath FP, Oberst FW. (1957). Inhalation effects (incapacitation and mortality) for monkeys exposed to GA, GB, and GF vapors. CWLR #2179. DISTRIBUTION LIMITED. Crook JW, Koon WS, McGarth FP, Oberst FW. (1952). Acute inhalation toxicity of GB vapors to pigs exposed for ten minutes. MLRR, 150. DISTRIBUTION LIMITED. DeCandole CA, McPhail MK. (1957). Sarin and paraoxon antagonism in different species. Can J Biochem Physiol 35:1071–83. Denac M, Sporri H, Beglinger R. (1977). The Gottingen minipig as a laboratory animal. Resp Exp Med 170:283–8. Finney DJ. (1971). Probit analysis. 3rd ed. Cambridge: Cambridge University Press. Fox J. (1997). Applied regression analysis, linear models, and related methods. Thousand Oaks, CA: Sage Publications, Inc. Grob D, Harvey JC. (1958). Effects in man of the anticholinesterase compound sarin (isopropyl methyl phosphonofluoridate). J Clin Investig 37:350–68. Haber FR. (1924). Zur Geschichte des Gaskrieges. In: Funf Vortrage aus Jahren, 1920–1923. Berlin, Germany: Springer, 76–92. Hulet SW, Sommerville DR, Crosier RB, et al. (2006a). Comparison of low-level sarin and cyclosarin vapor exposure on pupil size of the Gottingen minipig: effects of exposure concentration and duration. Inhal Toxicol 18:143–53. Hulet SW, Sommerville DR, Crosier RB, et al. (2006b). Low-level sarin (GB) vapor exposure in the Gottingen minipig: effect of exposure concentration and duration on pupil size. US Government Technical Report ECBC-TR-450. DISTRIBUTION UNLIMITED. Manthei JH, Muller AJ, Heitkamp DH, et al. (1996). Synthesis and acute toxicity of EA 5988. US Government Technical Report ERDECTR-316. DISTRIBUTION LIMITED. Marrazzi AS, Horton RG, King EA. (1951). Correlation between G agent toxicity and chemical structure: an interim report. Medical Division Research Report 37. DISTRIBUTION LIMITED. Marzulli FN, Conn LW, Cope OB, Wiles JS. (1955). Objective signs of poisoning in rabbits following cutaneous application and intravenous injection of GB. MLRR 408. DISTRIBUTION LIMITED. Marzulli FN, Williams MR. (1953). Studies on the evaporation, retention and penetration of GB applied to intact human and intact and abraded rabbit skin. MLRR 199. DISTRIBUTION LIMITED. McGrath FP, Fuhr I. (1948). LC50 of GB to pigeons, rabbits, rats and mice. MDR 140. DISTRIBUTION LIMITED. McGrath FP, Oberst FW. (1952). Acute inhalation toxicity of GA and GB vapors to cats exposed for ten minutes. MLRR 136. DISTRIBUTION LIMITED.

184

S. W. Hulet et al.

McGrath FP, von Berg VJ, Oberst FW, et al. (1953). Toxicity and perception of GF vapor. Report No. 185; Chemical Corps Medical Laboratories, Army Chemical Center: Edgewood, MD. DISTRIBUTION LIMITED. McPhail MK. (1953). Sex and the response to G agents; Suffield Technical Paper No. 38. DISTRIBUTION LIMITED. Mioduszewski RJ, Manthei JA, Way RA, et al. (2002a). Low-level sarin vapor exposure in rats: effect of exposure concentration and duration on pupil size. ECBC-TR-235. US Army Edgewood Chemical Biological Center, Aberdeen Proving Ground, MD. DISTRIBUTION UNLIMITED. Mioduszewski RJ, Manthei JA, Way RA, et al. (2002b). Interaction of exposure concentration and duration in determining acute toxic effects of sarin vapor in rats. Toxicol Sci 66:176–84. Mioduszewski RJ, Reutter SA, Miller LL, et al. (1998). Evaluation of airborne exposure limits for G-agents: occupational and general population exposure criteria. ERDEC-TR-489, US Army Edgewood Chemical Biological Center: Aberdeen Proving Ground, MD. DISTRIBUTION UNLIMITED. Muir A, Callaway S, Burgess F. (1949). The toxicity of the G compounds. Part VIII. Studies on the toxicity of GFT.2139. Porton Technical Paper No. 130. DISTRIBUTION LIMITED.

Inhal Toxicol, 2014; 26(3): 175–184

Shih T-M, McDonough JH. (1999). Organophosphorous nerve agentsinduced seizures and efficacy of atropine sulfate as anticonvulsant treatment. Pharm Biochem Behav 64:147–53. Smith CP (ed.). (2000). Information Resources on Swine in Biomedical Research 19902000. Beltsville, MD: United States Department of Agriculture, USDA. Available from: http://www.nal.usda.gov/awic/ pubs/swine/. Sommerville DR. (2003). Relationship between the dose response curves for lethality and severe affects for chemical warfare nerve agents. Proceedings of the 2003 Joint Service Scientific Conference on Chemical & Biological Defense Research, 17–20 November 2003. US Army Edgewood Chemical Biological Center, Aberdeen Proving Ground, MD. ten Berge WF, Zwart A, Appelman LM. (1986). Concentration-time morality response relationship of irritant and systematically acting vapours and gases. J Hazard Mater 13:301–9. U.S. Department of the Army. (1974). Chemical agent data sheets, vol. 1. Edgewood Arsenal Special Report, EO-SR 74001. Alexandria, VA: Defense Technical Information Center. DISTRIBUTION LIMITED. Whalley CE, Benton BJ, Manthei JH, et al. (2004). Low-level cyclosarin (GF) vapor exposure in rats: effect of exposure concentration and duration on pupil size. ECBC-TR-407. US Army Edgewood Chemical Biological Center, Aberdeen Proving Ground, MD. DISTRIBUTION UNLIMITED.

Copyright of Inhalation Toxicology is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.