DISTRIBUTION OF HEOD (DIELDRIN) IN

MAMMALS:
Ill. TRANSPORT--TRANSFER
F. T. LINDSTROM, 1 J. W . GILLETT, 2 and S. E. RODECAP 3
Oregon State University, Corvallis, Oregon 97331

A mathematical model simulating the blood transport and tissue residue distribu-
tions of the highly toxic and highly lipid soluble pesticide dieldrin in mammals is
presented. This model is a significant improvement over our previously published
preliminary model for dieldrin distribution in mammals. The assumptions and working
hypotheses of the model are presented and used in generating a set of differential
equations based upon mass balance principles. Two simulation cases are examined.
The first simply demonstrates the gross features of: 1) Transport limiting conditions; 2)
equal transport-equal membrane transfer conditions, and 3) membrane transfer limit
conditions. The second studies a single tissue (the blood-brain barrier case) example of
the above mentioned conditions. All simulations made were conducted for a hypotheti-
cal mature male rate of the average Wistar | type eating food ad lib.

In Part I of this series (Lindstrom et al. 1974) a mathematical model simulating

the transport of the drug dieldrin (HEOD) 4 in mammalian tissue systems was put
forth. This preliminary model was based mainly on assumptions of (1) almost
exclusive lipid phase transport of the drug due to its extremely high lipid/water
partition coefficient, and (2) extremely rapid transfer of dieldrin from blood lipid to
lipid in various tissues (flow-limited model, Bischoff and Dedrick 1968). Although
toxicological and pharmacological literature and data on dieldrin distributions sup-
ported these and the other assumptions, assumption (2) was actually a working
hypothesis as to what may be taking place at the capillary-tissue cell boundary,
where explicit information is lacking.

As shown in the application of the model simulations (Lindstrom et al. 1974 and
1975) reasonable values for adipose lipid-phase dieldrin residues are well-
approximated in the mature male rat and human being. The general validity of the
model and of the hypothesis embodied in assumption (2) do not appear to extend to
all situations, for realistic simulation of all tissue regions is not achieved under

1Associate Professor of Statistics and Mathematic~

ZAssoeiate professor of Agricultural Chemistry; Present Address: National EcologicalResearch Laborat-
ory, U.S. Environmental Protection Agency, Corvallis, OR.
3Research Assistant in Mathematics
4Dieldrin will be used to represent HEOD (1,2,3,4,10,10- hexachloro-6,7-epoxy- 1,4,4a,
5,6,7,8,8a-oetahydro- 1,4:5,8-endo-exo- dimethanoaphtalene) which constitutes at least 85% of the
active in~redi_ents in technical dieldrin.

Archives of EnvironmentalContamination 257

and Toxicology,Vol. 4, 257-288(1976)
9 1976by Springer-VedagNew York Inc.
 258 F. T. Lindstrom et al.

conditions of chronic exposure for extended periods. Characteristically, the first

form of the model (Lindstrom et al. 1974) yields identical lipid-phase concentrations
in all tissue regions. Moreover it can not simulate apparent compartmentalization
(Barron and Walton 1971) in adipose tissue.

Recent evidence (Zabik and Schemmel 1973, Barron and Walton 1971, Yadrick
et al. 1971 and unpublished studies by Gillett 1974) seems to indicate that several of
the compartments (e.g., adipose and brain) may not be flow limited but rather
membrane transport limited. This certainly comes as no surprise for it is well known
that lipid turnover in the brain is low while the lipid turnover in the liver is high.
Other organs (e.g., kidney) probably lie somewhere in between. The effect then of
including membrane transport kinetics as well as flow dynamics is to produce a
mixed pharmacodynamic model for dieldrin distribution in mammalian tissue sys-
tems (Dedrick et al. 1973).

The assumption that dieldrin is associated in its pharmacokinetic distribution

almost exclusively with lipid, of course, is based not only on the measurable
partition coefficient (Wauchope 1971), but also takes into consideration the pre-
sently insufficiently precise measurements of binding to proteins and/or lipids in
lipoprotein mixtures (e. g., membranes). Distributions of the various members of the
DDT family (e.g., DDT, DDE, DDD) follow diverse patterns not exactly parallel to
dieldrin and possibly related to specific binding to phospholipids (Haque et al.
1973). Hence the model for pharmacokinetics of highly lipid soluble drugs should
take into account the likelihood that such interactions might contribute significantly
to compartmentalization.

Transport--transfer model system

Figure 1 shows the revised conceptual scheme for the transport of highly lipid-
soluble drugs in man, derived basically from Bischoff and Brown (1966). Due to the
very large number of symbols (e.g., for masses, flows, concentrations, etc.) used in
Figure 1 and throughout the model equations, the reader is urged to consult the
nomenclature section at the end of this paper for lists of symbol definitions and
meanings.

The assumptions on which this model system is based are:

(1) Dieldrin is transported both dissolved in the lipid interior of the chylomic-
ron and bound to lipo-protein surfaces in the blood (e.g., bound to RBC,
serum lipo-protein) (Moss and Hathaway 1964, Pocock and Vost 1973).
Dieldrin may be bound to lipo-p~oteins in the intracellular portion as well;
(2) Dieldrin transfer across the blood-tissue boundary may or may not be rapid.
The speed is highly compound/tissue specific, and the various passive and
active elements in the transfer process will be represented by effective
forward and reverse lipid transfer rates;
 ), 111,,
oo o~
-I -I
o
-I -t
"T1
~z
"11 t"~ ~g
N:.~_
"I'I M ~g
_~.~
3g
g~
_o
1-
__.o __.~
-4 ~ r
~v -4
j I1] (1]
"11 ~
~ m "11
~ +
~ g
0
0
6g~ lOpOl~ uo.l]nq.Ialst.(] u.laplo~( ]
260 F. T. Lindstrom et al.

(3) Metabolism of dieldrin is carried out exclusively in the liver, and hepatic
microsomal induction is allowed;
(4) Dieldrin, being tied to lumen lipids, is almost totally absQrbed from the
intestines via lipid absorption (Tidwell 1958) and not readily, excreted in
feces and urine;
(5) Tissues can be assigned to large compartments of generally similar constitu-
tion;
(6) Each compartment behaves as a well-stirred chemical reactor (both capillary
+ extracellular material and intracellular compartments);
(7) The complete open system is representable by a system of mass balance
equations;, and
(8) The effective blood lipid mass flow (q~x), the "effective" forward and
reverse lipid transfer rates kx+ and kx, and the "effective" compartment
lipid mass (~x) coefficients are once continuously differentiable functions
of time (0 ~< t < oo).
Note that these assumptions admit to the possibility of being relaxed, in certain
instances, without leading to disintegration of the model. Assumption (8) allows for
time- or status-dependent changes in compartment sizes and flow rates ( e . g . , a
growing animal ingesting dieldrin).

The complete drug mass balance equation for each compartment has the form:
first for the extracellular compartment (see Figure 2)

(Blood Capillaries f Extra Cellular Fluid I ~.. Qx

(interstitial Fluid .- -- (Blood)

t
Kx_ Kx+

Intra Cellular Fluid

(Tissue)

Excretion Metabolism
(if any) (if any)
Figure 2. Schematic of a single large tissue region showing its intra- and extracellular fluid
portions.
 Dieldrin Distribution Model 261

dUxB UB
dt Cx
~J,~ m B

[rate of accumulation, of drug] [delivery of new drug via blood]

(free + bound) (free + bound)

Ux T
+ kx
-~J~xT

[transfer of stored drug in tissues to blood]

(free + bound)

UxB
--kx + ~.~x B

[transfer of drug from blood to tissues]

(free + bound)

Ux B
- r ~xB
[removal of drug in tissue blood]
(free + bound)

and for the tissue compartment (intracellular) itself

dUxT UxB
dt kx+ ~2~xB
[rate of accumulation of drug] [forward transfer of drug
(free + bound) from tissue blood]
(free + bound)

UxT
-- k X
-- ~J~xT

[backward transfer of drug from tissue blood]

(free + bound)

UxT
-- R ~ x l
~xT
[excretion (if any)]
 262 F. T. Lindstrom et al.

UxT
-- F x SJ~xT

[metabolism (if any)]

Using this format then as the guide, the complete drug mass balance system
(assumption (7)) has the form shown in Table I.

The mass balance system may be cast into matrix equation form, such as:

dt Ao,KU + ~ , U(o) = 5 , (2)

where
-- (UGIB,UGIT,ULB,ULT,UKB,UuT ,UKT,UBRB,UBRT,
UFB,UFT,UMB,UMT,UOTB,UoTT,UB)T,

and ~ = (0~,o,o...o) r.

The 15 x 15 transport-transfer matrix AQ K, contains 225 entries, but has only 44

non-zero elements as is shown by an explicit listing of AQ ~ in Figure 3.

Because equation system (2) is a non-linear coupled system of first order

integro-differential equations [the integral and non-linearity enter via the definition
of F, (LINDSTROM e t a l . 1974) ] no exact closed form solution to the general case
of this system appears possible at this time. Also, the character of the solution,
which can be shown to exist (DAVIS 1962), is highly dependent upon the growth
characteristics of the effective compartment masses:02xB and~-~xT as well as the
effective compartment blood flow rates ~bx. However, we choose to simulate the
system (approximate a solution) by the same method used in LINDSTROM e t a l .
(1974 and 1975).

S i m u l a t i o n o f t h e m o d e l s y s t e m . Let U(t,) = U(nAt) = U(n), n = 0, I, 2 . . .

denote the column matrix (15 x 1) of dependent variables all evaluated at time t,,
depending upon the n-value. The numerical simulation is now accomplished by
using a backward difference in time on At (the difference approximation of dt ).
This is one of the classical Pade' approximations, which for [' > 0 yields an
unconditionally stable difference scheme (Varga 1962). It is also O(At 2) accurate
locally and O(At) globally (Varga 1962). Thus
~(~ + 1 )_~(~)
At = A ~ K u(n+l) + ~ (n+ 1),
(3)
U(o) = o,

n = 0,1,2 . . . .
 Dieldrin Distribution Model 263

Table I. Mass balance equations.

Compartment Equation

d U G I B - ~GI ( UB -- UGIB )
Blood
dt B ']RGIB

. UGIB UGIT
m),
- kGl+ ( - ~ - ~ 7GI (la)
GI

dUGIT UGIB UGIT

Tissue dt - kGI+ ( '2R'GIB -- ")'GI ~ )

ULT
+/~ ( ~ + r - - ),
sLT

dULB UC~B UB ULB

Blood - -dt - -- (r + ~L)
~GI '~~RGIB
4 - r '2R~B ~LB

( ULB ULT )
-- kL+ 'Jg~LB -- ~/L 'Jg~LT '
Liver (Ib)
dULT ( ULB ULT
Tissue dt - kL+ 'ffdLB -- "/L - ' ~ )

ULT

dUxB UB UxB )
Blood
dt - ~bx ( ' ~ - ~ ~xB

X ~ . UxB UxT
Kidney, -kx+ l-~xB -- "/x ~ )
Brain, (1 c-g)
Fat,
Muscle, dUxT - kx+ " UxB UxT UxT
, Other, Tissue dt ('-~'~xB -- "Yx , ~ x T ) - Rexx ~9~xT '

General Blood Pool dUB _ (~bGi +~bL) ULB 4- ~bk UK-.-..~

B + ~bBR UBRB
dt ;lJ~L B '~92KB '2)2BRB

UFB + ~M UMB + ~5OT UOTB

4- r '.~FB '~RMB '~0~OTB

UB
- q~B (lh)

where ~ S - ' ~ G X + ~ L * ~ K + ~ B R + ~ F + ~ M +~oT"

264 F. T. Lindstrom et al.

~" 0 0 0 0 0 0 0 0 0 0 0 0 ,.~o ~" 0

0 0 0 0 0 0 0 0 0 0 0 0 o o
I

0 0 0 O 0 0 0 0 0 0 ~ 0 0 0
I

~. 0 0 0 0 0 o o o o o o o
i
.b
o 0 0 0 0 0 0 0 0 0 0 0 0 0
I

~'l | ~
0 0 0 0 0 0 o o ~ o o o o ~]~
I

0 0 0 0 0 0 0:| if| 0 0 0 0 0 0 0

§
r~ 0 0 0 0 o o § ~. ~| | o o o o o o ~
'

0 0 0 0 ~ ~ 0 0 0 0 0 0 0 0 0

0 0 0 0 ~ ~~ ~~, 0 0 0 0 0 0 0 0 ~I~
(J

,~ o ~ ~( 0 0 0 0 0 0 0 0

m o o
§
#~,
]i' ~,#, o o o 0 0 0 0 0 0 0 ~ 9

r~
0 0 0 0 0 0 0 0 0 0 0 0 0
I
~~ §
-|
~- ~ 0 0 0 0 0 0 0 0 0 0 0 0

k. ' /
II

d
 Dieldrin Distribution Model 265
m

Upon solving for the unknown U(n+O in matrix equation (3) obtains

~(n+t) = II_A~,)K At}--l(~(n) + ~ ( n + l ) A t ) , (4)

n = 0,1,2 . . . .

It is to be noted here that the superscript (n) on the transport-transfer matrix Ao, K
means that all explicit time dependencies (e.g., flow coefficients ~bx) are evaluated
at time tn~.l and all implicit time dependencies (e.g., dependent variables) are
evaluated at time t n. This effectively linearizes the system and renders it solvable at
each time step. To be sure, there are numerous other difference methods available
for use; however, none of the other simulation schemes will yield any different
character of the approximate solution from the simple method used here (Richtmyer
and Morton 1967). A discussion of the accuracy of this numerical simulation method
is being prepared for discussion elsewhere (Rodecap and Lindstrom 1975). Suffice it
to state here that the accuracy obtainable with this numerical method is much more
than adequate to cover the experimental uncertainties.

The numerical simulation of the liver microsomal protein mass and microsomal
lipid are taken directly from Lindstrom et al. (1974). Hence

u(n)
LT
p~n) + kcpi Atp}O) + /~omp
~'Pi :).)~(n)
p.(n+ 1) = LT , i = 1,2... 5, (5)
t 1 q- kcp i At

p(n+l) = E p(n+l) _p(n+1)

Ms i --3 ' (6)
i=l

,322(n+D = 1 p ( n +
Ms ~ -Ms 1), a a constant, (7)

~l~(LnT1 ) = ql?(n+l) 4" ~1~ 4- q- ,

"-" Ms N ~J'd~Cy :l~ Mc (8)
(assumed constant)

F(n+l)= F(n+l)Ms+ P(cn;I) =E KDi p(n+l)i (9)

(active protein)
266 F. T. Lindstrom et al.

NOMENCLATURE

Subscripts.

G.I. = Gastrointestinal
L = Liver
K = Kidney
BR = Brain
Compartment x F = Adipose
M = Muscle
OT = Other
EB = General b l o o d pool (chambers o f heart, lungs, large veins
and arteries, etc.)
B = Tissue blood
T = Tissue
BI = Bile
EBL = Blood lipid

TP
EBP = Blood lipo-protein
]~BS = Blood saline
TL = Tissue lipid
= Tissue lipo-protein
1 General blood pool

TS = Tissue saline
BL = Tissue b l o o d lipid
BP = Tissue blood lipo-protein
BS = Tissue b l o o d saline
Ms = Microsomal
Cy = Cytoplasmic
N = Nuclear
Me = Mitochondrial

Masses.

UxB = Dieldrin mass (g) in the extracellular portion o f compartment x.

UxT = Dieldrin mass (g) in the intraceUular portion o f compartment x.
Pi = ith protein mass in liver cells (g).
Lipid, lipo-protein, or saline solution mass in either the extracellular (blood
portion) or intracellular (tissue portion) o f compartment x (g).
 Dieldrin Distribution Mode[ 267

MMs, Mcy, MN, and MMc = Respectively the microsomal, cytoplasmic, nuclear, and
mitochondrial lipid masses of the liver cells (g lipid).

Mass flows.

qx = Blood flow (g blood/hr) through compartment x.

qBI = Bile flow (g bile fluid/hr) from liver to gut lumen.
Rex = Lipid equivalent of kidney excretion rate of dieldrin (g lipid/hr).

Membrane transfer coefficients.

kx+ = Lipid equivalent of "effective" forward transfer of dieldrin across cell mem-
branes in compartment x (g lipid/hr).
kx_ = Lipid equivalent of "effective" backward transfer of dieldrin across cell mem-
branes in compartment x (g lipid/hr).

Binding coefficients.
xBP = Average value of the lipid equivalent of lipo-protein binding of dieldrin in
the blood subcompartment of compartment x (g lipid/g lipo-protein)
XTP = Average value of the lipid equivalent of lipo-protein binding of dieldrin in
the tissue subcompartment of compartment x (g lipid/g lipo-protein)
~ 8P = Average value of the lipid equivalent of lipo-protein binding of dieldrin to
lipo-protein surfaces in the general blood pool mass (g lipid/g lipo-protein)
8I = Average value of the lipid equivalent of lipo-protein binding of dieldrin to
bile matter (g lipid/g bile matter)

Partition coefficients.
xBs, ~ XTS, and ~ ~:8s = Respectively the average true solution partition coefficients
(25~ of dieldrin for a water/oil system in subcompartments blood saline and
tissue saline portions of compartment x and the saline portion of the general
blood pool ((g HEOD/g oil)/(g HEOD/g H2 O)).

Fractional mass coefficients (Tables II and III in text).

fxB, fxT = Respectively, g of lipid per g blood or tissue in each subcompartment
of compartment x (g lipid/g material).
fxL, fxP, fxs = Respectively, the lipid fraction (g lipid/g whole subcompartment), the
Lipo-protein fraction (g Lipo-protein/g whole subcompartment), and the
saline fraction (g saline solution/g whole subcompartment).

Miscellaneous coefficients.
7x = kx_/kx+ (pseudo-partition coefficient).
 268 F.T. Lindstrom et al.

/~Pi = ith liver cell microsomal induction coefficient ( g protein )

g dieldrin hr
g lipid
kepi = i TMliver cell microsomal catabolism coefficient (L/hr).

KDi = i th liver cell proportionality coefficient for dieldrin-metabolizing proteins.

0 = lumen-G.I, tissue transport efficiency coefficient (dimensionless).
= dieldrin source function (g dielddn/hr).
P = lipid phase "effective" dieldrin destruction rate coefficient (g lipid/hr).
Fp = dieldrin flux to portal system (g dieldrin/hr).
FTD = dieldrin flux to thorasic duct (g dielddn/hr).

Discussion
In order to be able to use this model in a realistic situation (e.g., laboratory rat of
a certain size being dosed with dieldrin) certain a p r i o r i physiologic and biochemi-
cal data are needed. For example, to calculate the effective subcompartment masses
a p r i o r i knowledge of the subcompartment lipid mass, tissue protein mass, lipid-
water-dieldrin partition coefficient, lipo protein binding coefficient, and the total
saline solution mass is necessary for every subcompartment. In addition to having
all this data we must have a p r i o r i knowledge of the blood flow rates through each
compartment as well as the blood volume of every compartment. Quantitative
knowledge of dieldrin transport in the blood and blood lipo protein binding coeffi-
cients as well as partition coefficients, are needed. The following section outlines
how the effective mass and flow coefficients are obtained from the above mentioned
physiologic data.

Effective flow and mass coefficients. I. The effective subcompartment masses (g

lipids), based upon the assumption of extremely rapid, linear binding, equilibrium
kinetics, are defined as:
Blood ~ x B = MxBL + MxBP ~xBP + Mxes l
~-~xBS'
I
Tissue ~ x T MxTL + MxTP ~xTP + MxTS ~xTS'
where x = G.I., liver, kidney, brain, depot fat, muscle and other. The remaining
effective blood mass (heart chambers = lung chambers = large veins and arteries) is
defined as
1
~'IR2~B = My-BE + M~Bp ~ . B P + M2:BS ~EBS

II. The effective blood lipid flow coefficients (g lipid/hr) through the tissue blood
capillaries are defined as

1
~x = qx(fBL + fBP ~BP + fas ~ZaS )'
 Dieldrin Distribution Model 269

with x assuming the same values as listed above under I, and

1
~bB = qB(fBL + fBP ~ZBP + fBS ~ )

is defined as the effective lipid cardiac output with

qB = ~ qx, (grn blood/M)

x
defined as the true cardiac output.
III. The effective bile lipid flow (g lipid/hr) is defined as (when operative)
~BI = ~'~BI " qBI,
where

qal = true bile flow (g bile fluid/hr).

Physiologic and biochemical data. Tlie hypothetical cases to be discussed will

make use of the physiological and biochemical data summarized in Tables II, III,
and IV. In Table II, physiologic data for a mature 300-g male rat eating food and
water ad lib. is presented. The lipid masses M~aL or MxrL are calculated by the rule

MxBL t
{MxTL,} = { fxBLfxTL} ~MxT)
tMxBt

Note that the percentage of the whole compartment which is estimated (Krueger
1971) to be extracellular is shown in parenthesis immediately to the right of its
numerical'value ( e . g . , MGIB = (15.8)(30%) = 4.74 gins). The blood lipid flow rates
are calculated by the following rule

Qx -- (qx) " (f~aL), each x.

With these important physiologic parameters thus calculated, it is now possible to

estimate the effective mass and flow coefficients via the rules previously given in
the effective flow and mass coefficient section. The effective compartment mass and
blood flow parameters are summarized in Table III.

It is possible to find published values for most of the tissue protein and tissue
saline fractions fxP and fxs; but we were unable to find any published values for the
lipid equivalent of tissue protein binding of dieldrin with the sole exception of
dieldrin binding to RBC (Dale et al. 1966, Jakubowski and Crowder 1973, Morgan
et al. 1972, Moss and Hathaway 1964,, Pocock and Vost 1973). It appears that
dieldrin binds reversibly to the RBC very rapidly (of the order of seconds for 90%
equilibrium) and has a binding magnitude of about 40 to 60% to RBC. The
remaining plasma dieldrin seems to be in the four mg lipid/g blood saline fraction
(serum). Thus, using the 50% value for RBC binding of dieldrin in rat blood, the
value of 0.01 shown in Table III is obtained for ~xBP in the blood compartments.
 Table 11. Compartment mass and blood flow parameters for a hypothetical 300-g mature male rat.

Blood flow Blood lipid

Lipid fraction rate flow rate
Mass fx Lipid mass qx qx
Symbol Mx ( g lipid i MxBL, MxTL i g blood ,~ ( g blood lipid )
Compartment x (g) " ~ " (g lipid) "---h'T~" hr

GI Blood GIB 4.74 a (30%) 0.004 b 7% 0.02 450.0 1.80

Tissue GIT 11.06 0.10 Av 1.11

Liver Blood LB 2.95 (25%) 0.004 6% 0.012 90.0 0.36

Tissue LT 8.85 0.08 Av 0.71

Kidney Blood KB 0.81 (30%) 0.004 5% 0.003 360.0 1.44

Tissue KT I. 89 0. 07 Av 0.13

tO Brain Blood BRB 0.19 (10%) 0.004 8% 0.001 340.0 1.36

--d
Tissue BRT 1.71 0.09 Av 0.154

Fat Blood FB 2.16 (4%) 0.004 63% 0.009 100.0 0.40

Tissue FT 51.84 0.64 Av 33.18

Muscle Blood MB 12.0 (10%) 0.004 4% 0.048 240.0 0.96

Tissue MT 108.0 0.05 Av 5.40

Other
(skin, Blood OTB 6.0 (10%) 0.004 10% 0.024 130.0 0.52
etc.) Tissue OTT 54.0 0.11 Av 5.94

General
blood pool ~B 23.4 0.004 0.09 1710.0 6.84

aThese masses are based upon an estimate from Spector (1966) and Dedrick et al. (1973) on the percentage of whole tissue that is blood
capillary plus extra cellular fluid. The sum MGIB + MGIT = 15.8 g of tissue in accord with Lindstrom et al. (1974).
bThe blood lipid fraction value f~ B = 0.004 was assumed for lack of knowledge of lipid fraction on the sum of the blood capillaries plus
extra cellular fluid material in all the compartments. Also, it is assumed that fr~ B = fr.BL in this work.
 Table IlL Effective compartment mass and blood flow parameters for a hypothetical 300 g mature male rat.

Protein
Lipid fraction Saline Effective
fraction fxP fraction Effective blood lipid
fxL ( gprotein fxS lipid mass flow: ~bx
Symbol ( g lipid " g tissue ) [ R solution ~ ~ x B or ~)J~xT g lipid
Compartment x " ~ ) (RBC) " g tissue " (g)

GI Blood GIB 0.004 0.4 0.596 0.040 3.60

Tissue GIT 0. 10 1.110

Liver Blood LB 0.004 0.4 0.596 0.024 0.72

Tissue LT 0.08 0.710

Kidney Blood KB 0.004 0.4 0.596 0.006 2.88

Tissue KT 0.07 0.130
I'O
"--4
Brain Blood BRB 0.004 0.4 0. 596 0.002 2.72
Tissue BRT 0.09 0.154

Fat Blood FB 0.004 0.4 0.596 0.018 0.80

Tissue FT 0.64 33.180

Muscle Blood MB 0.004 0.4 0.596 0.096 1.92

Tissue MT 0.05 5.400

Other Blood OTB 0.004 0.4 0.596 0.048 1.04

Tissue OTT 0.11 5.940

General
blood pool ZB 0.004 0.4 0.596 0.187 13.68

Note: The lipid equivalent of protein binding ~ x a P = 0.01 ( g lipid/g protein) and ~ x T a = 0.0. This latter value is assumed in all simulations
in this paper since we were unable to find any definite literature values. ~xBS and ~xTS are assumed to be of the order of 3 x 10 s for HEOD
in a peanut oil-H20 system at 25~
272 F. T. Lindstrom et al.

In keeping with a philosophy of a priori parameter estimates, the " e f f e c t i v e "

subcompartment tissue lipid masses were calculated by setting ~xPa = 0 in the
effective mass equations. When suitable dieldrin-tissue protein binding experiments
have been performed, then those binding values could be used and the effective
subcompartment tissue lipid masses recalculated. Mathematically, it can be shown
that the net effect of setting ~XPR = 0 in the calculation of the effective subcompart-
ment tissue lipid masses simply speeds the approach toward the "steady state" by
shortening the residence time for dieldrin in any one compartment (e.g., the ~bx/~xT
ratio is larger).

The important biochemical data for dieldrin metabolism in the liver cells
(Lindstrom et al. 1975) is summarized in Table IV.

The following cdses are developed to explore hypotheses on the range or relative
levels of effective transfer coefficients in this revised model which would be
consistent with data on tissue residues and exposures to dieldrin. Before proceeding
further, it is important to point out that, as the level of sophistication of the model
increases, there are geometrically increasing needs both for data and validating
experimentation and for more sophisticated exposition of significant subsystems
within the model. These increasing complexities, such as an intimate knowledge of
changes in physiologic function (e.g., blood flows) with growth of diurnal activity
patterns, are being approached in a function-by-function basis. We must acknow-
ledge the possibility for the interaction between systems to change as these systems
are viewed at different levels or within different time frames. Pragmatically however
we have elected to operate on discrete portions and specific assumptions of the
model in a step-wise manner.

Case 1. Transport-transfer limitation extremes. The revised model can be

employed to expose the patterns of pharmacodynamic residue disposition for a given
chemical and species by simulation computer runs employing available physiologic
data and subordinate assumptions (Tables II to IV) combined with assumptions
delineating these patterns for dieldrin in a hypothetical 300-g male rat. These
effective lipid masses and blood lipid flows are assumed constant over time, with
the only time-variant parameters being hepatic lipid in response to microsomal
induction (tip, = 1, /3p, = 100, Table IV) and the periodic daily functions 6f bile
flow (~bB0, intestinal absorption (0), and the dieldrin exposure source function (s)
(Table V). Under this schedule the rat consumes 500 /zg dieldrin/day during the
12-hr feeding portion of each of the first 30 days of simulation ( " f e e d - u p " ) .
Exposure then ceases and residues are simulated for the 20-day " f e e d - o f f " period.
Kidney excretion is assumed to be negligible (Rex = 0).

The remainder of the parameters used in the simulations are given in Table VI.
By arbitrarily setting values of the forward transfer coefficients (kx+) in relation to
compartmental blood lipid flows (~bx) so that the ratios differ by a factor of 102,
examples of these potential types of dispositions are simulated. At one extreme is
the transport- or flow-limited case (Case Ia: k+/~b = 100), corresponding to the
 Table IV. Liver cell dieMrin destruction rate coefficient parametersa f o r a hypothetical 300 gm mature male rat.

Catabolism
Number Protein masses Induction coefficient coefficient Proportionality coefficient

Reomp g protein 1
i Pi(0) (gm protein) ~'Pi g dieldrin _ hr Kep i ( l / h r ) KDi g protein hr
g lipid g lipid t~

1 0.003 1.0 0.05 20.0

2 0.006 0.0 28.0

3 0.012 0.0 1.0

O
4 0.300 100.0 0.01 0.0 "~
K
O
5 0.291 0.0 0.0

Lipid masses (g)

MMs (0) = 0.355 ct = 1.724 (Equation 20, Lindstrom et al. 1974)
Mcy + Mn + MMe = 0.355

aThese estimates were obtained from data supplied by Davis et al. (1973) and Davis (1973).
 274 F. T. Lindstrom et al.

Table V. Periodic daily function~ a

Valve
Coefficient Units 0 < t ~< 12 his 12 < t < 24 hrs Duration

~bBi(t) g lipid/hr 0.32 0.032 50 days

0(t) (dimensionless) 0.995 0.05 50 days

~(t) ~ g dieldrin/hr) 41.67 0 first 30 days

0 0 last 20 days

afro = f(t + 24).

classical model suggested by Dedrick and Bischoff (1968) and employed in the
preliminary model by Lindstrom et al. (1974 and 1975). The membrane- or
transfer-limited case is illustrated in Case Ic (k+/(,b = 0.01), while an intermediate
case for relatively equal involvement of the transport and transfer functions is set
forth in Case Ib (k§ = ~b).

Data from a number of experiments (Claeys et al. 1969, Gillett 1974, Zabik and
Schemmel 1973) suggest that dieldrin lipid phase residue levels in the various tissue
compartments differ in a regular manner. One approach to achieving similar patterns
in the simulations is to set the valuesof 3,(k-/k+)arbitrarily, as has been done in
Table VI. An identical alternative would be to assume 3' constant for all tissues and
set k. arbitrarily so as to generate higher or lower residues in a given compartment.

The results of the simulations are shown in the computer-drawn figures (Figures
4, 5, and 6) for case Ia-c, respectively. The mean values of dieldrin lipid-phase
residues in the four main compartments of interest are summarized in Table VII,
which shows the residues at the end of exposure (30 days) and then after a
subsequent 20-day period without exposure.

The trio of computer-drawn figures (Figures 4, 5, and 6) all show the same
oscillatory build up in the tissue residues shown earlier by Lindstrom et al. (1974).
However, unlike the predictions o f the preliminary model (e.g., that as time prog-
ressed all lipid-phase residues would oscillate about the same stable mean value
irrespective of the tissue considered), the revised model now clearly demonstrates
the ability to predict different stable means about which lipid-phase dieldrin residues
in different tissues can oscillate.

The sharper and more extensive oscillations in the compartments of lower effec-
tive lipid mass contrast visiby with the very weak oscillations in the large adipose
c o m p a r t m e n t - - a direct manifestation of the " i n e r t i a " associated with the adipose
lipids. Also, the perfusion rate (~bx/~2xB) (the reciprocal of the compartment resi-
dence time) is very small for the adipose compartment. The apparent half-life of
dieldrin in the rat under these conditions is about seven days, which is within the
 Table VI. Physiologic and transfer parameters for Case I simulationz

Effective forward transfer coefficient lq.

Pseudo (g/lipid/hr)
Effective lipid Effective blood partition Equal tratlsport- Transfer-
Mass: lipid flow: ~b ratio a Flow-limit transfer limit
Compartment (B lipid) (g lipid/hr) 9' = k_/k+ Case la Case lb Case Ic

GI Blood 0.040 3.60 1.0 360.0 3.60 0.0360

Tissue 1.110

Liver Blood 0.024 0.72 1.0 72.0 0.72 0.0072

Tissue 0. 710

Kidney Blood 0.006 2.88 1.0 288.0 2.88 0.0288

Tissue O. 130

Brain Blood 0.002 2.72 4.0 272.0 2.72 0.0272

Tissue 0.154

Adipose Blood 0.018 0.80 0.75 80.0 0.80 0.0080

Tissue 33.180

Muscle Blood 0.096 1.92 1.0 192.0 1.92 0.0192

Tissue 5.400

Other Blood 0.048 1.04 1.0 104.0 1.04 0.104

Tissue 5.940

General
blood pool 0.180 13.68

aAssumed on the basis of long-term "steady state" residues (Claeys 1969) and short-term exposures (Zabik and Schemmel 1973, Gillett
1974).
 276 F.T. Lindstr0m et al.

range of values (4 to 30 days) found by Barron and Walton (1971) and Zabik and
Schemmel (1973) respectively.

The daily changes in shape of the liver dieldrin residue curves (Figures 4, 5, and
6) illustrate features of each pharmacokinetic pattern. The difference between the

1000

E Adipose

._~ 10
r
:L

5 10 15 20 25 30 35 40 45 50
Time (Days)

Figure 4. Semi-log plot of dieldrin residues for four selected tissues in the hypothetical rat
(Run 1, flow limit conditions).

.r,
~. ~oo

5 I0 15 20 25 30 35 40 45 50"

Time (Days)

Figure 5. Semi-log plot of dieldrin residues for four selected tissues in the hypothetical rat
(Run 2, equal flow-equal membrane transfer conditions).
 Dieldrin Distribution Model 277

11;1130

'E,
Adipose
> 100
.,-~

Brain

._~

::k

9 9 , . . . . . . . . . . , , h , i i , , , , . . . . . . . . . . . .

5 10 15 20 25 30 35 40 45 50
Time (Days)

Figure 6. Semi-log plot of dieldrin residue for four selected tissues in the hypothetical rat
(Run 3, transfer limit conditions).

patterns of the three cases derives in part from the liver having a periodic loss term
via bilary flow (12 hr on, 12 hr off). Under flow-limiting conditions, dieldrin is
transferred from the extracellular to the intracellular subcompartment so rapidly that
nearly instantaneous equilibrium is achieved. Thus, a substantial amount of the
" n e w " dieldrin delivered to the liver extracellular subcompartment via the portal
vein and/or the hepatic artery is immediately transported into the tissue portion. This
rapid transfer process tends to maintain the liver residue so as to reflect blood
changes (principally affected by tissue losses and absorptive gains) even during
periods when bile flow is occurring.

Table VII. Simulated residues of dieldrin in various compartments

following exposure and "feed-off"

(/~g HEOD/g lipid)

Case la Case lb Case Ic
Day 30 Day 50 Day 30 Day 50 Day 30 Day 50

Brain 15 1.5 22 4.8 230 95

Liver 58 5.7 50 11.5 15 6

Blood 60 6.2 85 19.1 920 370

Adipose 80 1.1 115 36.2 105 150

278 F. T. Lindstrom et al.

In Case Ib, where the contributions of transport and transfer are more equal
(Figure 5) due to assumption of a lower value for the forward transfer coefficient,
the effect of the biliary loss term on liver residues is clearly demonstrated. During
the period when bile flow is occurring more dieldrin is effectively lost from the liver
than can be replaced by transfer across the membrane. Dieldrin residues at day 30
are lower in Case Ib than in Ia (Table VII) and the decay during feed-off is retarded.
Since the forward and reverse transfer coefficients of all other compartments are
similarly reduced, the decreased ability of dieldrin to flow into the liver (the only
exit) results in higher residues at day 30 in all compartments of Case Ib in compari-
son to case Ia.

At the other extreme (transfer-limit conditions, Case Ic, Figure 6) the residues in
all compartments but the liver are very much higher than in either Cases Ia or Ib
both at day 30 and day 50 (Table VII). The liver does not reflect the blood
concentration (Case Ia), but clearly shows the biliary loss response. Liver residues
are even lower than in Cases Ia and Ib.'More significantly adipose tissue residues
continue to rise post-exposure, indicating the extreme deviation from the "steady-
state" assumed in deriving the rate of hepatic dieldrin metabolism. Dieldrin ac-
cumulates continuously under these conditions of very slow membrane transfer.

Mathematically the model of Case Ic can be viewed as in Figure 7. The Small

effective lipid masses of blood and extracellular compartments (~xB) are summed,
and dieldrin fed to the rat accumulates in the G.I. tissue compartment (1230 ~g
HEOD/g lipid at day 30) from which it is slowly released (k~I = 0.036 g lipid/hr).
Because of the slow transfer from the blood to the liver, which continues to lose
dieldrin via the bile, liver metabolism is greatly reduced.

Dieldrin pharmacokinetics follow neither extreme pattern (Cases Ia and c) but

appear likely to involve combinations of effective lipid mass, tissue perfusion rate,
and forward and reverse transfer coefficient arrayed so as to present an intermediate
case (Dedrick et a l . 1973). In the overall perspective these simulations indicate the
necessity of a time-wise series of measurements during exposure and continuing
post-exposure as a basis for evaluating pharmacokinetic mechanisms.

Case II. Blood-brain barrier simulation. In order to study the relationships

devolving from a compartment constrained by transfer limitations in a transport-
limited mammal, a series of simulations were executed using identical parameters as
in Case Ia, except for the brain. The blood-brain barrier can be supposed to be
characterized by two constants:k~, and Tan. The actual residence or turnover time
in the brain tissue compartment additionally will be dependent upon effective brain
lipid (MBar).

By straight forward methods it is easy to show that the lipid-phase dieldrin

residue in the brain tissue subcompartment can be expressed as
 Dieldrin Distribution Model 279

UBRT(t ) KBR+ t UBRB(r) KBR+TBR(t_r) }

= dT" (t0)
CBRT(t) = 'JJ~BRT ~BRT ~ 0 ~YJ~BRB exp '-~ BRT

If further it is assumed that during the " f e e d - o n " stage of the simulation

UBRB(t) = Co(1 -e-Xt), CO = ~ - (/.tg HEOD/gm lipid),

~.BRB
and

?~ = F/(total body lipid--brain tissue lipid) (L/hr),

the brain tissue lipid phase dieldrin residue is approximately

Co oxp(- t) exp ( -
"KBR+')'BRt }/
CBRT (t) -- "fBR (ll)
I- ~'~')~BRT + KBR+~'BR I)
(1-- KBR+.YBR") X~R,BRT

Fat Brain

UFT, ~'~FT UBRT, ~02BRT

KF§
t t F Rex
KBR+ KBR_
Other "*- KOT - KK-~- -- Kidney
General Blood Pool
UOTT, ~'~0 TT U KT, ~ K T
-- K O T + KK+--
UB'~J'~B =~J~T~B+~ff~GIB +~'~LB +~'~KB

Muscle 9=- KM_

+~'~BRB+~'~FB+qP
~'MB -~n~'
-~'OTB KGI..= G.I.
UMT ' ~0~MT UGIT, ~'~GI T
KGt +--
L, KM+ KL- KL+
t
Lumen 1 =- (1 - e ) F ~ ,
Liver ULT
r(metabolism)
85 + ~BI ~-ff~LT

Figure 7. Schematic of the extreme limit of membrane transfer for a hypothetical mammal.
280 F. T. Lindstrorn et al.

While the resultant of the integration indicated in equation (10) is exact (equation
(11)) for the stated condition on UBrm(t), this residue distribution is only an
approximation to reality. Nevertheless, it serves the useful purpose of allowing
estimates of residence times to be made. These residence times then serve as a guide
for the sequential study of the dieldrin pharmacodynamics in the brain compartment.

By rearranging the variables in equation (11) we obtain

CBRT(t) exp(-Xt) exp { - KBR+TB"~Rt

}~ffdBRT

1 - 7BR Co = X s.ffdBRT -- KBR+TB R (12)

- - 1)
(t KBR+TBR ) X ~-~BRT

Setting
1--TBR CBRT(tl/2) = 1
Co 2'

the following transcendental expression for tit.2 max, the time of half maximum
residue concentration, is obtained
KBR+TBR t 1/2 max
i exp(-kt 1/2 max) exp -- ~ff~BRT
T = X~E)~BRT -- KBR+')'BR (13)
(1 1)
K BR+')'BR X~-?ff~S R T

By choosing realistic values for k, KBR', Tsa, and~sRT, this equation can be solved
numerically to find tl/2rnax. This formula may be exploited as a guide as to when to
expect significant differences in the brain tissue lipid phase dieldrin residues from
those o f strict flow-limited conditions.

The t 1 ~ max x,alue reflects not only the membrane transfer parameters (KBa, and
Yea) but also shows that the effective brain tissue lipid mass is important. Recall
that the effective brain tissue lipid mass ~2BaT is defined as:
1
,~02BRT = M]~RTL q- MBRTP~3BRTP -F MBRTS ~BRT$
(from the preced~g section on effective flow and mass coefficients). Since the rat
brain tissue saline mass MBRTS is less than two g for our hypothetical rat, and since
PBRTS is in the order of 3 x 105, the third term in equation (15) is of the order 10 s
With MBRTL = 0.154 gm lipid and MBRT.P = 0.22 gm protein we may safely neglect
the saline fraction term. From Table VI, "YBR= 4.0 for dieldrin. Substitution of the
other mass values and the values: F = 0.27 g lipid/hr, k = 0.006 L/hr, and C o - 80
/zg/g lipid (Lindstrom e t a l . 1974) allows Table VIII, a table of half maximum
times, to be generated. Also included in Table VIII are values of the parameter
ratios which appear in equation (13),
 (o) (o),
Table VIII. Estimated half maximum times tH2 max (hr) amt parameter ratios: PBR = KBR + 7BR/'~t BR T and P (B~ = PER~7
for the rat brain tissue HEOD residue study.

t g lipid
KBR+ ~ hr
(Curve No. 1) (Curve No. 2) (Curve No. 3) (Curve No. 4)
~BRTP lO-I 10-2 10-3 10-4 Figure 8
g lipid (o) (l) (o) O) (o) O) (o) (l) (curve
( g protein ) tl/2 max ,OBR /9BR tl/2 max PBR ,OBR tl/2 max ,OBR ,OBR tl/2 max ,OBR #BR sets)

0 115 2.6 430 120 0.26 43 15~ 0.026 4.3 445 0.0026 0.43 top

0.1 116 2.3 380 121 0.23 38 164 0.023 3.8 481 0.0023 0.38 second

1.0 117 1.1 180 125 0.11 18 216 0.011 1.8 807 0.0011 0.18 third

10.0 121 0.2 30 178 0.02 3.0 504 0.002 0.3 3,617 0.0002 0.03 bottom
 282 F. T. Lindstrom et al.

KBR+TBR
p(O) (14)
BR -- ~BRT

and

p(l) = p(~/),
BR (15)

To illustrate the effects of the effective membrane transfer coefficient KBR, and
the pseudo brain partition coefficient 7as, both operating in conjunction with the
effective brain tissue lipid m a s s , B a T , Figure 8, has been prepared. The semi-log
plots of the brain tissue (intracellular subcompartment) dieldrin residue (/~g
HEOD/g effective lipid) versus time in days are shown in groups of four. The
appropriate Ksa, value used for each computer curve drawn is listed in Table VIII.
The effective brain tissue lipid mass is held constant with the actual effective lipid
mass value used in each set of four simulation runs given in Table VIII.

Figure 8 shows many interesting features pointing out the complex inter-
relationships of compartmentalization mechanics. A sequential cursory scan of the
curves labeled o n e from top to bottom reveals that with Ksa+ = 0.1 (g/hr) a rather
large increase in the effective lipid mass (creating a larger pool into which dieldrin
can be transferred) has very little effect in increasing the time of half maximum, at
least over the range of effective brain tissue effective lipid mass (0.154 g ~<~.lRsaT
~<2.354 g). The added inertial effect of large amounts of lipo-protin binding are
simply covered by the very rapid membrane transfer kinetics in this case. This
phenomena predicted with the complete simulator is also demonstrated adequately in
Table VIII.

Some increase in the values of t,/2 max can be seen for KBa+ = 10-3. This shift
toward longer times is especially noticeable in the bottom curve labeled two. Again
the estimator [equation (13)] gives good agreement on tl/,_ max. Another feature of
the increasing trend away from flow limit conditions and movement toward mem-
brane transport limit conditions and/or large effective reservoir sizes is the gradual
decrease in the amplitude of the daily oscillations. As KBR+ decreases and~2RBRT
increases, this effect becomes very clear. Notice that all the curves shown on Figure
8 with the label one still have significant daily oscillations on the feed-up portions
(first 30 days), whereas considerable reduction in oscillation amplitude and a
significant retardation in residue build up shows up under the conditions: KsR+ =
10 2 (g lipid/hr), 3'Ba = 4.0, and~BRT = 2.354 (g " e f f e c t i v e " lipid) (23BaTP = 10.0
g protein). Observe that the daily oscillations in all the cases considered are quite
small when KBR+ = 10 a (g lipid/hr). In the range of effective membrane transfer
parameter values where large daily fluctuations in the extracellular (blood capillary)
residue values seem to have only a small effect on a rather steady daily mean build
up, the retardation in build up of residue as well as in decay is now strongly affected
by the magnitude of effective lipid mass ~IRsrtT.
 Dieldrin Distribution Model 283

100

10 , 4

1' I " 2
"g_
Q.

==
E
o'l

0
UJ
-t-

10

I,-. -'
~m
I! 1 . ~ , , , , E , = .'t . . . . t . . . . ~ . , ~ .......... ~ , ,,, , I ; , , ~ I , ~ , , I

I.-
{z
0~
c~

r
0
tj

0
LU
,-r- 10= . ~ &

"0 V IIII , ...... , , ..... +

/

I..-

100

10

,c(/ ~4 ----~
0 5 10 15 20 25 30 35 40 45 50

Time in Days

Figure 8. HEOD residue simulations for rat brain tissue.

284 F. T. Lindstrom et al.

There appears to be some anomalous behavior in the brain residue curve in going
from~2Bav = 0.374 t o ' S a T = 2.354 g effective lipid, curves labeled three, in that
for~BaT ~< 0.374 g effective lipid the maximum residue occurs during the 30th day,
the last day on which dieldrin is administered, while the bottom curve has its peak
occurring at about day 35, five days after cessation of dietary dieldrin. We will
demonstrate v i a the use of the estimator (which is actually a very good approxima-
tion here) how this effect comes into being. First, it must be remembered that
equation (11) is valid only for times during which dieldrin is being administered.
For us this means 0 ~< t ~< 30 days. By adjoining to the blood " f e e d u p " dieldrin
residue distribution
UBRB(t)
CBRB(t) = ~ ] ~ B R B - - C ~ e - kt),

0 ~< t a T, (T = 30 days in our example), the blood " f e e d o f f " dieldrin residue
distribution
CBRB(t ) = Co(e- k(t--T) --e -- kt),

t > T (T = 30 days), it is possible by subgtituting these two definitions into equation

(I0) to obtain both " f e e d u p " and " f e e d o f f " approximate brain tissue lipid phase
dieldrin residue distributions.

The " f e e d u p " distribution is stated in equation (11). Continuing the integration
gives

- xt) (e -- P B R (
(0) t -- T) - - e-- "'~(O)t
BR )
i ,
C ~ t( e -k._(t.: T ) - - e
CBRT(t) = ~BR~ 1 -- 1
p(1) pO)_ 1 (16)
BR
t > T. By differentiating equation (16) with respect to " t " and setting the derivative
equal to zero, the following formula arises for the value of tmax,

In l (o) I
ep BR - I

e ~'T- 1 (17)
)~(1 ~- o(1)"~
trlla X
V'BRJ
Clearly, as PaR ~ 1, ( " o n e from the minus s i d e " ) , PBR "~ k, andtmax ~ + ~ , as
would be expected. Thus, PBa = 1, is the naturally arising dividing line between
when the expected peak in the residue curve will occur either during the last day of
" f e e d o n " or at some time later on after the " f e e d o f f " stage has been begun.
Hence, by careful estimation of PBR (Table VIII) it should be possible to predict
when the maximum brain tissue residue .will occur. In view of this reasoning then,
observe that in Table VIII for ksa + = 10 -a (L/hr) except for ~ BaT = 2.354 g lipid, PBR is
greater than one. Thus, tmax for all these curves occurs on day 30, whereas trnax for
the last case occurs about 35 days as a simple calculation using equation (17) will
reveal. A table of tmax values based upon rat brain parameters has been prepared
below (Table IX) to aid the reader.
 Dieldrin Distribution Model 285

The effects of membrane transfer limit conditions are clearly visible by inspection
of the set of curves labeled four in Figure 8. The increasing shift toward longer trnax
values, the reduction in dieldrin residue amplitude at day 30, the almost total lack of
any daily residue amplitude oscillation, and the progressively slower decay of brain
tissue dieldrin burden with increasing effective lipid mass~)~BaT are all clearly seen
in the set of curves labeled four in Figure 8. Again this is exactly what we would
expect by a quick scan of the column of tmax values in Table IX labeled " c u r v e 4 " .
All P6a values in this column are less than one in magnitude. Hence, we would
expect t~ax values in the actual full simulator to also give similar results, which they
do.
Because of space limitations we list in Table X the average brain tissue residue
values in/~g per g effective lipid at day 30 and day 50 for the various combinations
of K~R, and MaRT used in the brain simulation study.

Conclusions
The revised model for the pharmacokinetics of dieldrin and similar highly
lipid-soluble substances incorporates several features capable of simulating realistic
residues observed in various experiments. Unfortunately the basic data required for
explicit representation have not been developed, and those features (inter- and intra-
tissue compartmentalization) have not been fully explored for even one species and
compound. To validate the model as proposed herein, a series of feeding experi-
ments should be performed incorporating aspects of the designs of Barron and
Walton (1971) and of Zabik and Schemmel (1973). The rates of residue build-up and
decay should be determined along with changes in specific activity of 14C-labeled
dieldrin in several tissues of pre-dosed animals. Additionally, an unequivocal means
of determining lipoprotein and tissue binding of dieldrin must be found. Once these
patterns are established, the parameters of the model can be set or measured directly
with respect to age, physiological conditions, etc.
As illustrated in Case I examples, the pharmacodynamics and pharmacokinetics
of dieldrin are best simulated by assumptions providing for an intermediate level of
transport and transfer processes. The patterns seen for various relationships in Case
II simulations, although shown for the brain, illuminate the area of differences
between compartments. For the rat receiving dieldrin it would appear that some
compartments being transfer-limited (e.g., G.I. tract, adipose, brain) and others
being transport-limited (e.g., liver, muscle) provide the complex changes seen in
chronic exposure.
The distributions of several lipid-soluble chemicals (e.g., DDT and its cogeners)
are not the same as dieldrin, but these differences most likely rest in the values of
the tissue binding coefficients and membrane-specific transfer coefficients. Dieldrin
does not interact specifically with phospholipid, whereas DDT does (Haque 1974,
Pocock and Vost 1974). Thus in vitro studies can be helpful adjuncts to whole
animal feeding trials.
In initiating modeling studies of the pharmacodynamics and pharmacokinetics of
highly lipid-soluble chemicals, the authors intend to continue developing systems of
 Table IX. tmax [days] values estimated from equation (17). bO
OQ
O',

Curve 1 Curve 2 Curve 3 Curve 4

Effective KBR + = 10-1 KBR+ = 10 -2 KBR + = 10 -3 KBR + = 10-4 Figure 8
mass g lipid .g fipid g fipid g lipid
(g lipid) hr hr hr hr

0.154 30 30 30 30 top

0.176 30 30 30 32 second

0.374 30 30 30 36 third

2.354 30 30 32 44 bottom

B
Table X. 30 and 50 day average HEOD brain tissue residues lag dieldrin/g effective lipid.

Curve 1 Curve 2 Curve 3 Curve 4

BRT KBR + = 10-1 KBR + = 10-2 KBR + = 10 - 3 KBR+= 10-4
(g effective g lipid g lipid g lipid g lipid Figure 8
lipid) I~ hr hr hr (curve sets)

0.154 15.0 1.5 14.7 1.55 14.3 1.85 11 6.0 top

0.176 14.8 1.6 14.5 1.65 14.1 1.9 10 6.3 second
0.374 14.6 1.6 13.7 1.6 13.5 2.5 6.5 6.0 third
2.354 14.4 1.6 13.3 2.0 9.0 6.5 1.4 1.65 bottom
 Dieldrin Distribution Model 287

general relevance capable of predicting exposures of specific tissues in animals

chronically ingesting hazardous substances. Most importantly, the models exposited
herein need rigorous testing for validity and applicability. Such testing is beyond the
scope of any one laboratory, as has been the production of the data used in the
formulation of this system. We can only urge others to test and criticize this attempt,
to the end that a better understanding of the effects of such agents as presticides,
drugs, and industrial chemicals can be achieved.

Acknowledgements
The authors gratefully acknowledge the advice and encouragements of their
colleague Dr. Ian J. Tinsley, Professor of Chemistry, whose suggestions were most
helpful in preparing this update of the preliminary model. Also, this research was
supported in part by U.S. Public Health Service grants ES 00040 and ES 00210 from
the National Institute of Environmental Health Sciences. This paper is Technical
Paper No. 3914 of the Oregon Agricultural Experiment Station.

References
Barron, R. L., and M. S. Walton: Dynamics of HEOD (dieldrin) in adipose tissue of
the rat. Toxicol. Appl. Pharmacol. 18, 958 (1971).
Bischoff, K. B., and R. G. Brown: Drug distribution in mammals. Chem. Eng.
Progr. Symp. Ser. A. 66, 32 (1966).
Bischoff, K. B., and R. L. Dedrick: Thiopental pharmacokinetics. J. Pharmaceut.
Sci. 57, 1346 (1968).
Claeys, R. R.: Unpublished data (1969).
Dale, W. E., A. Curley, and C. Cipriano: Hexane extractable chlorinated insec-
ticides in human blood. Life Sciences, 5, 47 (1966).
Davis, H. T.: Introduction to Non-Linear Differential and Integral Equations, 566
pp. New York: Dover (1962).
Davis, J. L.: Personal communication (1973).
Davis, J. L., R. O. Morris, and I. J. Tinsley: Quantitative measurement of
microsomal sub-fractions isolated from livers of rats fed with 1, 1,
1-Trichloro-2,2:bis(p-chlorophenyl)ethane (DDT). Biochem. Pharmacol. 22,
869 (1973).
Dedrick, R. L., and K. B. Bischoff: Pharmacokinetics in applications of the
artificial kidney. Chem. Eng. Progr. Syrup. Ser. A, 64, 32 (1968).
Dedrick, R. L., D. S. Zuharko, and R. J. Lutz: Transport and binding of methotre-
xate in vivo. T. Pharmacent. Sci. 62, 882 (1973).
Gillett, J. W.: Unpublished results (1974).
Haque, R., I. J. Tinsley, and D. Schmedding: Lipid binding and mode of action of
compounds of dichlorodiphenyl-trichloroethane type: A proton magnetic reso-
nance study. Mol. Pharmacol. 9, 17 (1973).
288 F. T. Lindstrom et al.

Haque, R.: Personal communication (1974).

Jakubowski, T., and L. A. Crowder: Binding of 36Cl-dieldrin to suspected target
and non-target proteins in vitro. Bull. Environ. Contam. Toxicol. 10, 217
(1973).
Krueger, H.: Personal communication (1971).
Lindstrom, F. T., J. W. Gillett, and S. E. Rodecap: Distribution of HEOD (dieldrin)
in mammals: I. Preliminary model. Arch. Environ. Contam. Toxicol. 2, 9
(1974).
Lindstrom, F. T., J. W. Gillett, and S. E. Rodecap: Distribution of HEOD (dieldrin)
in mammals: II. Some applications of the preliminary model. Arch. Environ.
Contain. Toxicol. 3, (1975).
Morgan, D. P., C. C. Roan, and E. H. Paschal: Transport of DDT, DDE, and
dieldrin in human blood. Bull. Environ. Contam. Toxicol. 8, 321 (1972).
Moss, J. A., and D. E. Hathway: Transport of organic compounds in the mammal:
Partition of dieldrin and telodrin between the cellular components and soluble
proteins of blood. J. Biochem. 91, 384 (1964).
Pocock, M. E., and A. Vost: DDT absorption and lipoprotein transport in the rat. J.
Amer. Oil Chemicsts Soc. 50, 302A (1973).
Pocoek, D. M. E., and A. Vost: DDT absorption and chylomicron transport in rats.
Lipids. 9, 374 (1974).
Richtmyer, R. D., and K. W. Morton: D~ference Methods for Initial-Value
Problems. 405 pp. New York: Interscience (1967).
Rodecap, S. E., and F. T. Lindstrom: Numerical simulation of a compartmental
system for the distribution of lipid soluble chemicals in mammalian tissues.
Computers in Biology and Medicine 5, In press (1975).
Spector, W. S. (ed): Handbook of Biological Data. Fed. Amer. Soc. Exp. Biol. 584
pp. Philadelphia: Saunders (1956).
Tidwell, H. C. : Absorption of free and combined fatty acids. Proc. Soc. Exptl. Biol.
Med. 98, 12 (1958).
Varga, R. S.: Matrix Iterative Analysis. 322 pp. Englewood Cliffs, N.J.: Prentice
Hall, (1962).
Wauchope, D. R.: Private communication (1971).
Yadrick, M. K., M. E. Zabik, and F. Funk: Dieldrin levels in relation to total,
neutral, and phospholipid composition in selected pork muscles. Bull. Envi-
ron. Contam. Toxicol. 8, 289 (1971).
Zabik, M. E., and R. Schemmel: Dieldrin storage in obese, normal, and semi-
starved rats. Arch. Environ. Health 27, 25 (1973).

Manuscript received October 11, 1974; accepted January 5, 1975.