- Email: [email protected]
Assessing pyridostigmine efficacy by response surface modeling
FUNDAMENTAL
AND
APPLIED
TOXICOLLIGY
5,
Assessing Pyridostigmine
S242-S25 I (1985)
Efficacy by Response
D. E. JONES,* W. H. CARTER,
Surface Modeling
JR.,? AND R. A. CARCHMAN~
*U.S. Army Medical Research Institute of Chemical Defense9Drug Assessment Division, Aberdeen Proving Ground, Maryland 21010; and tDepartments of Biostatistics; Pharmacology and Toxicolofl, Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia 23298 Assessing Pyridostigmine Efficacy by Response Surface Modeling. JONES, D. E., CARTER, AND CARCHMAN, R. A. (1985). Fur&m. Appl. Toxicol. 5,5242-S25 1. The therapeutic efficacy of atropine sulfate/pralidoxime chloride (ATR/Z-PAM) treatment (im) therapy, and pyridostigrnine bromide (PYR) pretreatment (oral) therapy were evaluated in soman-challenged guinea pigs. ATR/Z-PAM efficacy was assessedas protective ratio (PR = treated soman LDSO/control soman LD50); PYR efficacy was assessedboth as PR and by response surface modeling (RSM) techniques. The optimal ATR/2-PAM treatment gave a PR of 3.78. PYR pretreatment (1 hr) produced a dose&&dependent (r = 0.96) inhibition of whole blood AChE and afforded significant (p < 0.05) increases in PR (with dosesgreater than 0.12 mg/kg PYR) against soman when followed by 64 mg/kg ATR/lOO mg/kg 2-PAM treatment. These PRs, however, were poorly correlated (r = 0.45) with the corresponding level of PYR-induced AChE inhibition. In contrast, RSM analysis of efficacyindicated that the optimal ATR/2-PAM dose combination varied as a function of both the soman-challenge level and the PYR pretreatment dose. Efficacy was therefore evaluated for varying PYR pretreatment doses in combination with the appropriate optimal ATR/ZPAM treatment (as determined by RSM for each soman challenge dose and PYR dose evaluated). When assessedin this manner, PYR efficacy(PYR) was found to be highly correlated (r = 0.97) with PYR-induced AChE inhibition. Since percentage of AChE inhibition was directly correlated with PYR dose (log), these results indicate that PYR pretreatment efficacy is a highly correlated, dose-dependent phenomenon, providing ATR/Z-PAM treatment is optimized. o 1985 say or W. H., JR.,
In this study, the use of response surface modeling (RSM) and protective ratio (PR) methods for evaluating pyridostigmine bromide (PYR) pretreatment (in conjunction with atropine sulfate/pralidoxime chloride [ATR/ZPAM] treatment) were examined for treatment of soman intoxication. RSM provides an improvement over currently available methods (Berry et al., 197 1; Boskovic et al., 1984; Boskovic and Stern, 1970; Clement and Lockwood, 1982; Dirnhuber et al., 1979; and Faff et al., 1976) and a scientifically valid procedure for optimizing therapy in this complex (fivedimensional) study. MATERIALS
AND
METHODS
Animals. Mixed-sex, 300 to 400-g Duncan Hartley Albino guinea pigs from Charles River Breeding Laboratories 0272-0590185 $3.00 Copyright 0 1985 by the Society of Toxicology. All rights of rxprcduction in any form reserved.
were used in all experiments. Containment and test facilities were maintained at constant temperature (2 I + 3”C), humidity (55 + 7%), and lighting (12-hr light/dark cycle). Nerve agent. Soman (GD, pinacolylmethylphosphonofluoridate) was prepared by the Chemical Research and Development Center, Aberdeen Proving Ground, Maryland. Purity was determined (by nuclear magnetic resonance) to exceed 95% for each GD lot. GD challenge doses were prepared in sterile isotonic saline and injected sc in the dorsal cervical area at a dose volume of 1.0 ml/k. Therapy compounds. ATR was purchased from C. H. Boehringer and Sons, Ingelheim, West Germany (Lot No. 352). 2-PAM was purchased from Ayerst Laboratories, New York, New York (Lot No. Z-799). ATR and 2-PAM concentrations (both single and combination therapies) were prepared in diititled, deionized water and injected im (rear limb muscle mass) at a dose volume of I.0 ml/ kg, given 1.Omin postGD challenge. Combination therapy injections of 2-PAM and ATR were admixed and administered as a single therapy injection. Pretreatment. PYR was diluted in sterile water and administered per OS(oral gavage) 1.0 hr prior to GD, at a dose volume of 5.0 m&g.
S242
SOMAN
PRETREATMENT/TREATMENT
THERAPY
S243
Preparation of test compounds. All test compound conwhere centrations (both nerve agent and therapy) were prepared by dilution from a single master stock concentration of JT = PO+ AXI + BzX2 + &X3 + B4X4 ; P,,X,z + B22X22 the appropriate compound. This procedure ensured that + b33x32 + ,%2x,x2 + &3x,x3 + &.,x,x., all agent-challenge and therapy doses were accurate to the + 823X2x3 + i324xZx4 + &x,x, + &23x,x2x3 same level of significance. + ~124xIxZx4 + /3134x,x3x4 + 8234X2x3x4 Acetylcholinesterase inhibition (AChEI). Whole blood + 161234xIxZx3x4 AChEI was assessedat 1.O hr post-PYR, using the method and of Siakotos et al. (1969). Soman lethality determination. Soman lethality for both dose of 2-PAM (mg/kg) XI untreated (control) and treated (PYR, ATR, and/or 2PAM) guinea pigs was assessedfrom 24-hr mortality redose of ATR (mg/kg) x2 sults, based on five soman challenge doses administered dose of PYR (mg/kg) x3 at equally spaced logarithmic (log) intervals to six guinea exposure level of GD (&kg) X4 pigs per challenge dose. LDSO estimates (for use in PR an unknown parameter associated with the prors, assessment) were determined by probit analysis (Finney, portion of untreated survivors 1971). an unknown parameter associated with the effect Protective ratios. PRs were assessedas the ratio of the 01 CD LD50 following therapy to the GD LD50 without on survival of 2-PAM therapy. Statistical difference in PRs were assessed by an unknown parameter associated with the effect 82 Newman-Keul’s analysis of variance (Steel and Torrie, on survival of ATR 1960). Percentage of AChEI/PYR dose correlation was an unknown parameter associated with the effect 83 determined by PROPHET System nonlinear regression on survival of PYR analysis (NIH, 1980); PR/PYR dose correlation and PR/ percentage of AChEI correlation were determined by linear an unknown parameter associated with the effect 84 regression analysis (Tallarida and Jacob, 1979). on survival of GD Optimal ATR/2-PAM therapy study. Soman antidotal an unknown parameter associated with the toxicity 0 II efficacywas asses.& for singular and combination therapy of 2-PAM doses of ATR and/or 2-PAM. ATR doses of 0,4, 8, 16, an unknown parameter associated witb the toxicity 32, and 64 mg/kg were combined with 2-PAM doses of B22 0,6.25, 12.5,25, 50, and 100 mg/kg and administered (as of ATR described above) to guinea pigs challenged with five difB33 an unknown parameter associated with the toxicity ferent soman challenge doses. of PYR Optimal PYR/ATR/2-PAM therapy study. Soman anan unknown parameter associated with the interB I2 tidotal efficacywas asses& for combination therapy doses action between 2-PAM and ATR of PYR (0.12, 0.47, 1.9, 7.5 mgjkg), ATR (8, 16, 32, 64 mgjkg) and 2-PAM (12.5, 25, 50, 100 mgjkg). PYR prean unknown parameter associated with the interB13 treatment doses and ATR/Z-PAM treatment doses were action between 2-PAM and PYR administered (as described above) to guinea pigs challenged an unknown parameter associated with the interB 14 with five different GD challenge doses.Therapeutic efficacy action between 2-PAM and GD was assessedby RSM analysis (see below). an unknown parameter associated with the interRSM; statistical analysis. RSM was employed to assess a3 combination PYR pretreatment and ATR/Z-PAM treataction between ATR and PYR ment of GDinduced lethality, using the methods described an unknown parameter associated with the inter84 by Carter et al. (1979) (for additional details see Carter et action between ATR and GD al., 1985). PYR pretreatment, ATR and 2-PAM treatment, an unknown parameter associated with the interL334 and GD challenge doses were evaluated as independent action between PYR and GD variables, and probability of survival assessedas the dependent variable. As a result of the nature of the agents B 123 an unknown parameter associated with the interused, it would be expected that the proportion of animals action between 2-PAM, ATR, and PYR surviving would increase to a point where treatment toxan unknown parameter associated with the inter8124 icity would exceed the therapeutic effect,and then decrease action between 2-PAM, ATR, and GD as treatment levels increased beyond that point. Using the same reasoning as presented in the previous paper (Carter B 134 an unknown parameter associated with the interet al., 1985), the following equation was used: action between 2-PAM, PYR and GD an unknown parameter associated with the interB234 action between ATR, PYR and GD
S244
JONES, CARTER,
AND CARCHMAN
P,234 an unknown parameter associated with the interaction between 2-PAM, ATR, PYR, and CD The model parameters were estimated from the experimental data by the method of maximum likelihood. The p value for the likelihood ratio test for the significance of the model was
RESULTS
I; 0.01
0.1
LOG DOSE
IO PYRlOOSTlGMlNE
too
(mg I kg)
FIG. 2. Pretreatment efficacy (PR) of PYR alone and in combination with ATR (64 mgjkg) and ATR (100 mgj kg) plus 2-PAM ( 100 mg/kg) treatments. PYR alone and with 2-PAM had no effect on efficacy. In combination with ATR, PYR doses greater than 0.47 mg/kg afforded significantly (p < 0.05) greater PRs than ATR without PYR. In combination with ATR plus 2-PAM, PYR doses greater than 0.12 mg/kg afforded significantly (p < 0.05) greater PRs than ATR/Z-PAM without PYR. (0) saline; (A) 2-PAM 100 mg/kg; (B) atropine 64 mg/kg; (0) atropine 64 mg/kg + 2-PAM 100 mgjkg.
The GD antidotal efficacy of combination ATR/ZPAM therapy is illustrated in Fig. 1 as assessed by PR methods. An analysis of this same data in the preceding paper (Carter et al., 1985) by RSM indicated that the “optimal” ATR/ZPAM treatment combination varied as a function of the GD exposure level. (p < 0.05) greater PR (3.78) than all other The more traditional PR assessment was em- treatment combinations evaluated. These ployed here, however, in an attempt to identify ATR and 2-PAM doses were therefore selected a single ATR/2-PAM treatment combination as optimal treatment standards for use in the that would provide “optimal” protection evaluation of PYR pretreatment efficacy. across a wide range of GD exposure levels. As As illustrated in Fig. 2, neither PYR alone shown in Fig. 1, treatment with 64 mg/kg ATR nor in combination with 100 mg/kg 2-PAM plus 100 mg/kg 2-PAM afforded a significantly had any effect on therapeutic efficacy (i.e. PRs). In contrast, PYR in combination with both 64 mg/kg ATR (PYR > 0.47 mg/kg) and 0.64 mg/kg ATR plus 100 mg/kg 2-PAM (PYR > 0.12 mg/kg) afforded significantly (p < 0.05) greater PRs than the corresponding ATR or ATR/2-PAM therapy without PYR pretreatment. As these results show, however, even though PYR afforded significantly enhanced protection, the PR responses were extremely variable, with no apparent dose-re1 I I I I 4om 8clm 12o.00 160xX) 20(3oo sponse relationship for PYR efficacy, either WSE P-PAM Cl hg/kg, alone or in combination with ATR and/or FIG. 1. Optimal ATR/Z-PAM dose combination for 2-PAM. treatment of soman; ATR 64 mg/kg/Z-PAM 100 mg/kg It was thought that the reason for this variafforded significantly (p < 0.05) greater protection (PR ability in response may have been due to the = 3.78) than all other dose combinations evaluated. (0) atropine SO, 4.0 mg/kg; (6) atropine SO, 16.0 mg/kg; (A) quaternary nature of PYR, thus resulting in atropine SO, 64.0 mg/kg; (Y) atropine So, 128.0 mg/kg. poor or erratic oral absorption. Inhibition of
SOMAN
PRETREATMENT/TREATMENT
whole blood AChE was therefore assessed as a function of the oral PYR dose; this in vivo pharmacological activity was intended to provide an indication of the variability in PYR oral absorption. As shown in Fig. 3, a highly correlated (t = 0.96) sigmoidal dose-response relationship was found to exist between PYR and the resultant level of AChEI. These results indicate that, regardless of the quatemary nature of PYR, oral administration affords precise and highly predictable in vivo pharmacologic activity, thus inferring a relatively consistent oral absorption profile. Table 1 presents these data in tabular form and correlates the doses of PYR used versus both the computed PR and the measured level of AChEI. As indicated, there was a very high correlation between AChEI and PYR dose (r = 0.96), but a very poor correlation between both the PR and the PYR dose (I = 0.45), as well as between the PR and the level of AChEI (I = 0.48). The inability of these studies to establish a definitive dose-response relationship for PYR pretreatment efficacy indicated that either PYR efficacy is not dose dependent, or the ATR/ZPAM doses used were inappropriate to show such a response. The second possibility is supported by the findings of the preceding paper (Carter et al., 1985), which
z
80
I=
5 60 z 5 40 z' y 20
DOSE
PYRIDOSTIGMINE
(mg/kg)
FIG. 3. Percentage of AChEI as a function of PYR dose. Although PYR is a quatemary compound, oral administration produced a highly correlated, dose (lo&dependent inhibition of whole blood AChE (r = 0.96).
S245
THERAPY TABLE 1 COMPARNNSOFPYR DOSE, PR,AND PERCENTAGEOFAC~EI 96 Inhibition AChE
DosePYR bw/kg)
3.1 + 8.9 5.4 f 2.1 16.4 + 9.5 26.4 f 20.6 34.0 f 11.2 52.4 + 6.5 71.9 + 7.1 79.1 -I 5.2 82.2 zk 6.1
0.06 0.12 0.23 0.41 0.94 1.90 3.15 7.50 15.0 l--r
PR
= 0.96*
4.16 5.62 5.02 6.38 4.91 5.21 5.16 6.33 5.95 r.= 0.48-I
Note. PRs assessedfollowing adjunct treatment with 64 mg/kg HTR plus 100 mg/kg 2-PAM.
indicated that no single optimal therapy dose exists for treatment of GD, rather, optimal therapy is dependent on the GD exposure level. To address these concerns, a three-tiered study utilizing multiple PYR pretreatment doses in combination with multiple ATR plus 2-PAM treatment doses was conducted as outlined under Methods. Since five separate variables were inherent to such a study (PYR, ATR, 2-PAM, GD, and survival), RSM was employed to assessand interpret the resultant data. The data presented in Table 2 provide RSM-generated estimates of the parameters for the various drug agent interactions. Unlike those reported in the previous study (Carter et al., 1985), not all ofthe potential therapeutic drugs and combinations were statistically significant. Notably, 2-PAM, ATR X 2-PAM, ATR X GD, ATR X 2-PAM, ATR X GD, ATR X 2-PAM X GD represent those agents and interactions which fell into this category. All three of the potential therapeutic agents (ATR, 2-PAM, PYR) exhibited the ability to
S246
JONES, CARTER, AND CARCHMAN TABLE 3
TABLE 2 ESTIMATION
ESTIMATION OF MODEL PARAMETERS
Variable Intercept Xl x2 x3 x4
XlSQ
x2SQ MSQ XIX2 x1x3 x1x4 x2x3 x2x4 x3x4 x1x2x3 x1x2x4 x1x3x4 x2x3x4 x1mx4
Parameter estimate
0% (0) (1)
0.1690
1.7 7.0 -2.7 -2.1 -2.0 -1.1 3.7 4.3 1.3 8.1 -7.6 3.2 -1.5 -2.0 -4.5 -3.8 9.7
(2)
<.oooOl 4001 <.OOOl 0.0002
X IO-* x x x x x x
10-4 10-3 10-I 10-4 10-3 10-4
x lo-’ x IO-’
X IO-’ x lo+ x 10-6
x 10-S X lO-5 x lo-’
(3) (4) (1, 1)
(2, 2) (3, 3) (1,2) (I>31 (1,4) (2, 3) (224) (334) (1,2, 3) (1,2,4)
(1,3,4) (2, 3, 4) Cl,& 3>4)
FOR
p value
-5.5 x 10-l 4.7 x 10-3 x 10-I x 10-I
OF OPTIMAL THERAPY SOMAN EXPOSURE
0.6172
<.ooo1 0.0000 0.0927 0.0076 0.0468 0.0085 0.9936 0.0000 0.0079 0.2678 0.0017 0.0500 0.0100
produce innate toxicity (i.e., pi&, ,&&, &&), being negative and statistically significant. Of potential therapeutic significance was the fact that the interaction of all three drugs with GD produced a positive and significant interaction (i.e., 1018283@44)The Nelder and Mead ( 1965) optimization procedure was utilized for this four-chemical interaction study. Results, presented in Table 3, demonstrate the relationships between GD exposure (28-498 peg), the therapeutic drugs, and the percentage of survival. Estimates of the optimal treatment regimens indicate that the antidotal agents can provide various degrees of protection. In addition, other features of this information are evident from the data provided in the following figures. Figure 4 depicts the relationship of therapy optimization and survivability as a function of GD exposure. As indicated, this optimization predicts that use of optimal therapy combinations should provide a large degree of
28.0 31.4 35.2 39.6 44.4 49.8 55.9 62.7 70.3 78.9 88.5 99.3
26.0 26.0 26.6 27.0 27.6 28.2 28.8 29.6 30.3 31.2 32.1 33.0 34.6 36.6 37.9 39.3 40.8 42.4 44.3 46.5 49.6 56.0 100.0 100.0 100.0
111.5 140.0 157.0 176.7 198.0 222.4 250.0 280.0 314.0 352.0 396.0 444.0 498.0
50.0 50.0 50.0 50.0 49.6 49.5 49.3 49.1 48.9 48.7 48.7 48.2 48.0 41.4 47.1 46.8 46.6 46.4 46.3 46.3 46.7 48.1 64.0 64.0 64.0
4.8 4.8 4.9 4.9 4.9 4.9 4.9 5.0 5.0 5.1 5.1 5.2 5.2 5.4 5.5 5.6 5.7 5.8 6.0 6.2 6.3 6.6 7.5 7.5
1.5
99.6 99.5 99.5 99.4 99.4 99.3 99.2 99.1 99.0 98.8 98.6 98.3 97.2 96.4 95.2 93.3 90.6 86.3 80.0 70.5 57.9 43.0 30.5 23.1 16.3
benefit. Indicated in Fig. 4 is the calculated LD50 for GD obtained in these studies as well as the predicted magnitude of shift in GD tox-
1 ZL -----------i ‘O-I
I
0’ rlosEwMAt4 (pg/Iql FIG. 4. Percentage of survival of guinea pigs pretreated with optimal pyridostigmine and treated with optimal atropine/2-PAM treatment against soman poisoning.
SOMAN
PRETREATMENT/TREATMENT
icity (11.9 X LD50) afforded by this optimal therapy. From the information presented in Tables 2 and 3 and Fig. 4, it was apparent that no single dose combination of PYR/ATR/ZPAM was “optimal” for treatment of soman intoxication; optimal doses for these therapeutic modalities were found to be dependent on interactions between the various therapeutic agents as well as on the level of GD exposure. In order to more simply view how the various therapeutic indices changed as a function of the GD exposure level, the effects of the various PYR pretreatment doses on the optimal ATR dose, 2-PAM dose, and survival were assessed as a function of the GD exposure level. The data depicted in Fig. 5 compare the relationship of various PYR doses to the optimal ATR component as a function of the GD exposure level. Across the GD doses and PYR doses tested, ATR remained relatively constant. This constancy was similar to that observed in the previous study in the absence of PYR. It is important to note that though the presence of PYR provided a qualitatively similar ATR optimal dose with respect to GD exposure, there were important qualitative differences; ATR optimum in the absence of
0
loo
200 300
DOSE SOWN
hg/kp)
RG.5.OptimalATR treatment as determined by RSM. Optimal ATR therapy remained relatively constant, independent of the PYR pretreatment dose and soman exposure level. (0) pyridostigmine 0.12 mg/kg; (+) pyridostigmine 0.47 mg/kg; (0) pyridostigmine 1.9 mg/kg; (A) pyridostigmine 7.5 mg/kg.
S247
THERAPY
Km
200 300
DOSE SOMAN hq/kg)
FIG. 6. Optimal 2-PAM treatment as determined by RSM. Optimal 2-PAM therapy increased as a function of increasing soman exposure levels, but decreased as the PYR pretreatment was increased. (0) Pyridostigmine 0.12 mg/kg; (+) pyridostigmine 0.47 mgjkg; (Cl) pyridostigmine 1.9 mg/kg; (A) pyridostigmine 7.5 mg/kg.
PYR at 84.6 pg/kg GD was equal to 168 mg/ kg (see Carter et al., 1985), whereas in the presence of optimal PYR, the ATR optimum = -50 mg/kg (Table 3). A similar comparison with respect to 2PAM is depicted in Fig. 6. At PYR doses of 0.12,0.47, and 1.9 mg/kg, the 2-PAM optimal dose(s) as a function of GD exposure were quite similar; the 2-PAM component increased in a linear manner as a function of GD exposure. At 7.5 mg/kg PYR, however, there was a slight qualitative and quantitative difference in this linear relationship. Overall though, as the dose of PYR increased, less 2PAM was required, but independent of the PYR dose the 2-PAM requirement increased as a function of GD dose. This was also observed in the previous study in the absence of PYR, and as was seen for ATR, PYR dramatically lowered the 2-PAM optimal component. Previous reports have shown ATR to be essential for PYR pretreatment (Dimhuber et al., 1979, Harris et al., 1980). The results presented here substantiate these previous findings; in addition, they provide additional definitive evidence for a positive therapeutic role for 2-PAM as an adjunct to PYR plus ATR therapy. These data indicate that PYR pre-
S248
JONES, CARTER, AND CARCHMAN
treatment dramatically reduces the amount of 2-PAM treatment needed to achieve optimal protection at any level of agent exposure. Although not graphically presented here, the data also supports the converse of this observation, i.e., 2-PAM treatment reduces the amount of PYR pretreatment needed. Figure 7 shows percentage of survival afforded by various pretreatment doses of PYR (in combination with the optimal ATR/2PAM treatment for each PYR dose and CD exposure level) as a function of GD exposure. Also shown are the effects of no therapy as well as optimal ATR/2-PAM without PYR. Similar to what was shown in Fig. 4 (following optimization of all the therapeutic modalities), percentage of survival decreased as a function of increasing GD levels. Increasing the PYR dose, however, resulted in a corresponding rightward shift in the mortality curve; the calculated GD ED50 doses (i.e., the soman LD50) for each PYR curve are also indicated. Since these values are analagous to the treated soman LD50 values traditionally used for assessment of PRs, these findings actually indi-
TABLE 4 COMPARISONOF PYR DOSE, PR, AND PERCENTAGEOF AChEI Dose PYR OWW
% Inhibition AChE
.12 .47 1.90 7.50
5.4 26.4 52.4 79.1
PR 5.6 11.9
~----po.974 Nofe. PRs calculated for each PYR pretreatment dose in combination with optimal ATR/Z-PAM treatment (as assessedby RSM).
cate that, ifATR/2-PAM therapy is optimized, PYR pretreatment affords a dose-dependent increase in the PR. A clearer perspective of this observation is presented in Table 4. Again, AChEI was highly correlated (I = 0.96) with the PYR pretreatment dose. Unlike the results presented in Table 1 (varied PYR pretreatment followed by a single treatment dose of ATR/2-PAM), Table 4 indicates that if ATR/ZPAM treatment is optimized, PYR alfords a highly correlated (r = 0.95) dose (1og)dependent increase in the PR. In addition, under these test conditions, PYR efficacy (assessedas PR) was found to be highly correlated (r = 0.97) with the level of PYR-induced AChEI. DISCUSSION
FIG. 7. Percentage of survival with varying PYR pretreatment doses followed by optimal ATR and 2-PAM therapy. Percentage of survival decreased as the soman exposure level increased, but increased (at all levels of soman exposure) as the PYR pretreatment dose was increased. (a * .) Indicates soman lethality with no therapy; (+) indicates the soman LD50 for each PYR pretreatment group; (--) pyridostigmine 0.12 mg/kg; (---) pyridostigmine 0.47 mg/kg; (-) pyridostigmine 1.9 mgjkg; (- - -) pyridostigmine 7.5 mpB; (- v-) no pyridostigmine/optimal ATR/Z-PAM.
Further information has been provided concerning the use of RSM to estimate therapeutic optima. In this example, an additional antidotal treatment (i.e., pyridostigmine pretreatment) was utilized compared with the antidotal data presented in the previous paper (Carter et al., 1985). Therapeutic optimal search procedures were utilized here to evaluate the entire experimental region; this was equivalent to evaluating all possible treatment
SOMAN
PRETREATMENT/TREATMENT
combinations and choosing the maximum. This procedure is more precise than altering fixed percentages of one drug in a combination, or using PRs to evaluate a relatively small number of treatment combinations. In addition, RSM utilizes all the experimental data, not just singular points (i.e., ED50). In a multidimensional study such as the one presented, (i.e., five dimensions), techniques that require limited data procedures (i.e., PR) or which require visualization of the response (Berry et al., 1971; Boskovic et al., 1984; Boskovic and Stern, 1970; Clement and Lockwood, 1982; Dimhuber et al., 1979; Faff et al., 1976; Jones et al., 1984; Koplovitz, et al., 1984; and Green, 1984) are inherently restricted in their ability to accurately assess the full dose-response surface. Use of the prediction equation to estimate the response of any (or all) treatment combination(s) within the experimental region makes RSM a powerful approach to analyze such complex experimental procedures. The estimation of the parameter (i.e., p) terms for the individual compounds and their interactions provides additional information. Though PYR pretreatment produced a large, positive, and statistically significant therapeutic response (as indicated by &; Table I), it cannot do so indefinitely (i.e., &, a reflection of the PYR toxicity). Unlike the previous study, 2-PAM by itself did not produce a significant therapeutic response; this is related to the fact that the experimental conditions (e.g., GD dose) covered a range > five times that used in the previous study. Further analysis revealed that PYR pretreatment reduced the ATR and 2-PAM requirements compared to the optimum observed in the previous study. This reduction for ATR was -2/3, and was relatively constant as a function of GD dose. In contrast, the 2-PAM dose optimum, though reduced, was found to vary inversely as a function of PYR pretreatment dose and progressively as a function of GD dose. This latter finding was consistent with the data from the previous study, i.e., ATR optimum was constant, whereas 2-PAM optimum increased as
THERAPY
S249
a function the GD dose. Overall, PYR pretreatment reduced the ATR/2-PAM optimum treatment, but did not appear to change the relationship between these two agents. The relationship of “efficacy” to PYR dose was highly correlated only if RSM was used to optimize the adjunctive ATR/ZPAM treatment doses. This was shown in Tables 1 and 4 and Fig. 7, with the use of PRs to evaluate PYR efficacy across a wide range of GD exposure levels. If percentage of survival following a single exposure level of GD was used as an alternative criteria (rather than PRs) for evaluating PYR efficacy, RSM analysis also provides the information necessary to make such an assessment. Although this approach was not specifically addressed in this paper, the information presented in Fig. 7 substantiates this claim. If percentage of survival were assessed(as the dependent variable) following a single GD exposure level (e.g., 200 clg/kg) it can be seen that a dose-related increase in percentage of survival occurs with increasing PYR pretreatment levels. Although the data assessment is not provided in this paper, this dosedependent response was found to be highly correlated (r = 0.99), (D. E. Jones, 1985, unpublished observations) by nonlinear regression analysis (NIH, 1980). This example of an alternative analytical approach for assessing PYR efficacy was provided in order to further emphasize the utility of RSM analysis. Similar to what was found with therapeutic efficacy (PRs) and PYR dose, PYR efficacy was also highly correlated with AChEI, but only if RSM was used to optimize the adjunctive ATR/2-PAM treatment. The biological significance of this observation is meaningful in that one can infer that PYR pretreatment, by reducing AChE, is providing a protective mechanism against GD-induced lethality. Although these observations and inferences by no means establish a definitive cause-effect relationship, they are nonetheless in agreement with the proposed mechanism of action for PYR as originally postulated by Berry and Davies (1970). Unfortunately, the
S250
JONES, CARTER,
role of whole blood AChE is poorly understood with regard to nerve agent intoxication. However, PYR-induced AChEI in other tissues has been shown to closely parallel that observed in whole blood (Harris et al., 1980; Hey1 et al., 1980; Karlsson et al. 1984). The approaches and findings presented here, in combination with these earlier observations, will hopefully serve to both reinforce and expand the mechanistic interpretations concerning the therapeutic role of pretreatment carbamates. In summary, the two studies presented here demonstrate the utility of RSM to analyze complex studies and to arrive at therapeutic optima. They also provide provocative insights into possible interactions and mechanisms. In addition to therapeutic optimization, RSM also provides an assessment of the entire doseresponse surface for all inherent variables (dependent and independent). Such features greatly expand the range and significance of information that can be derived from a single set of experimental data. RSM analysis of the data provided in these papers provides a means to accurately predict the therapeutic efficacy afforded by all possible combination of PYR/ ATR/ZPAM against any GD exposure level. Such information becomes extremely valuable for formulating rational treatment hypotheses, especially if used in conjunction with information provided by other experimental procedures (e.g., maximum tolerated doses of therapy compounds or anticipated level of GD exposure). Since candidate nerve agent therapies cannot be evaluated against nerve agent in a controlled clinical (human exposure) situation, potential therapeutic dosage regimens must be selected on the basis of nontoxic (e.g., nonincapacitating doses). Only by testing these dosages in nerve agent-exposed animal models can any meaningful assessment (albeit comparative) be made of their therapeutic potential. It is therefore essential that the most accurate and informative means available be used to make these assessments. RSM procedures, as oulined in these two papers, provide
AND CARCHMAN
the most accurate and comprehensive means yet reported for making these complex assessments of nerve agent antidotal efficacy. REFERENCES BERRY, W. K.. AND DAVIES, D. R. (1969). The use of carbamates and atropine in the protection of animals against poisoning by 1,2,2-trimethylpropylmethylphosphonofluoridate. Biochem. Pharmacol. 19,921-934. BERRY,W. K., DAVIES, D. R., AND GORDON, J. J. (197 1). Protection of animals against soman (1,2,2-trimethylpropylmethylphosphono-fluoridate) by pretreatment with some other organophosphorus compounds, followed by oxime and atropine. Biochem. Pharmacol. 20, 125-134. BOSKOVIC, B., KOVACEVIC, V., AND JOVANOVIC, D. (1984). PAM-2 Cl, HI-6, and HGG-12 in soman and tabun poisoning. Fundam. Appl. Toxicol. 4, S106-S 115. BOSKOVIC, B., AND STERN, P. (1970). Protective effects of oximes and cholinolytics against soman poisoning. Arch. Toxicol. 26, 306-3 10. CARTER, W. H., JR., STABLEIN, D. M., AND WAMPLER, G. L. (1979). An improved method for analyzing survival data from combination chemotherapy experiments. Cancer Res. 39, 3446-3453. CARTER, W. H.. JR., JONES, D. E., AND CARCHMAN, R. A. (1985). Application of response surface methods for evaluating the interactions of soman, atropine and pralidioxime chloride. Fundam. Appl. Toxicol. 5, OOO000. CLEMENT, J. G., AND LOCKWOOD, P. A. (1982). HI-6, an oxime which is an effectiveantidote of soman poisoning: a structure-activity study. Toxicol. Appl. Pharmacol. 64, 140-146. DIRNHUBER, P., FRENCH, M. C., GREEN, D. M., LEAD BEATER, L., AND STRATTON, J. A. (1979). The protection of primates against soman poisoning by pretreatment with pyridostigmine. J. Pharm. Pharmacol. 31, 295-299.
FAFF, J., BORKOWSKA, G., AND BAK, W. (1976). Therapeutic effectsof some cholinolytics in organophosphate intoxications. Arch. Toxicol. 36, 139- 146. FINNEY, D. J. (197 1). Probit Analysis, 3rd ed. Cambridge Univ. Press, Cambridge, Mass. GREEN, M. D. (May 1984). Guinea pig model for correlating efficacy with pralidoxime chloride (2-PAM) plasma levels. Proc. Annu. Chem. Dej Biosci. Rev. pp. 71-95.
HARRIS,L. W., STITCHER,D. L., AND HEYL, W. C. (1980). The effectsof pretreatments with carbamates, atropine and mecamylamine on survival and on soman-induced
SOMAN
PRETREATMENT/TREATMENT
alterations in rat and rabbit brain acetylcholine. Life Sci. 26, 1885-1891. HEYL, W. C., HARRIS, L. W., AND STITCHER,D. L. (1980). Effects of carbamate on whole blood cholinesterase activity: Chemical protection against soman. Drug Chem. Toxicol. 3(3), 3 19-332. JONES,D. E., KOPLOVITZ, I., HARRINGTON, D. G., HILMAS, D. E., AND CANFIELD, C. J. (May 1984). Models for assessingefficacy of therapy compounds against organophosphates (OP). Proc. Annu. Chem. Del: Biosci. Rev., pp. l-37. KARLSSON, N., LARSSON, R., AND Puu, G. (1984). Ferrocene-carbamate as prophylaxis against soman poisoning. Fundam. Appl. Toxicol. 4, S 184-S 189. KOPLOVITZ, I., JONES,D. E., HARRINGTON, D. G., HILMAS, D. E., AND CANL~ELD, C. J. (May 1984). Models for assessingefficacyof pretreatment compounds against organophosphates (OPs). Proc. Annu. Chem. Def: Biosci. Rev., p. 39-69. NELDER, J. A., AND MEAD, R. A. (1965). A simplex
THERAPY
S251
method for function minimization. Cornput. J. 7,308313. NIH (National Institutes of Health) (1980). Regression analysis. In Prophet Statistics: A User’s Guide to Statistical Analysis on the Prophet System. U.S. Department of Health and Human Services, Public Health Service, National Institutes of Health, Publication No. 80-2 169. Bolt, Baranek and Newman, Inc., Cambridge, Mass. SIAKOTOS, A. N., FILBERT, M., AND HESTER, R. (1969). A specific radioisotopic assay for acetylcholinesterase and pseudocholinesterase in brain and plasma. Biochem. Med. 3, l-9. STEEL, R. G. D., AND TORRIE, J. H. (1960). Analysis of variance I: The one-way classification. In Principles and Procedures of Statistics (Steel, R. G. D. and Tonie, J. H., eds.),Ch. 7, pp. 99-131. McGraw-Hii, New York. TALLARIDA, R. J., AND JACOB, L. S. (1979). The Dose Response Relation in Pharmacology pp. I- 17 and 8% 110. Springer-Verlag, New York.
AND
APPLIED
TOXICOLLIGY
5,
Assessing Pyridostigmine
S242-S25 I (1985)
Efficacy by Response
D. E. JONES,* W. H. CARTER,
Surface Modeling
JR.,? AND R. A. CARCHMAN~
*U.S. Army Medical Research Institute of Chemical Defense9Drug Assessment Division, Aberdeen Proving Ground, Maryland 21010; and tDepartments of Biostatistics; Pharmacology and Toxicolofl, Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia 23298 Assessing Pyridostigmine Efficacy by Response Surface Modeling. JONES, D. E., CARTER, AND CARCHMAN, R. A. (1985). Fur&m. Appl. Toxicol. 5,5242-S25 1. The therapeutic efficacy of atropine sulfate/pralidoxime chloride (ATR/Z-PAM) treatment (im) therapy, and pyridostigrnine bromide (PYR) pretreatment (oral) therapy were evaluated in soman-challenged guinea pigs. ATR/Z-PAM efficacy was assessedas protective ratio (PR = treated soman LDSO/control soman LD50); PYR efficacy was assessedboth as PR and by response surface modeling (RSM) techniques. The optimal ATR/2-PAM treatment gave a PR of 3.78. PYR pretreatment (1 hr) produced a dose&&dependent (r = 0.96) inhibition of whole blood AChE and afforded significant (p < 0.05) increases in PR (with dosesgreater than 0.12 mg/kg PYR) against soman when followed by 64 mg/kg ATR/lOO mg/kg 2-PAM treatment. These PRs, however, were poorly correlated (r = 0.45) with the corresponding level of PYR-induced AChE inhibition. In contrast, RSM analysis of efficacyindicated that the optimal ATR/2-PAM dose combination varied as a function of both the soman-challenge level and the PYR pretreatment dose. Efficacy was therefore evaluated for varying PYR pretreatment doses in combination with the appropriate optimal ATR/ZPAM treatment (as determined by RSM for each soman challenge dose and PYR dose evaluated). When assessedin this manner, PYR efficacy(PYR) was found to be highly correlated (r = 0.97) with PYR-induced AChE inhibition. Since percentage of AChE inhibition was directly correlated with PYR dose (log), these results indicate that PYR pretreatment efficacy is a highly correlated, dose-dependent phenomenon, providing ATR/Z-PAM treatment is optimized. o 1985 say or W. H., JR.,
In this study, the use of response surface modeling (RSM) and protective ratio (PR) methods for evaluating pyridostigmine bromide (PYR) pretreatment (in conjunction with atropine sulfate/pralidoxime chloride [ATR/ZPAM] treatment) were examined for treatment of soman intoxication. RSM provides an improvement over currently available methods (Berry et al., 197 1; Boskovic et al., 1984; Boskovic and Stern, 1970; Clement and Lockwood, 1982; Dirnhuber et al., 1979; and Faff et al., 1976) and a scientifically valid procedure for optimizing therapy in this complex (fivedimensional) study. MATERIALS
AND
METHODS
Animals. Mixed-sex, 300 to 400-g Duncan Hartley Albino guinea pigs from Charles River Breeding Laboratories 0272-0590185 $3.00 Copyright 0 1985 by the Society of Toxicology. All rights of rxprcduction in any form reserved.
were used in all experiments. Containment and test facilities were maintained at constant temperature (2 I + 3”C), humidity (55 + 7%), and lighting (12-hr light/dark cycle). Nerve agent. Soman (GD, pinacolylmethylphosphonofluoridate) was prepared by the Chemical Research and Development Center, Aberdeen Proving Ground, Maryland. Purity was determined (by nuclear magnetic resonance) to exceed 95% for each GD lot. GD challenge doses were prepared in sterile isotonic saline and injected sc in the dorsal cervical area at a dose volume of 1.0 ml/k. Therapy compounds. ATR was purchased from C. H. Boehringer and Sons, Ingelheim, West Germany (Lot No. 352). 2-PAM was purchased from Ayerst Laboratories, New York, New York (Lot No. Z-799). ATR and 2-PAM concentrations (both single and combination therapies) were prepared in diititled, deionized water and injected im (rear limb muscle mass) at a dose volume of I.0 ml/ kg, given 1.Omin postGD challenge. Combination therapy injections of 2-PAM and ATR were admixed and administered as a single therapy injection. Pretreatment. PYR was diluted in sterile water and administered per OS(oral gavage) 1.0 hr prior to GD, at a dose volume of 5.0 m&g.
S242
SOMAN
PRETREATMENT/TREATMENT
THERAPY
S243
Preparation of test compounds. All test compound conwhere centrations (both nerve agent and therapy) were prepared by dilution from a single master stock concentration of JT = PO+ AXI + BzX2 + &X3 + B4X4 ; P,,X,z + B22X22 the appropriate compound. This procedure ensured that + b33x32 + ,%2x,x2 + &3x,x3 + &.,x,x., all agent-challenge and therapy doses were accurate to the + 823X2x3 + i324xZx4 + &x,x, + &23x,x2x3 same level of significance. + ~124xIxZx4 + /3134x,x3x4 + 8234X2x3x4 Acetylcholinesterase inhibition (AChEI). Whole blood + 161234xIxZx3x4 AChEI was assessedat 1.O hr post-PYR, using the method and of Siakotos et al. (1969). Soman lethality determination. Soman lethality for both dose of 2-PAM (mg/kg) XI untreated (control) and treated (PYR, ATR, and/or 2PAM) guinea pigs was assessedfrom 24-hr mortality redose of ATR (mg/kg) x2 sults, based on five soman challenge doses administered dose of PYR (mg/kg) x3 at equally spaced logarithmic (log) intervals to six guinea exposure level of GD (&kg) X4 pigs per challenge dose. LDSO estimates (for use in PR an unknown parameter associated with the prors, assessment) were determined by probit analysis (Finney, portion of untreated survivors 1971). an unknown parameter associated with the effect Protective ratios. PRs were assessedas the ratio of the 01 CD LD50 following therapy to the GD LD50 without on survival of 2-PAM therapy. Statistical difference in PRs were assessed by an unknown parameter associated with the effect 82 Newman-Keul’s analysis of variance (Steel and Torrie, on survival of ATR 1960). Percentage of AChEI/PYR dose correlation was an unknown parameter associated with the effect 83 determined by PROPHET System nonlinear regression on survival of PYR analysis (NIH, 1980); PR/PYR dose correlation and PR/ percentage of AChEI correlation were determined by linear an unknown parameter associated with the effect 84 regression analysis (Tallarida and Jacob, 1979). on survival of GD Optimal ATR/2-PAM therapy study. Soman antidotal an unknown parameter associated with the toxicity 0 II efficacywas asses.& for singular and combination therapy of 2-PAM doses of ATR and/or 2-PAM. ATR doses of 0,4, 8, 16, an unknown parameter associated witb the toxicity 32, and 64 mg/kg were combined with 2-PAM doses of B22 0,6.25, 12.5,25, 50, and 100 mg/kg and administered (as of ATR described above) to guinea pigs challenged with five difB33 an unknown parameter associated with the toxicity ferent soman challenge doses. of PYR Optimal PYR/ATR/2-PAM therapy study. Soman anan unknown parameter associated with the interB I2 tidotal efficacywas asses& for combination therapy doses action between 2-PAM and ATR of PYR (0.12, 0.47, 1.9, 7.5 mgjkg), ATR (8, 16, 32, 64 mgjkg) and 2-PAM (12.5, 25, 50, 100 mgjkg). PYR prean unknown parameter associated with the interB13 treatment doses and ATR/Z-PAM treatment doses were action between 2-PAM and PYR administered (as described above) to guinea pigs challenged an unknown parameter associated with the interB 14 with five different GD challenge doses.Therapeutic efficacy action between 2-PAM and GD was assessedby RSM analysis (see below). an unknown parameter associated with the interRSM; statistical analysis. RSM was employed to assess a3 combination PYR pretreatment and ATR/Z-PAM treataction between ATR and PYR ment of GDinduced lethality, using the methods described an unknown parameter associated with the inter84 by Carter et al. (1979) (for additional details see Carter et action between ATR and GD al., 1985). PYR pretreatment, ATR and 2-PAM treatment, an unknown parameter associated with the interL334 and GD challenge doses were evaluated as independent action between PYR and GD variables, and probability of survival assessedas the dependent variable. As a result of the nature of the agents B 123 an unknown parameter associated with the interused, it would be expected that the proportion of animals action between 2-PAM, ATR, and PYR surviving would increase to a point where treatment toxan unknown parameter associated with the inter8124 icity would exceed the therapeutic effect,and then decrease action between 2-PAM, ATR, and GD as treatment levels increased beyond that point. Using the same reasoning as presented in the previous paper (Carter B 134 an unknown parameter associated with the interet al., 1985), the following equation was used: action between 2-PAM, PYR and GD an unknown parameter associated with the interB234 action between ATR, PYR and GD
S244
JONES, CARTER,
AND CARCHMAN
P,234 an unknown parameter associated with the interaction between 2-PAM, ATR, PYR, and CD The model parameters were estimated from the experimental data by the method of maximum likelihood. The p value for the likelihood ratio test for the significance of the model was
RESULTS
I; 0.01
0.1
LOG DOSE
IO PYRlOOSTlGMlNE
too
(mg I kg)
FIG. 2. Pretreatment efficacy (PR) of PYR alone and in combination with ATR (64 mgjkg) and ATR (100 mgj kg) plus 2-PAM ( 100 mg/kg) treatments. PYR alone and with 2-PAM had no effect on efficacy. In combination with ATR, PYR doses greater than 0.47 mg/kg afforded significantly (p < 0.05) greater PRs than ATR without PYR. In combination with ATR plus 2-PAM, PYR doses greater than 0.12 mg/kg afforded significantly (p < 0.05) greater PRs than ATR/Z-PAM without PYR. (0) saline; (A) 2-PAM 100 mg/kg; (B) atropine 64 mg/kg; (0) atropine 64 mg/kg + 2-PAM 100 mgjkg.
The GD antidotal efficacy of combination ATR/ZPAM therapy is illustrated in Fig. 1 as assessed by PR methods. An analysis of this same data in the preceding paper (Carter et al., 1985) by RSM indicated that the “optimal” ATR/ZPAM treatment combination varied as a function of the GD exposure level. (p < 0.05) greater PR (3.78) than all other The more traditional PR assessment was em- treatment combinations evaluated. These ployed here, however, in an attempt to identify ATR and 2-PAM doses were therefore selected a single ATR/2-PAM treatment combination as optimal treatment standards for use in the that would provide “optimal” protection evaluation of PYR pretreatment efficacy. across a wide range of GD exposure levels. As As illustrated in Fig. 2, neither PYR alone shown in Fig. 1, treatment with 64 mg/kg ATR nor in combination with 100 mg/kg 2-PAM plus 100 mg/kg 2-PAM afforded a significantly had any effect on therapeutic efficacy (i.e. PRs). In contrast, PYR in combination with both 64 mg/kg ATR (PYR > 0.47 mg/kg) and 0.64 mg/kg ATR plus 100 mg/kg 2-PAM (PYR > 0.12 mg/kg) afforded significantly (p < 0.05) greater PRs than the corresponding ATR or ATR/2-PAM therapy without PYR pretreatment. As these results show, however, even though PYR afforded significantly enhanced protection, the PR responses were extremely variable, with no apparent dose-re1 I I I I 4om 8clm 12o.00 160xX) 20(3oo sponse relationship for PYR efficacy, either WSE P-PAM Cl hg/kg, alone or in combination with ATR and/or FIG. 1. Optimal ATR/Z-PAM dose combination for 2-PAM. treatment of soman; ATR 64 mg/kg/Z-PAM 100 mg/kg It was thought that the reason for this variafforded significantly (p < 0.05) greater protection (PR ability in response may have been due to the = 3.78) than all other dose combinations evaluated. (0) atropine SO, 4.0 mg/kg; (6) atropine SO, 16.0 mg/kg; (A) quaternary nature of PYR, thus resulting in atropine SO, 64.0 mg/kg; (Y) atropine So, 128.0 mg/kg. poor or erratic oral absorption. Inhibition of
SOMAN
PRETREATMENT/TREATMENT
whole blood AChE was therefore assessed as a function of the oral PYR dose; this in vivo pharmacological activity was intended to provide an indication of the variability in PYR oral absorption. As shown in Fig. 3, a highly correlated (t = 0.96) sigmoidal dose-response relationship was found to exist between PYR and the resultant level of AChEI. These results indicate that, regardless of the quatemary nature of PYR, oral administration affords precise and highly predictable in vivo pharmacologic activity, thus inferring a relatively consistent oral absorption profile. Table 1 presents these data in tabular form and correlates the doses of PYR used versus both the computed PR and the measured level of AChEI. As indicated, there was a very high correlation between AChEI and PYR dose (r = 0.96), but a very poor correlation between both the PR and the PYR dose (I = 0.45), as well as between the PR and the level of AChEI (I = 0.48). The inability of these studies to establish a definitive dose-response relationship for PYR pretreatment efficacy indicated that either PYR efficacy is not dose dependent, or the ATR/ZPAM doses used were inappropriate to show such a response. The second possibility is supported by the findings of the preceding paper (Carter et al., 1985), which
z
80
I=
5 60 z 5 40 z' y 20
DOSE
PYRIDOSTIGMINE
(mg/kg)
FIG. 3. Percentage of AChEI as a function of PYR dose. Although PYR is a quatemary compound, oral administration produced a highly correlated, dose (lo&dependent inhibition of whole blood AChE (r = 0.96).
S245
THERAPY TABLE 1 COMPARNNSOFPYR DOSE, PR,AND PERCENTAGEOFAC~EI 96 Inhibition AChE
DosePYR bw/kg)
3.1 + 8.9 5.4 f 2.1 16.4 + 9.5 26.4 f 20.6 34.0 f 11.2 52.4 + 6.5 71.9 + 7.1 79.1 -I 5.2 82.2 zk 6.1
0.06 0.12 0.23 0.41 0.94 1.90 3.15 7.50 15.0 l--r
PR
= 0.96*
4.16 5.62 5.02 6.38 4.91 5.21 5.16 6.33 5.95 r.= 0.48-I
Note. PRs assessedfollowing adjunct treatment with 64 mg/kg HTR plus 100 mg/kg 2-PAM.
indicated that no single optimal therapy dose exists for treatment of GD, rather, optimal therapy is dependent on the GD exposure level. To address these concerns, a three-tiered study utilizing multiple PYR pretreatment doses in combination with multiple ATR plus 2-PAM treatment doses was conducted as outlined under Methods. Since five separate variables were inherent to such a study (PYR, ATR, 2-PAM, GD, and survival), RSM was employed to assessand interpret the resultant data. The data presented in Table 2 provide RSM-generated estimates of the parameters for the various drug agent interactions. Unlike those reported in the previous study (Carter et al., 1985), not all ofthe potential therapeutic drugs and combinations were statistically significant. Notably, 2-PAM, ATR X 2-PAM, ATR X GD, ATR X 2-PAM, ATR X GD, ATR X 2-PAM X GD represent those agents and interactions which fell into this category. All three of the potential therapeutic agents (ATR, 2-PAM, PYR) exhibited the ability to
S246
JONES, CARTER, AND CARCHMAN TABLE 3
TABLE 2 ESTIMATION
ESTIMATION OF MODEL PARAMETERS
Variable Intercept Xl x2 x3 x4
XlSQ
x2SQ MSQ XIX2 x1x3 x1x4 x2x3 x2x4 x3x4 x1x2x3 x1x2x4 x1x3x4 x2x3x4 x1mx4
Parameter estimate
0% (0) (1)
0.1690
1.7 7.0 -2.7 -2.1 -2.0 -1.1 3.7 4.3 1.3 8.1 -7.6 3.2 -1.5 -2.0 -4.5 -3.8 9.7
(2)
<.oooOl 4001 <.OOOl 0.0002
X IO-* x x x x x x
10-4 10-3 10-I 10-4 10-3 10-4
x lo-’ x IO-’
X IO-’ x lo+ x 10-6
x 10-S X lO-5 x lo-’
(3) (4) (1, 1)
(2, 2) (3, 3) (1,2) (I>31 (1,4) (2, 3) (224) (334) (1,2, 3) (1,2,4)
(1,3,4) (2, 3, 4) Cl,& 3>4)
FOR
p value
-5.5 x 10-l 4.7 x 10-3 x 10-I x 10-I
OF OPTIMAL THERAPY SOMAN EXPOSURE
0.6172
<.ooo1 0.0000 0.0927 0.0076 0.0468 0.0085 0.9936 0.0000 0.0079 0.2678 0.0017 0.0500 0.0100
produce innate toxicity (i.e., pi&, ,&&, &&), being negative and statistically significant. Of potential therapeutic significance was the fact that the interaction of all three drugs with GD produced a positive and significant interaction (i.e., 1018283@44)The Nelder and Mead ( 1965) optimization procedure was utilized for this four-chemical interaction study. Results, presented in Table 3, demonstrate the relationships between GD exposure (28-498 peg), the therapeutic drugs, and the percentage of survival. Estimates of the optimal treatment regimens indicate that the antidotal agents can provide various degrees of protection. In addition, other features of this information are evident from the data provided in the following figures. Figure 4 depicts the relationship of therapy optimization and survivability as a function of GD exposure. As indicated, this optimization predicts that use of optimal therapy combinations should provide a large degree of
28.0 31.4 35.2 39.6 44.4 49.8 55.9 62.7 70.3 78.9 88.5 99.3
26.0 26.0 26.6 27.0 27.6 28.2 28.8 29.6 30.3 31.2 32.1 33.0 34.6 36.6 37.9 39.3 40.8 42.4 44.3 46.5 49.6 56.0 100.0 100.0 100.0
111.5 140.0 157.0 176.7 198.0 222.4 250.0 280.0 314.0 352.0 396.0 444.0 498.0
50.0 50.0 50.0 50.0 49.6 49.5 49.3 49.1 48.9 48.7 48.7 48.2 48.0 41.4 47.1 46.8 46.6 46.4 46.3 46.3 46.7 48.1 64.0 64.0 64.0
4.8 4.8 4.9 4.9 4.9 4.9 4.9 5.0 5.0 5.1 5.1 5.2 5.2 5.4 5.5 5.6 5.7 5.8 6.0 6.2 6.3 6.6 7.5 7.5
1.5
99.6 99.5 99.5 99.4 99.4 99.3 99.2 99.1 99.0 98.8 98.6 98.3 97.2 96.4 95.2 93.3 90.6 86.3 80.0 70.5 57.9 43.0 30.5 23.1 16.3
benefit. Indicated in Fig. 4 is the calculated LD50 for GD obtained in these studies as well as the predicted magnitude of shift in GD tox-
1 ZL -----------i ‘O-I
I
0’ rlosEwMAt4 (pg/Iql FIG. 4. Percentage of survival of guinea pigs pretreated with optimal pyridostigmine and treated with optimal atropine/2-PAM treatment against soman poisoning.
SOMAN
PRETREATMENT/TREATMENT
icity (11.9 X LD50) afforded by this optimal therapy. From the information presented in Tables 2 and 3 and Fig. 4, it was apparent that no single dose combination of PYR/ATR/ZPAM was “optimal” for treatment of soman intoxication; optimal doses for these therapeutic modalities were found to be dependent on interactions between the various therapeutic agents as well as on the level of GD exposure. In order to more simply view how the various therapeutic indices changed as a function of the GD exposure level, the effects of the various PYR pretreatment doses on the optimal ATR dose, 2-PAM dose, and survival were assessed as a function of the GD exposure level. The data depicted in Fig. 5 compare the relationship of various PYR doses to the optimal ATR component as a function of the GD exposure level. Across the GD doses and PYR doses tested, ATR remained relatively constant. This constancy was similar to that observed in the previous study in the absence of PYR. It is important to note that though the presence of PYR provided a qualitatively similar ATR optimal dose with respect to GD exposure, there were important qualitative differences; ATR optimum in the absence of
0
loo
200 300
DOSE SOWN
hg/kp)
RG.5.OptimalATR treatment as determined by RSM. Optimal ATR therapy remained relatively constant, independent of the PYR pretreatment dose and soman exposure level. (0) pyridostigmine 0.12 mg/kg; (+) pyridostigmine 0.47 mg/kg; (0) pyridostigmine 1.9 mg/kg; (A) pyridostigmine 7.5 mg/kg.
S247
THERAPY
Km
200 300
DOSE SOMAN hq/kg)
FIG. 6. Optimal 2-PAM treatment as determined by RSM. Optimal 2-PAM therapy increased as a function of increasing soman exposure levels, but decreased as the PYR pretreatment was increased. (0) Pyridostigmine 0.12 mg/kg; (+) pyridostigmine 0.47 mgjkg; (Cl) pyridostigmine 1.9 mg/kg; (A) pyridostigmine 7.5 mg/kg.
PYR at 84.6 pg/kg GD was equal to 168 mg/ kg (see Carter et al., 1985), whereas in the presence of optimal PYR, the ATR optimum = -50 mg/kg (Table 3). A similar comparison with respect to 2PAM is depicted in Fig. 6. At PYR doses of 0.12,0.47, and 1.9 mg/kg, the 2-PAM optimal dose(s) as a function of GD exposure were quite similar; the 2-PAM component increased in a linear manner as a function of GD exposure. At 7.5 mg/kg PYR, however, there was a slight qualitative and quantitative difference in this linear relationship. Overall though, as the dose of PYR increased, less 2PAM was required, but independent of the PYR dose the 2-PAM requirement increased as a function of GD dose. This was also observed in the previous study in the absence of PYR, and as was seen for ATR, PYR dramatically lowered the 2-PAM optimal component. Previous reports have shown ATR to be essential for PYR pretreatment (Dimhuber et al., 1979, Harris et al., 1980). The results presented here substantiate these previous findings; in addition, they provide additional definitive evidence for a positive therapeutic role for 2-PAM as an adjunct to PYR plus ATR therapy. These data indicate that PYR pre-
S248
JONES, CARTER, AND CARCHMAN
treatment dramatically reduces the amount of 2-PAM treatment needed to achieve optimal protection at any level of agent exposure. Although not graphically presented here, the data also supports the converse of this observation, i.e., 2-PAM treatment reduces the amount of PYR pretreatment needed. Figure 7 shows percentage of survival afforded by various pretreatment doses of PYR (in combination with the optimal ATR/2PAM treatment for each PYR dose and CD exposure level) as a function of GD exposure. Also shown are the effects of no therapy as well as optimal ATR/2-PAM without PYR. Similar to what was shown in Fig. 4 (following optimization of all the therapeutic modalities), percentage of survival decreased as a function of increasing GD levels. Increasing the PYR dose, however, resulted in a corresponding rightward shift in the mortality curve; the calculated GD ED50 doses (i.e., the soman LD50) for each PYR curve are also indicated. Since these values are analagous to the treated soman LD50 values traditionally used for assessment of PRs, these findings actually indi-
TABLE 4 COMPARISONOF PYR DOSE, PR, AND PERCENTAGEOF AChEI Dose PYR OWW
% Inhibition AChE
.12 .47 1.90 7.50
5.4 26.4 52.4 79.1
PR 5.6 11.9
~----po.974 Nofe. PRs calculated for each PYR pretreatment dose in combination with optimal ATR/Z-PAM treatment (as assessedby RSM).
cate that, ifATR/2-PAM therapy is optimized, PYR pretreatment affords a dose-dependent increase in the PR. A clearer perspective of this observation is presented in Table 4. Again, AChEI was highly correlated (I = 0.96) with the PYR pretreatment dose. Unlike the results presented in Table 1 (varied PYR pretreatment followed by a single treatment dose of ATR/2-PAM), Table 4 indicates that if ATR/ZPAM treatment is optimized, PYR alfords a highly correlated (r = 0.95) dose (1og)dependent increase in the PR. In addition, under these test conditions, PYR efficacy (assessedas PR) was found to be highly correlated (r = 0.97) with the level of PYR-induced AChEI. DISCUSSION
FIG. 7. Percentage of survival with varying PYR pretreatment doses followed by optimal ATR and 2-PAM therapy. Percentage of survival decreased as the soman exposure level increased, but increased (at all levels of soman exposure) as the PYR pretreatment dose was increased. (a * .) Indicates soman lethality with no therapy; (+) indicates the soman LD50 for each PYR pretreatment group; (--) pyridostigmine 0.12 mg/kg; (---) pyridostigmine 0.47 mg/kg; (-) pyridostigmine 1.9 mgjkg; (- - -) pyridostigmine 7.5 mpB; (- v-) no pyridostigmine/optimal ATR/Z-PAM.
Further information has been provided concerning the use of RSM to estimate therapeutic optima. In this example, an additional antidotal treatment (i.e., pyridostigmine pretreatment) was utilized compared with the antidotal data presented in the previous paper (Carter et al., 1985). Therapeutic optimal search procedures were utilized here to evaluate the entire experimental region; this was equivalent to evaluating all possible treatment
SOMAN
PRETREATMENT/TREATMENT
combinations and choosing the maximum. This procedure is more precise than altering fixed percentages of one drug in a combination, or using PRs to evaluate a relatively small number of treatment combinations. In addition, RSM utilizes all the experimental data, not just singular points (i.e., ED50). In a multidimensional study such as the one presented, (i.e., five dimensions), techniques that require limited data procedures (i.e., PR) or which require visualization of the response (Berry et al., 1971; Boskovic et al., 1984; Boskovic and Stern, 1970; Clement and Lockwood, 1982; Dimhuber et al., 1979; Faff et al., 1976; Jones et al., 1984; Koplovitz, et al., 1984; and Green, 1984) are inherently restricted in their ability to accurately assess the full dose-response surface. Use of the prediction equation to estimate the response of any (or all) treatment combination(s) within the experimental region makes RSM a powerful approach to analyze such complex experimental procedures. The estimation of the parameter (i.e., p) terms for the individual compounds and their interactions provides additional information. Though PYR pretreatment produced a large, positive, and statistically significant therapeutic response (as indicated by &; Table I), it cannot do so indefinitely (i.e., &, a reflection of the PYR toxicity). Unlike the previous study, 2-PAM by itself did not produce a significant therapeutic response; this is related to the fact that the experimental conditions (e.g., GD dose) covered a range > five times that used in the previous study. Further analysis revealed that PYR pretreatment reduced the ATR and 2-PAM requirements compared to the optimum observed in the previous study. This reduction for ATR was -2/3, and was relatively constant as a function of GD dose. In contrast, the 2-PAM dose optimum, though reduced, was found to vary inversely as a function of PYR pretreatment dose and progressively as a function of GD dose. This latter finding was consistent with the data from the previous study, i.e., ATR optimum was constant, whereas 2-PAM optimum increased as
THERAPY
S249
a function the GD dose. Overall, PYR pretreatment reduced the ATR/2-PAM optimum treatment, but did not appear to change the relationship between these two agents. The relationship of “efficacy” to PYR dose was highly correlated only if RSM was used to optimize the adjunctive ATR/ZPAM treatment doses. This was shown in Tables 1 and 4 and Fig. 7, with the use of PRs to evaluate PYR efficacy across a wide range of GD exposure levels. If percentage of survival following a single exposure level of GD was used as an alternative criteria (rather than PRs) for evaluating PYR efficacy, RSM analysis also provides the information necessary to make such an assessment. Although this approach was not specifically addressed in this paper, the information presented in Fig. 7 substantiates this claim. If percentage of survival were assessed(as the dependent variable) following a single GD exposure level (e.g., 200 clg/kg) it can be seen that a dose-related increase in percentage of survival occurs with increasing PYR pretreatment levels. Although the data assessment is not provided in this paper, this dosedependent response was found to be highly correlated (r = 0.99), (D. E. Jones, 1985, unpublished observations) by nonlinear regression analysis (NIH, 1980). This example of an alternative analytical approach for assessing PYR efficacy was provided in order to further emphasize the utility of RSM analysis. Similar to what was found with therapeutic efficacy (PRs) and PYR dose, PYR efficacy was also highly correlated with AChEI, but only if RSM was used to optimize the adjunctive ATR/2-PAM treatment. The biological significance of this observation is meaningful in that one can infer that PYR pretreatment, by reducing AChE, is providing a protective mechanism against GD-induced lethality. Although these observations and inferences by no means establish a definitive cause-effect relationship, they are nonetheless in agreement with the proposed mechanism of action for PYR as originally postulated by Berry and Davies (1970). Unfortunately, the
S250
JONES, CARTER,
role of whole blood AChE is poorly understood with regard to nerve agent intoxication. However, PYR-induced AChEI in other tissues has been shown to closely parallel that observed in whole blood (Harris et al., 1980; Hey1 et al., 1980; Karlsson et al. 1984). The approaches and findings presented here, in combination with these earlier observations, will hopefully serve to both reinforce and expand the mechanistic interpretations concerning the therapeutic role of pretreatment carbamates. In summary, the two studies presented here demonstrate the utility of RSM to analyze complex studies and to arrive at therapeutic optima. They also provide provocative insights into possible interactions and mechanisms. In addition to therapeutic optimization, RSM also provides an assessment of the entire doseresponse surface for all inherent variables (dependent and independent). Such features greatly expand the range and significance of information that can be derived from a single set of experimental data. RSM analysis of the data provided in these papers provides a means to accurately predict the therapeutic efficacy afforded by all possible combination of PYR/ ATR/ZPAM against any GD exposure level. Such information becomes extremely valuable for formulating rational treatment hypotheses, especially if used in conjunction with information provided by other experimental procedures (e.g., maximum tolerated doses of therapy compounds or anticipated level of GD exposure). Since candidate nerve agent therapies cannot be evaluated against nerve agent in a controlled clinical (human exposure) situation, potential therapeutic dosage regimens must be selected on the basis of nontoxic (e.g., nonincapacitating doses). Only by testing these dosages in nerve agent-exposed animal models can any meaningful assessment (albeit comparative) be made of their therapeutic potential. It is therefore essential that the most accurate and informative means available be used to make these assessments. RSM procedures, as oulined in these two papers, provide
AND CARCHMAN
the most accurate and comprehensive means yet reported for making these complex assessments of nerve agent antidotal efficacy. REFERENCES BERRY, W. K.. AND DAVIES, D. R. (1969). The use of carbamates and atropine in the protection of animals against poisoning by 1,2,2-trimethylpropylmethylphosphonofluoridate. Biochem. Pharmacol. 19,921-934. BERRY,W. K., DAVIES, D. R., AND GORDON, J. J. (197 1). Protection of animals against soman (1,2,2-trimethylpropylmethylphosphono-fluoridate) by pretreatment with some other organophosphorus compounds, followed by oxime and atropine. Biochem. Pharmacol. 20, 125-134. BOSKOVIC, B., KOVACEVIC, V., AND JOVANOVIC, D. (1984). PAM-2 Cl, HI-6, and HGG-12 in soman and tabun poisoning. Fundam. Appl. Toxicol. 4, S106-S 115. BOSKOVIC, B., AND STERN, P. (1970). Protective effects of oximes and cholinolytics against soman poisoning. Arch. Toxicol. 26, 306-3 10. CARTER, W. H., JR., STABLEIN, D. M., AND WAMPLER, G. L. (1979). An improved method for analyzing survival data from combination chemotherapy experiments. Cancer Res. 39, 3446-3453. CARTER, W. H.. JR., JONES, D. E., AND CARCHMAN, R. A. (1985). Application of response surface methods for evaluating the interactions of soman, atropine and pralidioxime chloride. Fundam. Appl. Toxicol. 5, OOO000. CLEMENT, J. G., AND LOCKWOOD, P. A. (1982). HI-6, an oxime which is an effectiveantidote of soman poisoning: a structure-activity study. Toxicol. Appl. Pharmacol. 64, 140-146. DIRNHUBER, P., FRENCH, M. C., GREEN, D. M., LEAD BEATER, L., AND STRATTON, J. A. (1979). The protection of primates against soman poisoning by pretreatment with pyridostigmine. J. Pharm. Pharmacol. 31, 295-299.
FAFF, J., BORKOWSKA, G., AND BAK, W. (1976). Therapeutic effectsof some cholinolytics in organophosphate intoxications. Arch. Toxicol. 36, 139- 146. FINNEY, D. J. (197 1). Probit Analysis, 3rd ed. Cambridge Univ. Press, Cambridge, Mass. GREEN, M. D. (May 1984). Guinea pig model for correlating efficacy with pralidoxime chloride (2-PAM) plasma levels. Proc. Annu. Chem. Dej Biosci. Rev. pp. 71-95.
HARRIS,L. W., STITCHER,D. L., AND HEYL, W. C. (1980). The effectsof pretreatments with carbamates, atropine and mecamylamine on survival and on soman-induced
SOMAN
PRETREATMENT/TREATMENT
alterations in rat and rabbit brain acetylcholine. Life Sci. 26, 1885-1891. HEYL, W. C., HARRIS, L. W., AND STITCHER,D. L. (1980). Effects of carbamate on whole blood cholinesterase activity: Chemical protection against soman. Drug Chem. Toxicol. 3(3), 3 19-332. JONES,D. E., KOPLOVITZ, I., HARRINGTON, D. G., HILMAS, D. E., AND CANFIELD, C. J. (May 1984). Models for assessingefficacy of therapy compounds against organophosphates (OP). Proc. Annu. Chem. Del: Biosci. Rev., pp. l-37. KARLSSON, N., LARSSON, R., AND Puu, G. (1984). Ferrocene-carbamate as prophylaxis against soman poisoning. Fundam. Appl. Toxicol. 4, S 184-S 189. KOPLOVITZ, I., JONES,D. E., HARRINGTON, D. G., HILMAS, D. E., AND CANL~ELD, C. J. (May 1984). Models for assessingefficacyof pretreatment compounds against organophosphates (OPs). Proc. Annu. Chem. Def: Biosci. Rev., p. 39-69. NELDER, J. A., AND MEAD, R. A. (1965). A simplex
THERAPY
S251
method for function minimization. Cornput. J. 7,308313. NIH (National Institutes of Health) (1980). Regression analysis. In Prophet Statistics: A User’s Guide to Statistical Analysis on the Prophet System. U.S. Department of Health and Human Services, Public Health Service, National Institutes of Health, Publication No. 80-2 169. Bolt, Baranek and Newman, Inc., Cambridge, Mass. SIAKOTOS, A. N., FILBERT, M., AND HESTER, R. (1969). A specific radioisotopic assay for acetylcholinesterase and pseudocholinesterase in brain and plasma. Biochem. Med. 3, l-9. STEEL, R. G. D., AND TORRIE, J. H. (1960). Analysis of variance I: The one-way classification. In Principles and Procedures of Statistics (Steel, R. G. D. and Tonie, J. H., eds.),Ch. 7, pp. 99-131. McGraw-Hii, New York. TALLARIDA, R. J., AND JACOB, L. S. (1979). The Dose Response Relation in Pharmacology pp. I- 17 and 8% 110. Springer-Verlag, New York.