450 TIPS -November 1983

Analysis of combined constant fraction of a system. Consider the

case where, under specified conditions of
dose and experimental design, one drug

drug effects: inhibits a process 20%, i.e. fraction

affected 0ca) is 0.20. Since by definition (fa)
+ (fu) = 1, the fraction unaffected (fi)
a new look at a very old must therefore be 0.80 (80%). Similarly,
we find that a second drug inhibits the same
problem system by 40% (/Ca = 0.40), so that f~ =
0.60. According to the fractional product
concept, the actions of the two drugs have
Ting-Chao Chou* and Paul Talalayt been considered additive, if their combined
*Laboratory of Pharmacology, Memorial Sloan-Kettering Cancer Center, Cornell University Graduate effects leave 48% (60% of 80%) unaf-
School of Medical Sciences, NY, USA. fected. Or, more specifically (fu)x= = (fu)t
t Department of Pharmacology and Experimental Therapeutics, The Johns Hopkins University School of x (fu)2, where (fu),:, (fu),, and ~)2 refer to
Medicine, Baltimore, MD, USA. the fraction of the total activity unaffected
by the combination of the two drugs, by
drug 1 and by drug 2, respectively. If the
A simple, general and novel method for the quantitative analysis of combined drug observed effect of the two drugs acting
effects is presented. The procedure determines if the drug combinations are additive, simultaneously is larger or smaller than that
synergistic or antagonistic, regardless o f whether the dose--effect curves are calculated from the product expression, it
hyperbolic (first order) or sigmoidal (higher order), or whether the effects of the drugs has been assumed that synergism or
are totally independent or not. We believe that the proposed method (median-effect antagonism, respectively, has occurred.
ploO suffers from fewer restrictions and is more widely applicable than the Although this equality is intuitively attrac-
isobologram or the fractional product method. It provides insight into the nature of tive and its use is simple, it is not valid
the dose--effect relationship with relatively few measurements. We illustrate the unless both drugs are mutually nonexclu-
procedure by analysing the combined lethal effects of two insecticides on house-flies. sive, i.e. totally independent of one
another, and give hyperbolic dose--effect
The combined use of two or more therapeu- knowledge of the doses of a number of drug curves. Unfortunately, the isobologram
tic agents is often advantageous since it combinations which achieve the same method x and the fractional product method a
may permit lowering of drug dosages and end-point can only be obtained by making a have often been applied indiscriminately
consequently of toxicity, reduce the oppor- large number of measurements. (2) The without full appreciation of their limita-
tunities for the development of resistance plot gives a valid definition of summation tions.
by the target cells or organisms, and pro- of drug effects only if the two drugs have a
vide the potential for synergism of drug similar mode of action (see Chou and Proposal o f a n e w m e t h o d o f analysis
effects. One of the central problems in Talalay3). (3) For all practical purposes, The proposed method of analysis is the
assessing the effectiveness of drug combi- the analysis is limited to two drugs. outgrowth of the examination of the rela-
nations has been the lack of agreement on The conclusion that the summation of tion between concentration (or dose) and
how to analyse the experimental results in a effects of two or more drugs can be calcu- effect in a number of systems, and is based
quantitative manner that is simple, theoreti- lated by multiplying the residual fraction of on the derivation by Chou4,5 of the
cally sound, and does not require large the activity unaffected by each drug indi- 'median-effect principle' of the mass action
numbers of measurements. The two most vidually, is based on the view that two or law, and its application to multiple enzyme
widely used methods of analysis of such more agents whose actions are totally inde- inhibitions8'6, by a simple graphical
systems are the isobologram introduced by pendent o f one another (and are therefore method.
Loewe ~, and the fractional product method 'mutually nonexclusive') will each affect a It is well known that the dependence of
formalized by Webb 2. The isobologram
analysis requires the construction of
dose---effect curves for each drug alone and Definitions
for a series of mixtures of the drugs. The f~ = fraction of system affected by drug
doses of each drug alone and in combina- fu = fraction of system unaffected by drug sincefa + fu = 1, fu = 1 - fi,
tions which are required to achieve an (D)1, (D)2, (D)1,2 = dose of drug 1, drug 2 or mixture of drug 1 and drug 2 (in
arbitrarily-selected end-point (commonly specified ratio), respectively.
EDso, IC5o, or LD,o) are then used to con- (Din)l, (Din)2, (Dm)l.~ = dose of drug 1, drug 2 and mixture of drug I and drug 2 (in
struct the isobologram, as a secondary plot. specified ratio) required to produce a median effect, e.g. IC5o, EDso or LDso.
The isobologram, which is a curve of equi- m = the slope of the median-effect plot, in which y = log (fa/fu) and X = log (D).
effectiveness, reveals by inspection The magnitude of the m value is a measure of the sigmoidicity of the dose-effect
whether the drug mixture displays summa- curve and is analogous to a Hill coefficient. In an enzyme system it represents the
tion (additivity), synergism, or antagon- co-operativity coefficient of binding or the (minumum) number of binding site(s).
ism. There are several major drawbacks of When m = 1, a Michaelis-Menten type of hyperbolic dose--effect curve is observed.
the isobologram method. (1) Accurate
© 1983 Elsevier Sctence Publishers B V , Amsterdam 0165 - 6147/83/$01 00
TIPS -November 1983 451

biological effects on concentration (or dos-

¢.h~ = (f.h (f.)2=(D.), ~ (2) 100 (1R
u ~ 5P) (1R:lSP)
Mo~
age) conforms commonly to one of two (R)
general types: either the relationship is
hyperbolic (f'LrStorder, or Michaelis-Men- Similarly, if the slope (m) of the above plot
ten type), or it is S-shaped (sigmoidal or is not 1 (e.g. higher order system), then:
higher order). Two important parameters
characterize both systems. The fwst [¢,~,~l- r¢,),]~ rc.),l,-
parameter is the concentration (or dose) l-mhJJ =Lc-~,J+L~SJ" -~
× e0
(3)
required to achieve the median effect (Din),
(D)1 (D)2 _~ 50 d
which is synonymous with the LD~o, EDso,
= ( D , . ~ "~(D,.-----~
or IC~o, describing lethality, activational
effects of agonists, and inhibitory effects of A general equation for the combination of 7 _
antagonists, respectively. In a broader more than two drugs has also been derived 8.
sense, Dm is also analogous to the We are now ready to apply the analysis to 20
Michaelis--Menten constant (Kin), or the Ko an experimental situation.
value characterizing the ionization 10
(strengths) of acids or bases5. The second Lethal effects of two insecticides
parameter is the slope of a plot of log In order to apply the median effect o os 1o l.s 2o
C~-.2~~ a t ~ l o( I n a e c ~ d e
(fdfi)* with respect to the log D , the dos- method, experimental values are required (~r~"l
age or drug concentration. This slope is for dose-effect curves for each drug alone
designated as the m value (a measure of and for at least one fixed ratio mixture of Fig.1. Arithmetic dose-effectplot of the lethality of
rotenone (R) and pyrethrms (P) for house-flies. The
sigmoidicity). When the value of m is drugs, results (Table 1) were retrieved by Finney sfrom the
unity, the system obeys hyperbolic or In'st A series of quite old, but very careful expenmen~ findings of Le PeUey and Sullivan °.
order kinetics. measurements on the lethal effects of two The symbols for the ~st series (©, [:2) and the
The median effect relationship states insecticides (varied alone or in two combi- second series (0, II) of experiments designate
quite simply that for any dose D , the nation ratios) lend themselves to our rotenonealone, and pyrethrins alone, respectively.
fractional effect (fo) is given by the analysis. The experiments are described by The mixture in the first series contained rotenone to
pyrethri~ in a ratio of 1:5 (A) and in the second
expression8,4: Finney (1952) s, who also analysed them, series in the ratio of l:15 ( &), by weight.
as follows: 'Le Pelley and Sullivan (1936)
have reported the results of two series of tri-
V. = b--= o) als in which adult house-flies were sprayed Finneye from the original graphs of Le Pel-
with alcoholic solutions of rotenone, pyre- ley and Sullivan8 and agree closely with our
Hence, log (fo/f.) = m log ( D / D . ) = m thrins, a n d a m i x t u r e o f t h e t w o ; i n t h e f i r s t own analysis of these graphs. Arithmetic
log D - m log D . . Consequently, a plot of series the mixture contained rotenone and scale plots (Fig. 1) of the dose-effect rela-
y = log (fo/fu) with respect to x = log D pyrethrins in the proportion of 1:5 (by tionship for each insecticide alone and for
will be linear and will have a slope of m. weight), and in the second series the pro- the two mixtures (1:5 and 1:15) show
The magnitude of this slope will at once portion was 1:15. About 1 000 flies were clearly sigmoidal, rather than hyperbolic
reveal whether the system behaves in tested at five levels of each toxic prepara- relations between fly kill and concentra-
accordance with Michaelis-Menten predic- tion', tion. The same results plotted as log (f~/f.)
tions (m = 1) or not (m ~ 1). Furthermore, The results (Table I) were retrieved by with respect to log D are shown in Fig. 2,
the line will intercept the x-axis at y = log
(f,/fi) = 0. Consequently, on this a x i s f d f ,
TABLEI. Toxicityof rotenoneand pyrelta~ to house-flies(fromthe dataof Le Pelleyand Sullivan"as retrieved
= 1 and sincef. +I, = 1.0,/, = f , = 0.5. by Finneys)
The x-axis intercept is log D . and its anti-
log gives the value of D . . Thus, the x-axis Fire series Secondsines
can be considered the median effect axis. Concentration % Kill Concenttalkm % Kill
The plot has been referred to as the (mg ml-') (rag ml-')
median-effect (Chou) plop .~'. If a linear
Rotenone (R) Rotenone (R)
regression analysis is used, the value of D -
0.10 24 0.10 28
is based on all the experimental values. 0.15 44 0.15 51
Thus, this plot provides a useful way of 0.20 63 0.20 72
analysing the effect of ligands in many 0.25 81 0.25 82
systems. 0.35 90 0.35 89
We now extend this analysis to descdhe ~ ( p ) Py~thrins(P)
the combined effects of two pharmacologi- 0.50 20 0.50 23
0.75 35 0.75 44
cal agents, DI and Dz. In its simplest (first
1.00 53 1.00 55
order system; mutually exclusive) form, 1.50 80 1.50 72
the median effect equation predicts that the 2.00 88 ZOO 90
summation of effects of two drugs in given Mixture (IR:5P) Mixture(IR: 15P)
b~: 0.30 27 0.40 23
0.45 53 0.60 48
*Wheaanalysingexperimentalresults,it is not always 0.60 64 0.80 61
convenientto plotfail.,but to employ its algetnic 0.875 82 1.20 76
equivalents:(f.)/(1-f.), [(f.)-'- 1], Ot [(fa)-' - , !. 175 93 1.60 93
1] -~. The id~tity of th~¢ exlxessiom is shown easily
if o ~ remecnlie~that (fa + fu) 1.
= Mortalitywas sootedon day 3 ~ exposure.
452 T I P S - N o v e m b e r 1983

and their quantitative analysis by linear cumstances even i f m is unknown or differ-

regression is given in Table lI.
It is immediately obvious that the
median-effect equation and the median-
effect plot provide considerable diagnostic
and quantitative information about the
1.0
t _//I
/
Rotenone {R)

~//
#
Mature
(1R:5P)
/
/
&l
/#
/

/ /
ent m values are obtained for each
inhibitor. Consequently:

results and their interpretation. The conclu-
sions to be drawn are given below.
Linear regression, analysis
The system obeys the median effect equa-
tion remarkably well and the results are
highly reliable (undoubtedly because a
large number of flies were used). The linear
regression coefficients for the six
dose-response series (each at five concen-
trations) lie between 0.983 and 0.996, and
four of the values o f r are 1> 0.994. Since
the regression coefficients are close to 1,
the effect produced by a given dose or the
dose required to produce a given effect can
be calculated from a rearrangement of
Equation 1 and the knowledge that
f i + f i = l:
1
f" = 1 + (Din/D)m (4)
Higher order
As expected from the sigmoidal curves of
Fig. 1, the system is 'higher order'. The
slopes of plots for the single drugs and the
mixtures are remarkably similar (m =
2.52_+0.15). The parallel nature of these
lines suggests that the two insecticides are
'mutually exclusive' (e.g. have similar
modes of action). It is hazardous, if not
impossible, in this type of experiment to
ascribe a physical meaning to sigmoidlcity.
In isolated systems (e.g. enzymes, hemo-
globin), simpler than those involving death
as an end-point, this property is rationalized
by the presence of more than one binding
site for each ligand.
Median effect dose: absolute and relative
potencies
The reproducibility between two series is
excellent and accurate toxicity data are
obtained. Thus, theDm (EDso) values in the
two series for rotenone are 0.157 and 0.144
mg ml -~ and for pyrethrins are 0.910 and
0.868 mg mi-~, respectively. Furthermore,
(Dh. (Dh
e,--~'C', = 1
(Dm)l tu,,,)2
(5)
when (D)I + (D)2 concentrations are such
as to produce (f,)1.2 = 0.5. I f D 1 is rotenone
and D2 is pyrethrins, we can substitute the
values given in Table 11 into equation 5.
For the experiments of the first series
(ratio of rotenone to pyrethrins = 1:5), we
obtained (Table II) (Dmh = 0.157, (Dm)z
= 0.910, and (Dm)l,l = 0.450 mg mi-L
Now, for this series 0.450 mg mi-x of the
mixture consists of 0.075 mg mi-x of
Mlxtu
rotenone (1/6) and 0.375 mg mi-1 of pyre-
-10 -1'0 -o's ~ 0s thrins (5/6). Hence:
Log D (rag ml ')
( D h ~ (D)2 0.075 0 375
- - - - - +-'~'- =0.890
Fig. 2. Lethality of rotenone and pyrethrins for (Dm)l (Dm)2 0.157 0.910
.
house-flies. The results (Table land Fig. 1) are plot-
ted by the median-effect plot, i.e. log (fa/fu) with For the experiments of the second series
respect to log (D). The symbols are the same as for
Fig. 1. The lines are drawn by linear regression
(ratio of rotenone to pyrethrins = 1:15), we
analysis: first series ( - - ) , second series (- - -). The obtained (Table II) (Dm)l = 0.144, (Din)2
parameters obtained from the linear regression = 0.868, and (Dm)t,~ = 0.652 mg mi-L
analysis of the plots are summarized in Table IL Now, 0.652 mg mi-~ of the mixture con-
sists of 0.041 mg m1-1 of rotenone (1/16)
and 0.611 mg ml -x of pyrethrins (15/16).
A r e the combined effects o f rotenone and Hence:
pyrethrins additive, synergistic or
antagonistic? -
(D)I
- q-
(D)z
- -
0.041 0.611
This question has been the subject of debate (Dm)~ (Din)2 0.144 q-0.----~ = 0.989
by a number of investigators (see summary
by Finney s, p. 149). Analysis by the We conclude that the 1:15 mixture of insec-
median effect principle is now illustrated. ticides gives an effect which is strictly addi-
Since we have already shown that the tive (less than 1% deviation from unity in
system is higher order (m ~> 1 ), equation 3 the above equations), whereas the 1:5 mix-
should be used. Note, however, that at the ture is mildly synergistic (0.89) by 11%
specific dosage of the drug c o m b i n a t i o n under the specified conditions where the
producing a median effect, i.e. when (fi)x.2 mixture produces a 50% fly kill. This con-
= ffuh.2 = 0.5, the value of (fa)l,i/(fu)Ll = clusion is in agreement with those obtained
1, and the value of m becomes irrelevant by the application of probit analysis as dis-
since 1m = 1 for all values of m. Thus, cussed by Finney 8.
equation 3 can be evaluated under these dr- By the use of equation 3, we may calcu-
TABLE H. Analysis of toxicity of rotenone and pyrethrins, alone and in mixtures to house-flies.
The data from Table I were analysedaccordingto the relation, log (fa/fu) = m log D - m log Din, or the median-
effect plot, y = log (fa/fu)v. x = log (D) (see Fig. 2).
Median effect
dose a Lmear regression
Dm (EDs,) Slope of plot b coefficient
Insecticide ( m g m1-1) m r

the relative potency of rotenone to pyre-

Rotenone First series 0.157 2.757 0.995
thrins is 5.79 in the first series and 6.03 in Second series 0.144 2.500 0.995
the second series, a difference of only Pyrethrins First series 0.910 2.528 0.996
4.15%. These derived values are in close Second series 0.868 2.297 0.984
agreement with those calculated from the Mixture (1R:5P) First series 0.450 2.520 0.994
same data by Finney 8 who fitted parallel Mixture (1R:15P) Second series 0.652 2.533 0.983
probit lines to the data. In contrast to the
Menn value ofm for all insecticides (_+ S.D,) 2.523 -+ 0.146
median-effect equation which is based on
the mass-action law principle, the empiri- aDm can be calculated from the intercept of the median-effect plot on the x-axis (y = 0). Thus, Dm is the antilog of
cal probit equation assumes log-normal dis- ( - y intercept/m ).
tribution of the dose--effect relationships. bm is the slope of the linear regression line of the median-effect plot.
TIPS - November 1983 453

late the contribution of each drug in the TABLEIlL Comparisonof calculatedand observedlethalgiasof rotenoneand pyrethrmsm combination
combinations to the lethalities, assuming
Total dose (concentration) Calculated a Observed
summation of effects. This has been done of insecticides fracaonal fractional
for each concentration of the mixture of in combinations Dose conmbuted by lethality lethality
rotenone and pyrethrins for both series mg ml -~ rotenoue pyrethrms [tfah.~] [(faha]
(Table III). The fly kill predicted and that
Rotenone Pyrethnns
observed agree very closely for the series in First series (Rotenon¢: Pyrethrins = 1'5)
which the rotenone to pyrethrins ratio was 0.30 0.05 0.25 0.200 0,27
1:15. For four of the five concentrations 0.45 0.075 0.375 0.423 0.53
used, the difference between predicted and 0.60 0.10 0.50 0.611 0.64
calculated values was 5% or less, whereas 0.875 O. 1458 0.7292 0.809 0,82
for the fifth value, the difference was 8%. 1.175 O. 1958 0.9792 0.903 0,93
The differences showed no marked devia- Second series (Rotenone:Pyrethrms = 1.15)
tion from summation in the experimental 0.40 0.025 0.375 0.231 0.23
dose-effect range. In contrast, for the mix- 0.60 0 0375 0.5625 0.443 0.48
ture containing the insecticides in a ratio of 0.80 0.050 0.750 0.613 0.61
1.20 0.0750 I. 125 0.807 0.76
1 part rotenone to 5 parts pyrethrins, the
1.60 0.1(30 1.50 0.893 0.93
observed fly kill was always higher than
that predicted for summation. Thus, there aThe calculated fractional lethahty [(J'ah a] was obtained from the equauon 3:
was synergism at all concentrations; and
the synergism was most prominent at lowfo
(f.),j L(D,,,), (Dm),J
values: thus the value forfi = 0.27 (27%
kill) was 35% larger than predicted. where (D), and (D,,,)Ireferto to.none, and (D)I and (D,,,). referto pyrethrins. A meanvalueof slope(m) was
used, i.e. from Table IL for the first series m (rotenonc) ts 2.757 and m (pyrethrins) is 2.528. hence m = 2.643.
The isobologram procedure is difficult to Sample calculaOon for 0.60 mg ml ' of both drugs m first series.
apply to the available data because it (D,nh = 0.157 mg ml-'; (D,n)s = 0.910 mg m l - '
requires many more measurements. To the
very limited extent that it can be applied (fa),.s=F0.10 + 0 5 0 12.643 = 1.1864),() = 1.5711
(fuha L0 157 0.910 _1
without extrapolation, its conclusions are in
harmony with those obtained by the median (fa)xa = 1.5711 (fu)x.a, but since (fu)x,2 ~ [I - (fa),J]
effect concept. (,fa)~,~ = 1.5711 [(l -- ora),.s] = 0.6110
Note that in a hypothetical simafon when the values of m , and m s are markedly different, the use ofm = (m, 4-
The product of fractional activities gives
ms)/2 may not be a good apptoxinumon o f m l a , in this circumstance, the method of using equaUon 5 at tfa h.2 =
startlingly different resdts. There is one set 0.5 [i.e. at (Dmh.*] Is recommended (as iUnstrated m the text) since the magnitudes of m , and ms can be &sre-
of values from each series which can be garded under this resmcted con&tion.
used directly (Table IV), but the same con-
clusions apply to all values, if one interpol- effects has been derivedL I! has an addi- intercept the plot of the more potent of the
ates over the required range. In both series, tionalterm when compared with equation 3 two drugs. The theoretical plot for equation
this analysis indicates that the drugs show and is given by: 6 and experimental examples have been
very marked synergism. We believe this to provided recently (see Ref. 3 for details).
indicate that this method (based on
restricted mechanistic assumptions) is not ~'~J
[
(foh~l.-L (Dh
= (-D~.h + ~
(D)2 (Dh(D)2
+ (Dm),(D..~ General utility of median effect analysis
adequate to describe summation of com- We have demonstrated that application
(6) of the median effect equation to whole ani-
bined drug effects in this system. As indi-
cated earliers, the fractional product When the mixture of mutually nonexclu- mal toxicity data leads not only to rigorous
method applies only under the following sire drugs is used, the median-effect plot of definition of summation of drugs effects but
conditions: (i) the effects of the drugs are y ~- log [(fah~/(f,),~] with respect to x = also provides a cost-effective analysis and
mutually nonexclusive (i.e. the drugs act log [(Dh + (Dh] will give a concave additional information about the properties
independently of one another), (ii) the upward curve which at high doses will of the system. Unfortunately, there are very
dose-effect curves for both drugs are
hyperbolic (Michaelis-Menten or first TABLE IV. Comparison of the fractional product method and the ixesent method
order). In the present example, neither of Insect~kle F ~ fly ~1 ¢fi)
these criteria applies, and consequently the and dose Calculatedby Webb's Calculated by the present
fractional product method gives erroneous (nagm F ~) Obsctvc(I method assumingaddiuvity" method assuming ndditivay b
results.
First series
Romnone, 0. I 0.24
What happens if the effects of two drugs pmthrths, 0.5 0.20
are mutually nonexclnsive? Mixture, rotenone, 0.1
The above example for rotenoue and pins pyrethrins, 0.5 0.64 0.392 0.611
pyrethrins indicates that the effects of both Second seams
drugs are mutually exclusive (i.e. parallel Rotenon¢, 0.1 0.28
median-effect plot for the mixture as shown Pyrmhrim, 1.5 0.72
Mixture, rocenone, O. I
in Fig. 2). For the situation when the effects
plus pytethrins, l 5 0 93 0 798 0 893
of both drugs are mutually nonexclusive
(i.e. the drugs have different modes of • C_~culated ~ m [ I - (f,,),.) = [ I - (f,,).) 11 - (f.).) or q',,).. = ( f . ) , x ( f . ) . .
action or act independently) the general b Calculat~ froHI the eqttagion 8.s iUusuag~ Ill Table m . Note that all other combinauons that were calculated with
equation for describing the summation of the present method according to Table Ill cannot be calculated with Webb's method
454
few experiments on drug or toxicity effects
which have been designed appropriately for
this type o f analysis. O n e needs concentra-
tion-effect curves for each drug alone and
for at least one constant ratio mixture.
The identical principles have been
recently applied successfully to the simple
and efficient analysis o f the effects o f mul-
tiple inhibitors on e n z y m e s (Chou and
Talalay s, Kremer et al.ra; Finotti and
Palatinial; Steckel et al. ~). In each case the
method provided information on whether
or not s u m m a t i o n of inhibitor effects was
observed, whether the inhibitors were
mutually exclusive or nonexclusive, and
whether the system was f a s t or higher
order. All of these conclusions could he
drawn without knowledge of conventional
kinetic constants (gm, Vma~ or KL) and
involved no assumptions whether the
m e c h a n i s m of inhibition was competitive,
non-competitive or uncompetitivea,6,ra.
Furthermore, the median effect equation
has been used to obtain accurate values
o f ID6o, EDso or LDso in enzyme sys-
tems a-7,xo.aa,~a, in cellular systemsx4-x~,
and in animal systems xT,ra. An alternative
form o f the median-effect equation xa has
been used for calculating the dissociation
constant (K,) o f ligands for pharmacologi-
cal receptorsra. It has also permitted t h e
analysis of chemical carcinogenesis data,
and predicted especially accurately tumor
incidence at low dose carcinogen
exposuretL By using the median-effect
principle, the general equation for describ-
ing a standard radioimmunoassay or ligand
displacement curve has been derived
recently by Smith 2°. It has also been used to
show that there is marked synergism a m o n g
chemotherapeutic agents in the treatment of
hormone responsive experimental m a m -
mary caxcinomasra. It is perhaps surprising
that the median-effect analysis appears to
be valid for the analysis o f systems in which
both quantal (all-or-none) and graded
effects are being measured.
The future
It appears likely that the median effect
principle and plot will permit analysis o f
combinations o f effectors o f various other
types. In order to examine its validity, we
hope that other workers will bring to our
attention suitable data or will consult us
regarding the analysis o f their o w n data.
Acknowledgements
The authors receive research support from
the National Institutes of Health (AM 07422
and CA 27569) and from the American Cancer
Society (SIG-3).
Reading list
1 Loewe, S. (1957)Pharmacol. Rev. 9, 237-242
2 Webb, J. L. (1963) m Enzymes and Metabolic
lnhibitors, Vol. 1, pp. 66-79 and 487-512,
AcademicPress, New York
© 1983 ElsevterScience PubhshersB V , Amsterdam 0165- 6147/83/$01 00
TIPS - N o v e m b e r 1983
3 Chou, T.-C. and Talalay, P. (1981) Eur. J. 16 Chou, T.-C., Burchenal, J. H., Schmid, F. A.,
Biochem. 115, 207-216 Braun, T. J., Su, T.-L., Watanabe, K. A., Fox,
4 Chou, T.-C. (1976)J. Theor. Biol. 59,253-276 J. J. and Philips, F. S. (1982) Cancer Res. 42,
5 Chou, T.-C. (1977)J. Theor. Biol. 65, 345--356 3957-3963
6 Chou, T.-C. and Talalay, P. (1977) J. Biol. 17 Chou, T.-C. (1980) Carcinogenesis 1,203-213
Chem. 252, 6438-6442 18 Teller, M. N., Stock, C., Bowie, M., Chou, T.-C.
7 Bonnias, M. (1982) Biochem. Pharmacol. 31, and Budinger, J. M. (1982) Cancer Res. 42,
2769-2775 4408-4412
8 Finney, D. J. (1952) in ProbitAnalysis, 2ndedn, 19 Murphy, K. M. M. and Snyder, S. H. (1982)Mol.
pp. 146-153, Cambridge University Press, Cam- Pharmacol. 22, 250-257
bridge 20 Smith, S. W. (1982)J. Theor. Biol. 98, 475--499
9 Le Pelley, R. H. and Sullivan, W. N. (1936) J.
Econ. Entomol. 29, 791-797
Ting-Chao Chou was born m Taiwan, receivedhis
10 Kremer, A. B., Egan, R. M. and Sable, H. Z. Ph.D. in Pharmacologyfrom Yule Universityand
(1980)J. Biol. Chem. 255, 2405-2410 his postdoctoral training at The Johns Hopkins
11 Finotti, P. and Palatini, P. (1981)J. Pharmacol. University. He is now an Associate Member at
Exp. Ther. 217,784-790 Memorial Sloan-Kettering Cancer Center and
12 Steckel, J., Roberts, J., Philips, F. S and Chou, Associate Professor of Pharmacology at Cornell
T.-C. (1983) Biochem. Pharmacol. 32, 971-977 University GraduateSchool of MedicalSciences.
13 Chou, T.-C. (1974) Mol. Pharmacol. 10,
235-247
Paul Tululayis the John Jacob Abel Distinguished
14 Friedman, S. J. and Skehan, P. (1980) Proc. Natl Service Professor of Pharmacology and Experi-
Acad. Sci. USA 77, 1172-1176 mental Therapeuticsat Johns Hopkins University
Medical School and a former director of this
15 Rosenberg, R. (1981) Biochem. Biophys. Acta
649, 262-268 department.
Note added in proof:
After submission of this paper, an improved method for calculating the drug combination index (CI) at
various effect levels has been developed. Equation (5) has been extended to define CI for two drugs as follows:
CI = (D)l/(Dx)l + (D)~J(Dx)2,where (D01 and (D0"- are the doses required to achieve a gwen effect level
for each drug, Le. a specified value offa; (D)l and (D)e are the doses of each drug m a given mixture which
gives the same fa. When C1 values are 1, < 1, or > 1, It may be concluded that summation, synergism, or
antagomsm, respectively, have been observed at the particular effect level examined.
Computer programs for automated analysis have been written and may be obtained from the authors. A
more detaded description will be pubhshed (T.-C Chou and P. Talalay (1984)Advancesin Enzyme Regulanon
23, m press).
H o r m o n e s and
n e u r o t r a n s m i t t e r s in
milk
E. Hazum
Department of Hormone Research, The Weizmann Instituteof Science, Rehovot, 76100, Israel.
Several hormones and neurotransmitters are present in milk f r o m humans and other
mammals, e.g. gonadotropin-releasing hormone, thyrotropin-releasing hormone,
epidermal growth factor, morphiceptin and fl-casomorphin, morphine, TSH,
A C T H , prolactin, gonadotropins, etc. The presence of significant quantities o f these
hormones and neurotransmitters in milk raises the question not only o f their role but
also o f their origin. They can be synthesized or processed in the m a m m a r y gland or
excreted into milk through various pathways. In the suckling mammals, the hormones
and neurotransmitters may gain entrance into the body through the gastrointestinal
tract and could have physiological and pharmacological significance by interacting
with their specific receptors.
Introduction importance and have evolved differently,
Milk is a mammal-specific biological both in quality and quantity, to m e e t t h e
fluid. It is the f'ast and sole food of newborn specific requirements o f neonatal develop-
m a m m a l s for a certain period o f time. The ment in various m a m m a l s . The unique sig-
neonatal phase is characterized by rapid nificance o f milk as a food for the neonate
growth and development o f several organs has resulted in the biochemical aspects o f
including the liver, kidneys and endocrine milk being extensively investigated since
glands, which are not fully formed at birth. t h e early eighteenth century.
Therefore, the milk constituents are o f great The composition of milk changes with