Journal of Erhnopharrnacology.

32 ( 1991) 117-l 33 117

Elsevier Scientific Publishers Ireland Ltd.

Can et~nop~a~aco~ogy contribute to the development of new

anticancer drugs?

Geoffrey A. Cordell, Christopher W.W. Beecher and John M. Pezzuto

Programfor Collaborative Research in the Pharmaceutical Sciences, College of Pharmacy, Universiry of Minois at Clricago. Chicago.
Illinois. 60612 (U.S. A.)

Ethnopha~acolo~c and ethnom~cal information has been poorly utilized in the past in the search for new and effective
treatments for cancer. In spite of this, plants have been a very viable source of clinically useful compounds. leads for synthetic
modification and tools for mechanistic studies. In this paper, a new strategy for the discovery of anticancer agents from plants is
proposed in which ethnomedical information is correlated against pertinent published chemical and biological information. resulting
in a prioritization of plants for collection. Authenticated plants are extracted and the extracts tested in a broad array of more than
20 human cancer cel1 and mechanism-based assays through a cooperative research program involving a university, a research institute
and a pharmaceutical company. Bioactivity-directed fractionation will be carried out at all three sites. with a view to identifying novel
compounds which will serve as candidates for preclinicat testing,

Drug discovery has been a goal of mankind
since prehistoric times. In the past 100 years, the
sources of drugs have expanded to include higher
and lower plants, marine organisms, animals and
arthropods as well as total and partial synthesis. If
one decides that, because of structural and
biological diversity, terrestrial plants offer a uni-
que, renewable resource for the discovery of
potential new drugs and biologicat entities, the
most pertinent question is how to find the prover-
bial needle (active compound) in the haystack
(plant kingdom). Since conservative estimates in-
dicate that there are about 250,000 species of
flowering plants on this planet, of which it is
estimated that 155,000 are found in the tropics
(Prance, 1977), a rational strategy for drug
discovery is required for success.
Presented at the First International Congress on Ethnophar-
macofogy, Strasbourg, 5-9 June, 1990.
Correspondence to: Geoffrey A. Cordell, Program for Col-
laborative Research in the Pharmaceutical Sciences, College of
Pharmacy, University of Illinois at Chicago, Chicago, IL
60612, U.S.A.
Plants continue to be an exceptionally viable
source of biologically active natural products
which may serve as commercially significant en-
tities in themselves or which may provide lead
structures for the development of modified
derivatives possessing enhanced activity and/or
reduced toxicity. From the small portion of the
250,000 species of flowering plants that have been
investigated, about 120 therapeutic agents of
known structure are isolated for commercial pur-
poses from about 90 species of plants (Farnsworth
et al., 1985). Some representative examples are
given in Table 1. These entities are employed for
the treatment of a broad array of physiolo~cal
states, including cancer. Fully 74% of these 120
plant-derived therapeutic agents were discovered
based on ethnomedical records (Farnsworth et al.,
1985).
There are four basic approaches for the selec-
tion of plants that may contain new biological
agents from plants: the random, the taxonomic,
the phytochemical and the ethnomedical. In the
random approach, all available species are col-
lected, irrespective of prior knowledge and ex-

037~-~74l/SO3.50 0 1991 Elsevier Scientific Publishers Ireland Ltd.

Published and Printed in Ireland
118

TABLE 1 There are merits to each of these approaches

PRINCIPAL PLANT-DERIVED DRUGS IN THE UNITED depending on the particular circumstances. In this
STATES paper, we will discuss the ethnopharmacological
approach, and try to answer the question “Can
Plant Family Compound(s)
ethnopharmacology contribute to the develop-
Ananas comosus
ment of new anticancer drugs?” posed in the title.
Bromeliaceae Bromelain
Camelia sinensis Theaceae Caffeine Before we attack this issue in detail, however, we
Theophylline need to examine the history of previous drug
Cassia aculifolia Leguminosae Sennosides A&B discovery efforts for new plant anticancer agents
Cassia angustifolia
and discuss what might be differences between the
Catharanthus roseus Apocynaceae Vincristine
situations then, now and in the future.
Vinblastine
Cinchona species Rubiaceae Quinine It is well established that plants have been a
Quinidine useful source of clinically relevant antitumor com-
Colchicum autumnale Liliaceae Colchicine pounds. Indeed the Catharanthus alkaloids,
Digitalis purpkea Scrophul-
vincristine (Oncovin@) and vinblastine (Velban@),
ariaceae Digitoxin
Dioscorea species Dioscoreaeae Diosgenin
are two of the most important cancer
Duboisia myoporoides Solanaceae Atropine chemotherapeutic agents in current use, followed
Scopolamine by the podophyllotoxin derivative, etoposide. Col-
Ephedra sinensis Ephedraceae Ephedrine chicine derivatives are also used to treat cancer in
Papaver somniferum Papaveraceae Codeine
some countries of the world and the pyrrolizidine
Morphine
alkaloid monocrotaline is used topically in China
Papaverine
Physostigma venenosum Leguminosae Physostigmine to treat skin cancers. A number of other plant-
Pilocarpus jaborandi Rutaceae Pilocarpine derived antitumor agents have been subjected to
Podophyllum peltatum Berberidaceae Podophyllotoxin recent clinical evaluation, including taxol (Wirenik
Rauvolfia serpentina Apocynaceae Reserpine
et al., 1987), harringtonine (Li et al., 1983),
Rauvolfiu vomitoria Apocynaceae Deserpidine
homoharringtonine (Ajani et al., 1986; Legha et
Reserpine
Sanguinaria canadensis Papaveraceae Sanguinarine al., 1984; Neidart et al., 1986), lo-hydroxy-
Veratrum viride Liliaceae Cryptenamine camptothecin (Hsu and Yang, 1984), teniposide
(Muss et al., 1987) and ellipticine derivatives
(Caille et al., 1985; Einzig et al., 1985; Rouesse et
al., 1985).
There have been several efforts to discover new
perience. In the taxonomic approach, plants of a anticancer agents from plants, some organized by
given genus or family are deemed to be of interest, the National Cancer Institute and one by the Eli
and are sought from diverse locations and Lilly Company. The pharmaceutical company
evaluated. In the ethnomedical approach, credence sponsored program was based on folklore use (not
is given to information on the medicinal use of the necessarily for cancer) with testing for CNS, an-
plant, and, based on this information, the plant is timicrobial, insecticide and antitumor activity.
collected and evaluated. For each of the collec- Four important compounds came out of the 400
tions, the most rational method of proceeding in- plants that were initially analyzed: vincaleuko-
volves evaluating the material in a range of blastine, leurocristine, 9-methoxyellipticine and
bioassays. Active leads are then subjected to acronycine (Barrios and Farnsworth, 1979).
bioassay-directed fractionation procedures for The National Cancer Institute program was in-
procurement of the active principle(s). In the itiated by Dr. Jonathan Hartwell who began the
phytochemical approach. particular compound investigation of the May apple, Podophyllum
types, e.g. indole alkaloids, are regarded as being peltatum, a plant with a long and extensive
of biological interest, and plants likely to have folkloric use for the treatment of various kinds of
related compounds are collected and evaluated. cancer and warts. It was many years before the

related, clinically useful, products teniposide and
etoposide, emerged through synthetic efforts at a
pharmaceutical company. It was Hartwell too,
who assembled what remains today as the only
compilation of the ethnomedical use of plants
against cancer (Hartwell, 1967-1971). Hartwells’
list contains over 3000 different species of plants
and describes in considerable detail the various
uses of those plant parts.
For much of prehistory there is no documentary
evidence of the use of plants as anticancer agents.
The earliest written text is the Ebers Papyrus
which dates to 1550 B.C. Forty plants were recom-
mended in this work including barley, flax, ab-
sinth, coriander, figs, onions, garlic, dates, juniper
and grapes. Several of these plants are now known
to contain compounds toxic to cancer cells (e.g.
grapes, juniper and flax) (Perdue and Hartwell,
1969).
In 1957, the National Cancer Institute, through
the newly created Cancer Chemotherapy National
Service Center, began a program of screening of
plants against a leukemia model (Ll210 lym-
phocytic leukemia in mice), a carcinoma model
(adenocarcinoma 755 in mice) and a sarcoma
model system (sarcoma 180 in rats). Substantive
plant collection for this program began in 1960. In
1966, the Walker 256 carcinosarcoma replaced the
A755 and Sl80 assays, but this assay was later
found to be very sensitive to phytosterols,
saponins and tannins and was subsequently
dropped.
The program blossomed from being an in-
tramural program in the 1960s to an extramural
program with several contract sites in the 1970s.
Plant materials, provided by the United States
Department of Agriculture, were collected based
on all four of the strategies listed above, although
random collection predominated. The Hartwell
text was occasionally used as a source of
ethnomedical information. The assays used initial-
ly were the P-388 lymphocytic leukemia assay in
mice and the KB carcinoma of the nasopharynx in
cell culture. Both of these assays were conducted
at external contractual sites, and eventually, as
delays increased for the testing of fractions for the
bioactivity-directed fractionation, it became clear
that a quite different strategy was needed. In the
119
early 198Os, the NC1 discontinued this program
and then several years later reintroduced a new in-
tramural program in which the goal is to randomly
collect plants from three designated areas of the
world and screen them against a broad array of
human cancer cell assays.
The NC1 program in the period 1960-1986
screened over 35,000 plant species and 108,330 ex-
tracts against the murine tumors in vivo or in vitro
or the KB test system. Of these extracts, 4149
(3.83%) were active, representing 1410 genera and
2935 species. From among the 2000 crystalline
plant-derived compounds that were tested, 95 were
identified for special testing and 11 approved for
extensive tumor panel testing (Douros and Suff-
ness, 1980). These compounds included brucean-
tin, maytansine, colubrinol, indicine-N-oxide,
tripdiolide, taxol, homoharringtonine, ellipticine
and bouvardin. A number of other plant-derived
compounds, including tylocrebrine, lapachol.
camptothecin, acronycine, emetine, thalicarpine
and tetrandrine, were examined in this program in
various clinical trials.
Two features of anticancer drug discovery from
plants have become apparent: (i) the broad range
of plant families with active species and (ii) the
broad range of natural product structure types
which demonstrate in vivo activity.
As an example of the first feature, consider a
study of the results available to 1975. It was shown
(Barclay and Perdue, 1976) that in the
Pteridophyta 14 of 99 genera (14.1%) had active
species, the Gymnosperms had 25 of 60 genera
(41.7%) tested active, and in the Angiosperms 1066
of 4557 genera (23.4%) showed some type of an-
ticancer activity. Based on the criteria developed
at that time, a number of plant families were iden-
tified as being of high interest. What is lost in this
analysis of course is that the same compound may
be responsible for activity in a substantial number
of genera; podophyllotoxin would be an example.
With respect to the second feature, there have
been any number of review articles which have ex-
amined the breadth of natural product structure
types displaying some type of anticancer activity
(Danieli, 1971; Jewers et al., 1973; Hartwell, 1976,
Cordell, 1977; Stiffness, 1985). The conclusion has
been drawn that almost every group of terpenoid,
120
alkaloid, lignoid, shikimate-derived or acetate-
derived natural product has a member which
shows anticancer activity (Cordell and Farn-
sworth, 1976; Suffness and Douros, 1979). In some
instances, it has been established that within a
given plant there may be two, or even three, classes
of compounds which display activity.
In terms of ethnopharmacology, there are im-
portant implications. One is that the extensive
prior literature may already allow for some ra-
tionalization (positively or negatively) with respect
to biological potential of a reputed use. A second
important feature is that diversity of plant material
should not predicate biological potential, i.e. that
activity could be found almost anywhere. Indeed,
Hartwell’s review of the literature covered reports
on 184 families of higher plants.
A retrospective analysis of the NC1 program
(Spjut and Perdue, 1976) showed that the percen-
tage of active leads based on ethnomedicine was
substantially above that based on taxonomy,
which itself was more than the active leads iden-
tified through random screening. Thus, anthelmin-
tics (29.3%), fish poisons (38.6%), arrow poisons
(52.2%) gave higher activity than the Leguminosae
(18.4%) and the overall activity rate (3.8%) from
random collection. The only published study in
which plants were investigated for anticancer ac-
tivity based on ethnopharmacology is a report by
Hostettmann’s group (Chapuis et al., 1988) in
which 75 species of East African plants used
ethnomedically were evaluated for their cytotoxici-
ty in a human colon cancer cell line. Twenty-nine
of these (38.7%) were active according to establish-
ed criteria. In unpublished studies in our
laboratories (Cordell and Pezzuto, unpublished), a
group of 30 plants used in Thai traditional
medicine were evaluated in three cell lines (P-388,
KB and KB-vinblastine resistant). Of these, 15
species (50%) displayed cytotoxicity at 10 &ml or
less in one or more test systems.
But all that is in the past, for while there has
been substantial success in identifying agents that
will be effective against a variety of systems, a
number of tumor types have proven resistant to
chemotherapy using the existing agents. Thus a
new strategy for the discovery of plant-derived
natural products which have anticancer activity is
called for. The National Cancer Institute has
ongoing an intramural program in which random-
ly collected plants are screened against an exten-
sive battery of cultured cell types (Alley et al.,
1988; Scudiero et al., 1988; Shoemaker et al., 1988;
Suffness, 1987). We have chosen a strategy that
combines cellular and molecular assays. As
described previously (Suffness and Pezzuto, 1991),
extreme care must be taken in designing such a
program, taking into account mechanisms of
potential cancer inhibition. The group that is in-
volved in the program is based at the University of
Illinois at Chicago, with partners at Research
Triangle Institute under the leadership of Dr.
Monroe Wall, and at Glaxo Group Research Ltd.
in England, under the leadership of Dr. Timothy
Harris.
New strategies for anticancer drug discovery
What then are the key factors in developing a
successful program aimed at drug discovery of new
anticancer agents from plants based on ethnophar-
macological information? We believe that there
are seven essential components:
(i) Rational plant selection and collection,
followed by authoritative taxonomic
verification.
(ii) Availability of diverse, proven cytotoxicity
and mechanism-based bioassays.
(iii) Unequivocal dedication to bioactivity-
directed fractionation.
(iv) Ability to perform not only effective struc-
ture determination of a broad array of novel
compounds, including new natural product
skeleta, but also the ability to synthesize
analogues to potentiate activity.
(v) Ability to develop new strategies in the
botanical, chemical and biological areas.
(vi) Critical decision-making capabilities regar-
ding potential candidates for future
development.
(vii) Knowledge and successful experience in
preclinical and clinical development of
primary lead compounds.
There are several novel aspects to such a pro-
gram that we believe are essential in order to suc-
cessfully exploit ethnopharmacologic information,
three of which will be discussed. The first is a two-
pronged strategy for the selection and collection of
the plants to be tested. Another is a broad array of
cell-based and mechanistically oriented bioassays.
The third is a computerized literature surveillance
and evaluation program which is critical for the in-
itial selection of candidate plants for collection,
and to continuously evaluate the available
literature, leading to a more effective decision
making process, including prioritization of active
leads.
(a) Plant selection and collection
Two novel strategies for the collection of plants
have been developed. One of these involves a net-
work of collaborating botanists/collectors in 15
different countries in primarily tropical regions of
the world. These collaborators will obtain plants
based on whether the plant is used ethnomedically
for the treatment of certain cancer-related, specific
diseases. A set of ethnobotanical usages (Table 2)
has been developed that reflects disease states
bearing some relevance to cancer or a cancer
symptom. The second approach to plant selection
involves the use of the NAPRALERT database,
and particularly the ability of this system to con-
duct relational searches which can then be
prioritized based on selected weighting factors in-
volving existing experimental data. This process
will be described subsequently.
Plant collection must be organized and/or con-
ducted by experienced plant collectors and tax-
onomists for a program to be successful.
World-wide field experience and botanical contact
network, coupled with capabilities regarding iden-
tification and authentication, are essential, par-
TABLE 2
ETHNOMEDICAL USAGE CLAIMS THAT WILL BE
QUERIED
(9 Cancer treatment
(ii) Immune disorders
(iii) infectious diseases
(iv) Parasitic diseases
(v) Viral diseases
titularly since local names may refer to several
different plants. Complete authentication and
documentation of voucher specimens at recogniz-
ed herbaria are also needed.
(b) Primary and secondary bioassay capabilities
The second novel aspect to the program con-
cerns the scope of primary and secondary
bioassays to be conducted. For many years, lead
compounds for the chemotherapy of cancer were
pursued on the basis of general cytotoxicity (P388
or KB) or antitumor-based assays as described
above. As more has been learned about the evolu-
tion of the cancerous state, and the mechanisms of
action of agents which might interfere with that
process, it has become increasingly clear that a
drug discovery process which involves mechanism-
based assays, and which would select for specific
target processes, would provide a very useful alter-
native strategy.
In some areas of drug discovery, straightfor-
ward methods of bioassay are apparent. As an ex-
ample, fundamental structural differences between
prokaryotic and eukaryotic cells have led to the
discovery of therapeutic agents (antibiotics) that
can be used with near complete impunity by the
host. Similar to tests that are routinely performed
to determine drug-susceptibility of invading
microorganisms, attempts have been made to
assess the susceptibility of human cancer cells by
means of excising malignant tissue and performing
in vitro (stem cell) assays (Von Hoff and Weisen-
thal, 1980; Von Hoff et al., 1981). However, these
procedures have led to limited therapeutic advan-
tage. An additional approach has been to explore
the development of drugs on the basis of certain
differences that have been noted between “nor-
mal” and “cancer” cells (e.g., asparagine starva-
tion (Chabner and Myers, 1982) biological
response modifiers (Foon, 1989), cell differen-
tiating agents (Bloch, 1984), cell compartment
specific strategies (Sartorelli, 1988) prodrugs ac-
tivated at the site of the tumor (Pezzuto et al.,
1988)). Thus far, however, agents demonstrating
impressive specificity have not emerged. The fun-
damental problem is as follows: no explicit dif-
ferences between “normal” and “cancer” cells
122
have been defined that are suitably specific to
allow unequivocal therapeutic exploitation. The
common characteristic of presently available an-
ticancer agents is a poor therapeutic index. Thus,
in devising a program for the discovery of an-
ticancer agents, a number of philosophical issues
must be considered from the outset.
One strategy is to consider the search for drugs
that could modulate disease-specific processes that
are required for progression. One notable example
is the process of metastasis. Since a localized
primary tumor (not associated with a vital body
organ) can often be definitively resolved by means
of surgery and/or radiation therapy, the
phenomenon of metastasis is of key importance in
leading to the serious consequences associated
with this disease. A number of assays have been
constructed to screen for antimetastatic agents
(Hendrix et al., 1987; Liotta, 1986; Yamanishi et
al., 1973). A second impo~ant example of ostensi-
ble tumor-specificity is the process of angiogenesis.
Tumor-derived factors can stimulate the prohfera-
tion of endothelial cells and induce angiogenesis
(Folkman, 1986). The discovery of specific
an~ogenesis inhibitors (Crum et al., 1985;
Folkman, 1985) therefore, is a valid endeavor of
substantial interest.
Given the reality of finite resources, and the
ethical and fiscal constraints associated with using
in vivo models for screening of antitumor activity,
the question must ultimately be posed: what assay
(or battery of assays) would be most deductive for
screening plant extracts? In a fairly thorough
treatise dealing with this question (Suffness and
Pezzuto, 1991), it was eventually concluded that
no concise answer for this question is available.
Beyond the philosophical considerations that
are imbedded in the assay systems selected for
drug discovery, there are always elements of per-
sonal interest, capability and physical resources.
The battery of drug discovery assays that will be
described in the remainder of this section was
selected by our group after extensive introspective
analysis. As a result, we are confident that enabl-
ing technology has been assembled and that our
group is now in a position to discover novel
naturally occurring agents that may be useful for
the treatment and characterization of human
cancer. For our particular set of circumstances,
and based on the current state of knowledge, these
assays will maximize the probability of a signifi-
cant discovery. However, as new information
becomes available, either through the literature or
our personal experience, the nature of this battery
of tests will likely change. Further, we do not mean
to imply that other groups of researchers should
necessarily attempt to adopt the same approach; it
is most likely that other test procedures would be
more apropos for other groups of researchers.
A summary of our test systems follows:
- Eight human cancer cell lines (melanoma, sar-
coma, lung, colon, squamous cell carcinoma,
hormone-dependent and hormone-inde~ndent
breast, and hormone-dependent prostate)
suitable for cytotoxicity testing
- Human fibroblasts to assess selectivity
- ASK ghoma cells (to evaluate antimitotic ac-
tivity (Suffness and Douros, 1982) and ability
to elevate intracellular cyclic AMP concentra-
tion (Swanson et al., 1988))
- KB and KB-VI (drug-resistant) cells
- A reversal of drug resistance using KB-Vl cells
- P388 cells
- Topoisomerases I and II
- Protein kinase C
- Tubulin binding
- DNA nicking
- Aromatase and C-17,204yase assays
- Steroid hormone receptor binding assays
- Tyrosine kinase assays
Although the assays listed above may appear
rather diverse, the use of a combination of cellular-
and mechanistic-based assays was thoroughly con-
templated; there are several areas in which the
assays are complementary, and other areas
wherein we simply did not want to over-restrict
our ability to detect potentially important leads. A
brief description of some of the more notable
elements of this strategy will be presented below.
Some of the mechanism-based in vitro assays
were designed in direct analogy with the types of
molecular responses mediated by known (clinically
effective) antitumor agents. Thus, we will monitor
effects similar to those known to be mediated by
 123

Vinca alkaloids (tubulin depolymerization), taxol resistance also inhibit the binding of vinblastine or
(tubulin stabilization), camptothecin (topo- doxorubicin to the P-glycoprotein (Akiyama,
isomerase I inhibition), adriamycin (topoiso- 1988). Thus, attempts will be made to enhance
merase II inhibition) and bleomycin (DNA vinblastine-mediated cytotoxicity with drug resis-
cleavage). These are all reasonable avenues toward tant KB-Vl cells in culture.
novel drug discovery. Similarly, taking into ac- Although mechanism-based bioassays clearly
count recent advances in areas such as in- offer great hope in drug discovery programs, some
tracellular signal transduction, proto-oncogene possible shortcomings should also be deliberated.
expression and function, and cell-cell interactions, For example:
additional sites for the targeting of drugs become (1) With few exceptions which require addi-
obvious, as do experimental assay systems which tional clarification/development (see above), any
are amenable to large numbers of samples. Thus, toxic mechanism known to be facilitated by an an-
we will monitor protein kinase C activity, tyrosine titumor agent is not limited to malignant cells.
kinase activity, and intracellular CAMP concentra- Thus, additional factors are involved in yielding a
tion Additionally, antagonism of hormone- demonstrable therapeutic index, and these factors
dependent tumors with appropriate types of “an- are either unknown or impossible to accurately
tiestrogens” is of known clinical value, and our assess with uncomplicated in vitro test systems.
battery of assays is reflective of this important fact. (2) Good progress has been made in defining
An ancillary approach toward the discovery of certain mechanisms of antitumor activity, but it is
agents useful in the treatment of cancer involves unlikely that many known agents function through
drug resistance. It is well-known that resistance to a single mode of action, Thus, even though one
multiple drugs may develop in a clinical setting, mechanism may predominate (and be accurately
and essentially all patients who become refractory reflected in an in, vitro assay), secondary
die of their disease (Bellamy et al., 1988). A mechanisms may contribute in a synergistic man-
number of mechanisms may lead to drug ner toward overall cytotoxic and antitumor ac-
resistance, such as enhanced metabolic detoxifica- tivities.
tion of chemotherapeutic agents (e.g. conjugation (3) The possible number of mechanisms through
with glutathione) or decreased permeability. A which potential antitumor agents may kill cells are
mechanism that appears particularly important, either too numerous or too complicated to totally
however, involves increased expression of the mimic with in vitro test systems.
multidrug resistance (mdr) locus (Choi et al., 1988; (4) Natural products have proven useful in
Shen et al., 1986) the product of which has been defining mechanisms on which to base in vitro test
termed P-glycoprotein. It is hypothesized that the systems. It is likely, however, that additional
P-glycoprotein decreases intracellular drug ac- mechanisms which are at present totally unknown
cumulation by means of an energy-dependent drug remain to be uncovered.
efflux mechanism. This type of basic information Based on considerations such as these, a broad
has permitted the generation of new strategies for overall scope of our bioassay capability has been
drug discovery, two of which will be used in our retained. One adjunct procedure is the evaluation
program. First, it should be of interest to compare of test materials with a battery of human tumor
the cytotoxic potential of substances with cell lines. Although the magnitude of this endeavor
multidrug-resistant cells and the parent cell line. is substantially smaller than the recent NC1 in-
Activity of equivalent potency with these cell lines itiative (Alley et al., 1988; Scudiero et al., 1988;
would imply the presence of an agent to which the Shoemaker et al., 1988; Suffness, 1987), the
“resistant” cell line is susceptible, and this would philosophy is somewhat similar. First, we
be worthy of characterization. A second approach recognize the fact that cytotoxicity is neither
toward therapeutic reversal of drug resistance in- necessary nor sufficient for antitumor activity.
volves alteration of P-glycoprotein-mediated drug However, cytotoxicity is an activity that is consis-
efflux. Most agents that reverse multidrug tent with antitumor activity, and interference with
124

any mechanism required for cell survival will
mediate a positive response. Thus, in conjunction
with the results of other biological assays, these
results will aid in establishing correlations, and
help in deciding which materials to subject to frac-
tionation procedures.
Second, although cytotoxicity is demonstrated
by a large number of compounds that are obvious-
ly of little therapeutic interest, it may also be noted
that nonspecific cytotoxicity is also demonstrated
by virtually every known naturally occurring an-
titumor agent currently used in the clinic. This is
clearly illustrated by the data summarized in Table
3. Thus, if there is some potentially useful agent in
a plant extract, it is likely that we will detect a
positive response in our cell culture systems. As
described below, plants will eventually be rank
ordered and then subjected to fractionation pro-
cedures. Based solely on non-specific cytotoxic ac-
tivity, a plant will generally not be given a high
priority, but it also will not be abandoned.
It is conceivable that we may change our
primary screen of mechanism-based assays in due
course, and a plant placed on hold due to
nonmechanistically-defined cytotoxicity could
then be re-evaluated. Therefore, these procedures
will help to more fully characterize the biologic
potential of the plant materials we select for
evaluation, and this is considered a more judicious
procedure than ranking a potentially clinically
useful material as a “false negative” due to the
unavoidable situation of utilizing a highly
discriminatory screen.
Third, the use of a battery of human cell lines
derived from a variety of human tumor-types also
offers great theoretical and practical value. The
cells can be carried as solid tumors in athymic
mice. Thus, more advanced testing of active prin-

TABLE 3
EVALUATION OF THE CYTOTOXIC POTENTIAL OF KNOWN ANTITUMOR AGENTS

Compound Cell line tested (ED,, &ml)

Breast Sarcoma Lung Melanoma Colon KB KB-V 1 P-388 ASK

Actinomycin 0.0077 0.0027 6.5 8.4 0.0044 0.005 0.77 0.013 -

CBeta-hydroxy- I.1 0.056 1.0 0.24 0.085 0.13 4.7 0.24 -
withanolide
Bruceantin 0.003 0.0014 0.001 0.0074 0.00007 0.62 0.0067 -
Camptothecin 0.043 0.0095 0.0024 0.14 0.047 0.0064 0.0064 0.015 -
Colchicine 0.022 0.047 0.043 0.18 0.0071 0.0033 0.67 0.017 +
Daunomycin 0.029 0.1 0.073 0.12 0.041 0.06 0.6 0.012 -
Ellipticine 0.12 0.23 0.38 0.77 0.23 0.13 0.19 0.14 -
Harringtonine 0.017 0.01 0.036 0.077 0.014 0.0023 0.2 0.011 -
Homoharringtonine 0.012 2.15 0.019 0.045 0.0076 0.00073 0.23 0.0065 -
IO-Hydroxycamptothecin 2.0 0.01 0.030 0.10 0.14 0.026 0.51 0.0044 -
lndicine N-oxide >40 >40 >40 >40 >40 >40 >40 >40 -
Mithramycin A 0.16 0.22 >4 >4 0.24 0.26 1.4 2.98 -
Mitomycin C 0.61 1.14 0.0961 2.0 0.61 0.19 >20 0.072 -
Nitidine HCI 0.63 0.17 0.084 1.6 0.37 0.06 >20 0.66 -
Phyllanthoside 0.043 0.00071 0.059 0.0047 0.012 0.0038 23 0.083 -
Podophyllotoxin 0.041 0.012 0.0181 >20 0.01 I 0.013 0.059 0.0031 +
Taxol 0.0032 0.00041 0.038 0.025 0.0034 0.00041 2.3 0.031 -
Teniposide 0.32 0.27 0.28 5.0 0.075 0.13 >20 0.013 -
Thalicarpine 8.7 9.1 11.7 11.4 I.5 8.3 12.7 1.1 -
Vinblastine 0.0016 0.013 0.009 1.7 0.013 0.0066 1.9 0.012 +
Vincristine 0.013 0.013 0.533 0.38 0.008 0.0064 9.6 0.023 +

ciples is obviously facilitated. In addition, it is
hoped that certain test materials will demonstrate
cell-type specificity, and this will lead to isolation
and identification of selectively cytotoxic agents.
Based on structure and activity, such a selectively
cytotoxic agent could certainly be a candidate for
more advanced testing. Irrespective of the out-
come, however, the concept of selective cytotoxici-
ty implicitly suggests the presence of a cell-specific
receptor that differentiates one tumor-type from
another. Such a discovery would be of monumen-
tal importance in terms of developing tumor-
specific therapeutic strategies, and a cytotoxic
agent specific for one cell-type would greatly aid in
identifying the appropriate subcellular target. In
other words, this approach purports that dif-
ferences exist among these cell types which can be
exploited to therapeutic advantage, and we will at-
tempt to discover novel compounds that can be us-
ed to uncover these differences.
Once a group of plants is evaluated with our col-
lective battery of bioassays, some will be classified
“active”. Further, it is extremely likely that the
total number of active leads will exceed our im-
mediate fractionation capacity. Therefore, a deci-
sion network will be required to rank order the ac-
tive leads. Clearly, plants will need to be con-
sidered on an individual basis, and the collective
experience of the personnel involved will con-
tribute greatly in this matter. In certain cases, it is
expected that a plant will demonstrate a unique
response in a particular test system (e.g. a par-
ticularly intense response, or a particularly in-
teresting dose-response), and the investigator will
intuitively believe the plant should be fractionated.
This would automatically place a plant into the.
first category of highest priority. In general,
however, decisions need to be based on quan-
titative data, and standard criteria will be utilized.
Although it is not possible to devise an immutable
set of rules, guidelines must be established and re-
evaluated at periodic intervals. Since many of
these bioassay procedures are novel, the entire
issue response intensity cannot be rigorously
discussed until sufficient data are accumulated.
Obviously, response intensity is further com-
plicated by the fact that it is dependent on the ac-
tual concentration of active principle that is
125
present in the extract. Nonetheless, in terms of a
general starting point, the following aspects are ex-
amples of factors that are taken into account in
ranking active plants:
Priority I
- Active in one test system only (except P388).
(This type of response suggests the presence of
a novel active principle.)
- Active in more than one test system in which
there is no obvious correlation of the
demonstrated activities (e.g. protein kinase C
inhibition and DNA nicking). (This type of
profile suggests the presence of two distinct ac-
tive principles.)
- Active only in test systems wherein mechanism-
based activities correlate (e.g., ASK reversal +
tubulin polymerization; KB-V 1 + antagonism
of vinblastine binding; inhibition of growth
factor binding + inhibition of oncogene-
transformed cells).
- Active in one cell culture system only (including
P388) and any mechanism-based assay.
Priority 2
- Active in more than one cell culture system, but
not P388.
- Active in more than one cell culture system (not
P388), and also active in any mechanism-based
assay.
Priority 3
- Generally cytotoxic (including P388) and also
active in a mechanism-based assay.
Once the active leads are ranked ordered, a
single bioassay is selected to direct the fractiona-
tion to yield the active principle(s). After the active
principles are structurally defined, they are screen-
ed through the entire battery of assays. In con-
sidering the results, there are some specific
correlative factors that should be observed. The
ability to establish these correlations, when taken
in conjunction with the chemical structure of ac-
tive isolates, will aid in formulating hypotheses
126
regarding mechanism of action, and further con-
tribute toward an informed decision network
regarding more advanced antitumor testing. Some
examples follow:
(a) Agents isolated on the basis of tubulin
polymerization assays should also be active in the
test involving ASK cells. Otherwise, it may be sug-
gested that the agent does not enter the cell.
(b) Agents isolated on the basis of hormone
receptor antagonism should be active in certain
hormone-dependent cell culture systems. Similar-
ly, agents isolated on the basis of tyrosine kinase
mediated activity should be active with other
classes of cells.
(c) Of particular interest, agents that are iden-
tified as antiestrogens may also function to inhibit
protein kinase C activity (Bignon et al., 1989;
O’Brian et al., 1985, 1986; Suet al., 1985), and vice
versa.
(d) Agents isolated on the basis of selective
cytotoxic activity may function by one of the
mechanisms assessed in the remainder of the test
battery.
Finally, it will be of value to determine the
overall cytotoxic potential of the isolates obtained
on the basis of mechanistic assays. Some of the
isolates may not function by mechanisms that
would be reflected through cytotoxicity (e.g.
CAMP induction, estrogen antagonism), but
others will be procured on the basis of cytotoxicity
with a single cell line, or on the basis of
mechanisms that are consistent with a cytotoxic
response. In these cases, cytotoxic potential would
be determined with our primary battery of cell
lines, but the evaluation would also be expanded
to include additional human cell lines derived from
the same classes of human tumors. When consider-
ing the possibility of more advanced testing, the
spectrum of activity would be of importance. This
information should provide the strongest indica-
tion of the type of tumor to be used for initial
evaluations of efficacy using in vivo test systems.
(c) Literature surveillance
The third essential aspect of any drug discovery
program concerns the computerized literature
surveillance of those plants which show activity
for compounds which may be responsible for the
observed bioactivity. Such a dereplication process
markedly enhances the ability to isolate novel
bioactive compounds, since time expended on the
isolation and characterization of known active
compounds is minimized. Coupled with the em-
phasis on previously uninvestigated genera with
ethnomedical and/or experimental biological ac-
tivity, the approach will afford a high probability
of discovering new skeletal structures possessing
biologically interesting activity.
Use of NAPRALERT in the plant selection and
dereplication processes
The key to the plant selection and literature
surveillance programs is the NAPRALERT
(NAtural PRoducts ALERT) database initiated by
Dr. Norman R. Farnsworth in 1975.
NAPRALERT is a database of natural product in-
formation (Loub et al., 1985). It includes records
containing the pharmacology, biological activity,
taxonomic distribution and chemistry of natural
products (Farnsworth et al., 1981). Many aspects
of this database, in addition to its subject matter,
make it unique, including the fact that the
database is structured in a fully normalized rela-
tional format, which makes it possible to compose
extremely complex queries. The database grows at
an average rate of 500 articles each month, based
on articles culled from the surveillance of some 700
national and international journals (Farnsworth
and Loub, 1982); additional scanning of some
abstracting services is also performed. Most infor-
mation is stored in two ways, both as it appears in
the article and in a coded form that is a distillation
of the text. This strategy makes it very easy to
research any relevant topic and also provides
much of the power and utility of the
NAPRALERT system. The database currently oc-
cupies 800 megabytes of disk storage and has
records on some 90,000 scientific articles, 80,000
compounds and 40,000 organisms. Information is
regularly included on plant extracts and
ethnomedical preparations as they appear in the
literature. The database does not ignore early
literature, and has references to work as early as
the mid 1600s.
Plant selection and literature surveillance
Over the past 15 years, the staff of
 127

NAPRALERT has systematically gathered data Plant selection

on natural products (plant materials and their
isolates) from the global literature, including their
taxonomic distribution and pharmacological ac-
tivity. Accommodation into a relational data
structure allows for a selection and prioritization
process. The use of NAPRALERT, combined
with sophisticated search routines, can provide a
highly enriched selection of potential candidates
for novel cancer chemotherapeutic agents.
One of the prior uses of the NAPRALERT
system was a retrospective search on plants
reported to have an ethnopharmacological activity
related to mammalian reproduction (Griffen,
1988). The original literature search of
ethnomedical references afforded a primary list of
some 5000 reputed active plants. Following com-
puter analysis, 400 were judged to have suffkient
merit for collection and further experimental
analysis. When these 400 were tested in a single in
vivo assay, 30 were found that could reproducibly
demonstrate the specific, unique biological activity
that was sought (anti-implantation effects). Pure
active compounds were isolated and identified
from 15 of these plants (Fig. 1). If we accept the
conservative estimate that there are currently ap-
proximately 250,000 species of terrestrial plants,
this search strategy rapidly narrowed the field, and
produced an enriched list in which almost ten per-
cent of the plants screened were active. This
represents a tremendous savings of time, effort and
money.
The primary and secondary literature contains
three relevant bodies of information that need to
be coordinated. Firstly, there are many articles
that record the use of a plant for specific cancer-
related treatments by various peoples of the world.
Secondly, there are numerous references to plants,
their extracts or specific chemical compounds
which, when tested in experimental systems,
demonstrate biological activities potentially
associated with cancer chemotherapeutics. Finally,
there is a vast natural products literature on the
isolation and characterization of compounds from
various sources. In some cases, several of these
pieces of information are contained in the same ar-
ticle, identifying a particular compound in a par-
ticular plant with a particular biological activity
and thereby justifying an ethnomedical observa-
tion. In other instances, these three aspects (the
ethnomedical observations, the isolation of com-
pounds or the biological evaluation of a plant or
compound) may be presented in literature
references that are diversely located.
NAPRALERT is a unique resource that
recognizes the existence of each of these entities
and has acquired substantial retrospective
literature on anticancer natural products.
The search pattern described below is designed
to correlate the biological activity records with
both the botanical and chemical records in order
to develop a series of datasets of plants and com-
pounds which demonstrate chemotherapeutic ac-

1
5,000
400
plant plant 30
species plant
species
~...‘_~_’
::::..‘..
nb ‘::
w

P 2%
t
0.16% f 0.012%
Lab Results
NAPRALERT

Fig. 1. Plant selection and biological evaluation process in the WHO fertility regulating program.
128
tivities. These datasets can then be correlated
against one another in order to determine the
plants that have been reported to exhibit a perti-
nent biological activity in an experimental system
where that activity cannot be associated with any
chemical known to be in that plant. Finally, these
datasets can be correlated with the ethnobotanical
data to determine those ethnobotanical observa-
tions that can be supported by the experimental
literature and those which may suggest further ex-
perimental examination.
(i) Database analysis. The search patterns to be
followed require the establishment of four basic
datasets: goodplant, goodcom, goodall, and ethno
(see Fig. 2). These datasets are correlated to build
a final dataset, the weighted list, which represents
the initial step in the ranking process.
Since correlations can only be performed on like
objects this means that the correlations can only be
performed between the lists of plants; specifically,
with active
GOODCOM GOODALL
Plants
reported
used i
\ ethnomedical claim and provide a measure of the
GOODPLANTS
ETHNO
Fig. 2. The four datasets that will be determined
NAPRALERT database.
the three lists of plants are the goodplant, goodall,
and ethno lists. The sole purpose of the goodcom
list is the production of the goodall list.
NAPRALERT is then used to:
- Create a goodplant list, i.e. a list of plants
reported to exhibit a potentially useful
therapeutic activity in a published experimental
system.
- Create a goodcom list, i.e. a list of compounds
that are reported to exhibit a potentially useful
therapeutic activity in an established ex-
perimental system.
- Create a goodall list, i.e. a list of all plants that
have been reported to contain any of the com-
pounds in the goodcom list.
- Create an ethno list, i.e. a list of all plants that
have been reported to be used for a relevant
ethnomedical purposes.
The process culminating in the weighted list
begins with a list of all the plants identified in
ethno. A list of the experimentally-determined
biological activities that are referenced in
NAPRALERT has been examined, and from these
a list of those activities that are likely to be seen in
bioassays and fit a chemotherapeutic profile has
been developed (Table 4). The biological activities
on this list will undergo a continual review pro-
cess, and the list will be modified as new activities
are judged relevant. Each plant is given an initial
score of zero points; points will then be added or
subtracted according to specific rules. For exam-
ple, where there is experimental evidence that the
extract is active, points will be added. Where there
is a known active compound that has been iden-
tified in the species the plant will lose points, and
where an active compound is isolated from within
the genus the plant will also lose points. The sum
of assigned points will validate both the
need for further examination of a particular plant.
(ii) Search strutegy. The search strategy relies on
the assumption that, at a most fundamental level,
all forms of cancer chemotherapeutic agents in-
using the volve the modulation of some cellular biochemical
process, either directly through control of the
TABLE 4

BIOLOGICAL ACTIVITIES THAT WILL BE QUERIED

The activities selected are all regarded by NC1 as relevant to cancer chemotherapy. NAPRALERT has reported these activites and
many more for the past live years to NCI. Furthermore. NAPRALERT has conducted retrospective searches on these activities and
therefore can be regarded as a reasonably complete data source.

Acting through non-specific means (2) DNA dependent RNA polymerase inhibition
(3) DNA disruption effect
(1) Carcinostatic activity DNA intercalating effect
(4)
(2) Metastasis inhibition
(5) DNA methylase inhibition/stimulation
(3) Plant antitumor effect
(6) DNA polymerase inhibition (I, Il. III etc.)
(4) Crown gall tumor inhibition
(7) Interaction with DNA
(5) Prolongation of lifespan Malondialdehyde stimulation
(8)
(6) Tumor promotion inhibition
(9) Nucleic acid inhibition
(7) Wound healing acceleration
(10) Purine synthesis inhibition
(8) Wounded cell recovery effect
(11) Terminal deoxynucleotidyltransferase inhibition
(9) Anti-invasive activity Thymidine kinase inhibition
(12)
(10) Antimetabolic activity
(13) Thymidine transport rate decreased
(11) Antitumor activity
(14) Thymidylate diphosphate kinase inhibition
(12) Antiviral-induced tumor activity
(15) Thymidylate monophosphate kinase inhibition
(13) Antitumor-promoting activity Thymidylate synthase inhibition
(16)
(17) Topoisomerase inhibition (I or II)
Affecting cellular communications

(1) Adenyl cyclase stimulation Antimitotic

(2) Epidermal growth factor inhibition
Histone phosphorylation inhibition (1) Antimitogenic activity
(3)
Nonhistone phosphorylation inhibition (2) Antimitotic activity
(4)
(3) Antitubulin activity
(5) Phorbol ester antagonist
(4) Binding effect to tubulin
(6) Phorbol receptor modulation
(5) Microlilament disruption
(7) Platelet aggregation inhibition
(6) Microtubule assembly or disassembly inhibition
(8) Protein kinase C inhibition
(7) Microtubule depolymerization
(8) Microtubule effects
Affecting drug resistance
(9) Tubulin aggregation effect
(I) Ca2+ pump inhibition (l(J) Tubulin binding effect
(2) Glutathione transferase inhibition (II) Tubulin polymerization (enhancement or inhibition)
(3) Membrane permeability enhancement
(4) DNA repair inhibition RNA effects

Cytotoxicity (1) Interaction with RNA

(2) RNA binding
(I) Cell Cycle cytotoxicity (All phases) (3) RNA polymerase inhibition (I. II or III)
(2) Cytolytic activity (4) RNA scission effect
(3) Cytotoxic activity (5) RNA synthesis inhibition
(4) Human tumor stem cell inhibition (6) Uridine kinase inhibition
(5) Antiproliferative activity (7) Uridine transport rate decreased
(6) Cell proliferation inhibition
onal effects
Protein synthesis
(1) Cholesterol esterase stimulation
(I) Peptidyl transferse inhibition
(2) Cholesterol synthesis inhibition
(2) Polypeptide chain initiation factor I inhibition
(3) Estrogen synthetase inhibition
(3) Polyuridine-directed polyphenylalanine synthesis
(4) Progesterone I I-cy-hydroxylase inhibition
inhibition
(5) Hormone receptor binding decreased
(4) Protein synthesis inhibition (any)
(6) Steroidogenesis inhibition
(7) Testosterone modulating activities
DNA effects Tyrosine aminotransferase modulation
(8)
( 1) DNA dependent DNA polymerase inhibition (9) Tyrosine hydroxylase modulation
130

growth and replication process, or indirectly

through a biochemical control mechanism.
It is important to note that the activities listed in
Table 4 are divided into categories of those which
are similar. Each division reflects approximate
biological boundaries and correlates to one or
more of the bioassays proposed. We have created
these divisions since the results of an assay
reported in the literature may reflect more than
one activity, for instance an assay to detect
“Tubulin Binding Effect” may, or may not, corres-
pond to a “Micro~lament Disruption” assay.
Since another research group may seek
microtubule binding agents, the need exists for us
to identify all test results that affect microtubules;
therefore, the search strategy will score for each
occurrence of a category of activity and addi-
tionally score the specific report of an assay that
directly corresponds to the group bioassay, in this Fig. 3. An overview of the potential relationships between the
case “Tubulin Binding Effect”. three plant-related databases.
NAPRALERT will be searched for all known
natural substances that exhibit one or more of
these activities. In doing so, two goals will be ac-
complished: (i) the establishment of a list of active
compounds (the goodcom list) and (ii) the available, the ethnopharmacologic and
establishment of a list of active plants (the ethnomedical literature or current traditional
goodplant list). The criteria for the selection of a medical practices have been poorly utilized in the
plant or compound for either goodplant or good- past in the search for new and effective treatments
corn, respectively, are that: (a) the recorded activi- for cancer. In spite of this, plants have been a very
ty be one of the activities listed in Table 4 and (b) viable source of clinically useful compounds, leads
the activity was reported in an experimental for synthetic modification and tools for
system. mechanistic studies. Nowever, there is clear indica-
Once the lists, goodplant, all of the active plants, tion that ethnopharmacologic information pro-
and goodcom, all of the active compounds, have vides a substantially increased chance of finding
been determined, NAPRAL~RT will proceed to active plants relative to a random collection ap-
produce a list of all plants that contain any of the proach. In this paper, a new strategy for the
active compounds (the goodall list). The criteria discovery of anticancer agents from plants is pro-
for the selection of a plant into goodall requires posed in which ethnomedieal information is
that there is a report of an isolation of one of the critically evaluated against existing chemical and
compounds identified in goodcom in a part of the biological information and plants are prioritized
plant (Fig. 3). for collection. Authenticated plants are extracted
In this way, we anticipate that time spent in and the extracts tested in a broad array of more
isolation of known active compounds will be kept than 20 human cancer cell and mechanism-based
to a minimum, thereby allowing us to focus on assays through a cooperative research program in-
novel isolates of potential preclinical interest. volving a university, a research institute and a
pharmaceutical company. Bioactivity-directed
Conclusions fractionation will be carried out at all three sites,
with a view to identifying novel compounds which
Even though there is a wealth of information will serve as candidates for preclinical testing.
 131

Acknowledgements of protein kinase C by triphenylacrylonitrile antiestrogens.

Biochemical and Biophysics Research Communicarions 163,
1377-1383.
The authors are grateful to the Drug Synthesis Bloch, A. (1984) Induced cell differentiation in cancer therapy.
and Chemistry Branch, Division of Cancer Treat- Cancer Treatmenr Reports 68, 199-205.
ment, National Cancer Institute, for supplying Caille, P., Mondesir. J.M., Droz, J.P., Kerbrat P., Goodman,
samples of known antitumor agents that were A., Ducret, J.P., Theodore, C., Speilmann, M., Rouesse,
J.G. and Amiel. J.L. (1985) Phase II trial of ellipticinium in
evaluated for cytotoxic potential. This aspect of
advanced renal cell carcinoma. Cancer Treatmem Reports
the work was supported, in part, by grant CA 69, 901-902.
33047 awarded by the National Cancer Institute. Chabner, B.A. and Myers, C.E. (1982) Clinical pharmacology
JMP is the recipient of a Research Career of cancer chemotherapy. In: V.T. DeVita, Jr., S. Hellman
Development Award from the National Cancer In- and S.A. Rosenberg (Eds.). Cancer. Principles and Prac-
tice of Oncology. J.B. Lippincott Co., Philadelphia, PA, pp.
stitute (1984-1989) and a Fellowship from the
156197.
Alexander von Humboldt Foundation (1990- Chapuis, J.-C., Sordat, B. and Hostettmann. K. (1988) Screen-
199 1). We appreciate the input of our colleagues at ing for cytotoxic activity of plants used in traditional
UIC (Drs. Farnsworth, Kinghorn and Soejarto), at medicine. Journal of Ethnopharmacology 23, 273-284.
RTI (Drs. Wall, Wani and Cobb) and at Glaxo Choi, K., Chen, C.-J., Kriegler, M. and Roninson, I.B. (1988)
An altered pattern of cross-resistance in multidrug-resistant
Group Research, Ltd. (Drs. Harris, O’Neill and
human cells results from spontaneous mutations in the
Tait) in developing some of the ideas expressed mdrl (P-Glycoprotein) Gene. Cell 53, 519-529.
here. Cordell, G.A. (1977) Recent experimental and clinical data
concerning antitumor and cytotoxic agents from plants. In:
Literature cited H. Wagner and P. Wolff (Eds.) Nen* Natural Products and
Plan1 Drugs with Pharmacological, Biological or
Ajani, J.A., Dimery, I., Chawaia, P.J., Pinnameni, K., Ben- Therapeutical Activity. Springer Verlag, Berlin, pp. 55-82.
jamin, R.S., Legha, S.S. and Krakoff, I.H. (1986) Phase 11 Crum, R., Szabo. S. and Folkman, J. (1985) A new class of
studies of homoharringtonine in patients with advanced steroids inhibit angiogenesis in the presence of heparin or a
malignant melanoma, sarcoma, and head and neck, breast heparin fragment. Science 230, 1375-1378.
and colorectal carcinomas. Cancer Research 70. 375-379. Danieli. B. (1971) Recenti acquisizioni nel campo degli an-
Akiyama, S.-l., Cornwell, M.M.. Kuwano, M., Pastan. I. and titumorali di origine vegetale. Fitorerapia 42, 1 l-39,
Gottesman, M.M. (1988) Most drugs that reverse multidrug 94-120.
resistance also inhibit photoafftnity labeling of P- Einzig, A.I., Gralla. R.J.. Leyland-Jones, B.R., Kelsen, D.P..
glycoprotein by a vinblastine analog. Molecular Phar- Cibas, I.. Lewis, E. and Greenberg, E. (1985) Phase I study
macology 33. 144-147. of ellipticinium (2-N-methyl-9-hydroxyellipticinium).
Alley, M.C., Scudiero, D.A., Monks, A.. Hursey, M.L.. Czer- Cancer Investigations 3, 235-24 1.
winski, M.J., Fine, D.L.. Abbott, B.J., Mayo, J.G., Farnsworth, N.R., Akerele, 0.. Bingel, A.S., Soejarto, D.D.
Shoemaker. R.H. and Boyd, M.R. (1988) Feasibility of and Guo, Z. (1985) Medicinal plants in therapy. Bulletin of
drug screening with panels of tumor human cell lines using the World Health Organization 63, 965-981.
a microculture tetrazolium assay. Cancer Research 48. Farnsworth, N.R. and Loub, W.D. (1982) Information gather-
589-60 I. ing and databases that are pertinent to the development of
Barclay, A.S. and Perdue, R.E., Jr. (1976) Distribution of an- plant derived drugs. Workshop Proceedings of
the 0ff;ce of
ticancer activity in higher plants. Cancer’Treafmem Reports Technology Assessment, December, 1982.
60, 1081-I 113. Farnsworth, N.R., Loub, W.D.. Soejarto, D.D.. Cordell, G.A.,
Barrios, V. and Farnsworth, N.R. (1979) The value of plants Quinn, M.L. and Mulholland, K. (1981) Computer services
indicated by traditional medicine in cancer therapy. Pro- for research on plants for fertility regulation. Korean Jour-
ceedings from a Consulration of the Potentials for the Use of nal of Pharmacognosy 12, 98-109.
Planrs Indicated by Traditional Medicine in Cancer Therapy. Folkman. J. (1985) Tumor angiogenesis. Advances in Cancer
WHO. Geneva. pp. 4-6. Research 43, 175-230.
Bellamy, W.T., Dalton, W.S.. Kailey, J.M., Gleason, M.C., Folkman, J. (1986) How is blood vessel growth regulated in
McCloskey, T.M., Dorr. R.T. and Albert& D.S. (1988) normal and neoplastic tissue? - G.H.A. Clowes memorial
Verapamil reversal of doxorubicin resistance in multidrug- lecture. Cancer Research 46. 467-473.
resistant human myeloma cells and association with drug Foon, K.A. (1989) Biological response modifiers: a new im-
accumulation and DNA damage. Cancer Research 48, munotherapy. Cancer Research 49, 1621-1639.
6303-6308. Giovanella. B.C.. Stehlin. J.S.. Wall, M.E.. Wani, M.C.,
Bignon, E., Ogita, K., Kishimoto, A., Gilbert, J., Abecassis, J., Nicholas, A.W., Liu, L.F., Silber, R. and Potmesil, R.
Miquel, J.-F. and Nishizuka, Y. (1989) Modes of inhibition (1989) DNA topoismerase I-targeted chemotherapy of
132

human colon cancer in xenografts. Science 246, I.B. (1985) Inhibition of protein kinase C by tamoxifen.
1046-1048. Cancer Research 45, 2462-2465.
Griffin, P.D. (1988) Plants for fertility regulation. In: E. Dicz- O’Brian, C.A., Liskamp, R.M., Solomon, D.H. and Weinstein,
falusy, P.D. Griffin and J. Khanna (Eds.), Research in LB. (1986) Triphenylethylenes: a new class of protein kinase
Human Reproduction Biennial Report of the WHO Special C inhibitors. Journal of the National Cancer Institute 76,
Program of Research, Development and Research Training in 1243-1248.
Human Reproduction. World Health Organization, Geneva, Perdue, R.E., Jr. and Hartwell, J.L. (1969) The search for plant
pp. 229-242. sources of anticancer drugs. Morris Arboretum Bulletin 20,
Hartwell, J.L. (1967-71) Plants against cancer. A survey. 35-58.
Lloydia 30, 363463; 31, 71-170; 32, 79-107; 32, Pezzuto, J.M., Shieh, H.-L., Shaughnessy, E. and Beattie, C.W.
153-205; 32, 247-296; 33, 98-194; 33, 288-392; 34, (1988) Approaches for drug development in treatment of
103-160; 34, 204-255; 34, 31&361: 34, 386438. advanced melanoma. Seminars in Oncology 15, 578-588.
Hartwell, J.L. (1976) Types of anticancer agents isolated from Prance, G.T. (1977) Floristic inventory of the tropics: where do
plants. Cancer Treatment Reports 60, 1031-1067. we stand? Annals of the Missouri Botanical Garden 64,
Hendrix, M.J.C., Seftor, E.A., Seftor, R.E.B. and Fidler, I.J. 659-684.
(1987) A simple quantitative assay for studying the invasive Rouesse, J.G., Le Chevalier, T., Caille, P., Mondesir, J.M.,
potential of high and low human metastatic variants. Sancho-Garnier, H., May-Levin. F., Speilmann, M., De
Cancer Letters 38, 137-147. Jager, R. and Amiel, J.L. (1985) Phase 11 study of ellip-
Hsu, B. and Yang, J.L. (1984) Hydroxycamptothecin as an an- ticinium in advanced breast cancer. Cancer Treatment
titumor agent. In: H.M. Chang, H.W. Yeung. W.W. Tso Reports 69, 707-708.
and A. KOO (Eds.), Advances in Chinese Medicinal Materials Sartorelli, A.C. (1988) Therapeutic attack of hypoxic cells of
Research. World Scientific Press, Philadelphia, pp. solid tumors: presidential address. Cancer Research 48,
377-389. 775-778.
Jewers, K., Manchanda, A.H. and Rose, H.M. (1973) Scudiero, D.A., Shoemaker, R.H., Paull, K.D., Monks, A.,
Naturally-occurring antitumor agents. Progress in Tierney, S., Nofziger, T.H., Curren, M.J., Seniff, D. and
Medicinal Chemistry 9, l-63. Boyd, M.R. (1988) Evaluation of a soluble tetrazoliumfor-
Legha, S.S., Keating, M., Picket, S., Ajani, J.A., Ewer, M. and mazan assay for cell growth and drug sensitivity in culture
Bodey, G.P. (1984) Phase 1 clinical investigation of using human and other tumor cell lines. Cancer Research
homoharringtonine. Cancer Treatment Reports 68. 48, 48274833.
1085-1091. Shen, D.-W., Fojo, A., Chin, J.E., Roninson, I.B., Richert, N.,
Leigh, E.G.. Jr. (1982) Introduction: why are there so many Pastan, 1. and Gottesman, M.M. (1986) Human multidrug-
kinds of tropical trees? In: E.G. Leigh, Jr.. A.S. Rand and resistant cell lines: increased mdrl expression can precede
D.M. Windsor (Eds.), The Ecology of a Tropical Forest. gene amplification. Science 232, 643645.
Smithsonian Institution, Washington, D.C. Shoemaker, R.H., Monks, A.. Alley, M.C., Scudiero, D.A.,
Li, Y.H, Guo, S.F., Zhou, F.Y., Xu, S.Z. and Zhang, H.L. Fine, D.L., McLemore, T.L., Abbott, B.J., Paull, K.D.,
(1983) Combined harringtonine or homoharringtonine Mayo, J.G. and Boyd, M.R. (1988) Development of human
chemotherapy for acute nonlymphocytic leukemia in 25 tumor cell line panels for use in disease-oriented drug
children. Chung-Hua I Hsueh Tsa Chih (English edn.) 96. screening programs. In: T.C. Hall (Ed.), Prediction of
303-305. Response to Cancer Therapy. Alan R. Liss, Inc.. New York,
Liotta, L.A. (1986) Tumor invasion and metastases - role of N.Y. pp. 265-286.
the extracellular matrix - Rhoads memorial award lecture. Soejarto, D.D. and Farnsworth, N.R. (1989) Tropical rain
Cancer Research 46, l-7. forests: potential source of new drugs? In: Perspectives in
Loub, W.D., Farnsworth, N.R., Soejarto, D.D. and Quinn, Biology and Medicine 32, 2-256.
M.L. (1985) NAPRALERT: Computer handling of natural Spjut, R.W. and Perdue, R.E. (1976) Plant folklore: a tool for
product research data. Journal of Chemical Information and predicting sources of antitumor activity. Cancer Treatment
Computing Sciences 25, 99-103. Reports 60, 979-985.
Muss, H.B., Bundy, B.N., Yazigi, R. and Yordan, E. (1987) Su, H.-D., Mazzei, G.J., Vogler, W.R. and Kuo, J.F. (1985) Ef-
Teniposide in squamous cell carcinoma of the cervix: a fect of tamoxifen, a nonsteroidal antiestrogen, on
phase II trial of the genecological oncology group. Cancer phospholipid/calcium-dependent protein kinase and
Treatment Reports 7 I, 873-874. phosphorylation of its endogenous substrate proteins from
Myers, N. (1984) The Primary Source. W.W. Norton and Co., the rat brain and ovary. Biochemical Pharmacology 34,
New York. 3649-3653.
Neidart, J.A., Young, D.C., Kraut, E., Howenstein, B. and Suffness, M. (1985) The discovery and development of an-
Metz, E.N. (1986) Phase I trial of homoharringtonine ad- titumor drugs from natural products. In: A.J. Vlietink and
ministered by prolonged continuous infusion. Cancer R.A. Dommisse (Eds.), Advances in Medicinal Plant
Research 46, 967-969. Research. Wissenschaftliche Verlagsgesellschaft mbH, Stut-
O’Brian, C.A., Liskamp, R.M., Solomon, D.H. and Weinstein, tgart, pp. IOl--133.

Suffness, M. (1987) New approaches to the discovery of an-
titumor agents. In: K. Hostettmann and P.J. Lea (Eds.),
Biologically Active Natural Products, Annual Proceedings of
the Phyrochemical Society of Europe. Clarendon Press, Ox-
ford, pp. 8S104.
Suffness, M. (1989) Development of antitumor natural pro-
ducts at the National Cancer Institute. Gann Monograph on
Cancer Research 36, 21-44.
Suffness, M. and Douros, J. (1979) Drugs of plant origin. In:
V.T. DeVita and H. Busch (Eds.), Methods in Cancer
Research. Vol. 16, Part A. Academic Press, New York, pp.
73-126.
Suffness, M. and Douros, J. (1982) Current status of the NC1
plant and animal product program. Journal of Natural Pro-
ducts 45, l-14.
Suffness, M. and Peuuto, J.M. (1991) Assays for cytotoxicity
and antitumor activity. In: K. Hostettmann (Ed.), Methods
in P/ant Biochemistry, Vol. 6, London, Academic Press, pp.
71-133.
133
Swanson, SM., Jiang, J.-X., de Souza, N.J. and Pezzuto, J.M.
(1988) A rapid and sensitive bioassay involving cultured rat
glioma ceils to screen for substances capable of elevating in-
tracellular cyclic AMP concentration. Journal of Natural
Products 5 1, 929-936.
Von Hoff, D.D., Casper, J., Bradley, E., Sandvack, J., Jones,
D. and Makuch, R. (1981) Association between human
tumor colony-forming assay results and response of an in-
dividual patient’s tumor to chemotherapy. American Jour-
nal of Medicine 70, lO27- 1032.
Von Hoff, D.D. and Weisenthal, L. (1980) In vitro methods to
predict for patient response to chemotherapy. Advances in
Pharmacology and Chemotherapy 17, 133-l 56.
Wiernik, P.H., Schwartz, E.L., Strauman, J.J., Dutcher. J.P.,
Lipton, R.B. and Paaietta, E. (1987) Phase I clinical and
pharmacokinetic study of taxol-I. Cancer Research 47,
24862493.