-
Vol. 4, 1079 -1086, May 1998 Clinical Cancer Research 1079
Special Lecture
There Are No Bad Anticancer Agents, Only Bad Clinical Trial
Designs-Twenty-first Richard and Hinda Rosenthal
Foundation Award Lecture1
Daniel D. Von Hoff 2,3Institute for Drug Development, Cancer Therapy and ResearchCenter, and Department of Medicine, University of Texas HealthScience Center at San Antonio, San Antonio, Texas 78245
Abstract
Unfortunately, the vast majority (90%) of new antican-
cer agents designed in the laboratory never make it into
routine clinical use. The hypothesis of this lecture is that
many new agents fail in the clinic because the appropriate
clinical trial(s) that could exploit the attributes of the new
agent are not performed. An appreciation that both bench
and clinical investigations are difficult endeavors should aid
in improving clinical trial designs and give the best chance
for new agents to be added to our therapeutic armamen-
tarium.
IntroductionA very common story is that a basic scientist works for
years and years to develop a new agent with a unique mecha-
nism of action to treat tumor cells; but that new agent never
makes it into clinical trials, or if it does, it almost always fails to
make an impact on the disease.
The hypothesis of today’s lecture is that many new agents
fail because the appropriate clinical trials to exploit the at-
tributes of the new agent have not been performed. This is
because of a basic uncoupling or disconnection between the
bench scientist, who has worked long and hard on his or her
hypothesis but who may not understand the difficulties of and
creativity required for clinical trials, and the clinical scientist,
who may not understand how difficult the basic discovery has
been and how careful one must be before abandoning the new
concept or new agent from the bench.
The magnitude of the disconnection between the bench
scientist and the bedside clinician cannot be assessed solely by
how many new agents don’t make it into general use in the
oncology community. The percentage of agents, however, that
do go to clinical trial and are adopted for clinical use is disap-
Received 2/3/98; accepted 3/4/98.1 Presented at the 88th Annual Meeting of the American Association for
Cancer Research, April 15, 1997, San Diego. CA.
2 To whom requests for reprints should be addressed, at Institute forDrug Development, Cancer Therapy and Research Center, 14960Omicron Drive, San Antonio, TX 78245. Phone: (210) 677-3880; Fax:(2 10) 677-0058; E-mail: [email protected] Dr. Von Hoff is a founder and minority owner of ILEX Oncology,
which is working on clinical trials with MGBG and DFMO. He is alsoa member of the Scientific Advisory Board for Eli Lilly, which owns
and markets gemcitabine.
pointingly low. Fig. 1 details the number of new agents brought
into clinical trials from 1975 to 1994 and the number of those
agents eventually approved for clinical use. Overall, there were
280 new agents brought into Phase I (dose-finding) clinical
trials in patients. Only 29 of them (10%) were eventually ap-
proved for use in patients.
There are three main reasons that emerge in examining
why the other 90% of new agents don’t make it into routine use.
These include: (a) the toxicities of the agent were too great; (b)
there was a lack of efficacy; and (c) no attention was paid to the
mechanism of action of the compound when the clinical trials
were designed and conducted.
In the discussion below, we will explore all three of these
major reasons why new agents don’t make it into general
clinical use. In addition, examples will be given of how to
bypass these problems so that a greater percentage of agents
brought into clinical trials will actually make it to general use.
Why Compounds Don’t Make It-ToxicitiesTable 1 gives examples of severe toxicities that halted the
development of several new anticancer agents. These toxicities
basically stopped the development of the new agents in their
tracks, in most instances, with a complete suspension of their
clinical trials. However, because of the clinical activity of the
agent that had been seen early on, e.g. , responses to the agent
paclitaxel in patients with a variety of malignancies, clinicians
persisted in trying to find a way to circumvent the toxicity of
paclitaxel. The thesis is that one can circumvent toxicities of the
new agent if the drug has clinical antitumor activity. Table 1
also details the methods that were used to circumvent toxicities
for each of the agents.
The observation of our drug development team in San
Antonio, TX, is that in dealing with the phenomenon of dose-
limiting toxicities, stopping the drug and then the clinician
finding a way to circumvent those toxicities, we have devised
the life cycle of a new anticancer drug as shown in Fig. 2 (1). As
can be seen in the figure, a drug is discovered and brought into
clinical testing. Almost inevitably, a crisis in the development of
the agent occurs in which either a severe toxicity is noted with
the agent or no antitumor activity is noted with the agent. The
agent then undergoes programmed drug death, which we like to
call “pharmacoptosis.” Fortunately, methods can sometimes be
developed to circumvent the toxicity of the new agent, and the
agent then emerges from the state of pharmacoptosis and is
rediscovered or reborn. As is noted in Table 1, this ability to
reverse pharmacoptosis is a very common phenomenon.
An Example of Recovery from Pharmacoptosis Because
of Toxicity-Mitoguazone. A recent example of recovery
from pharmacoptosis is the redevelopment of mitoguazone.
Mitoguazone was synthesized in 1898 by Thiele et a!. (2). It also
has the names MGBG, methylglyoxal bis(guanylhydrazone),
Research. on June 13, 2021. © 1998 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
http://clincancerres.aacrjournals.org/
-
40
.nC.)
� 30rI)
.�
I-
5)
75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94
Year
1080 There Are No Bad Anticancer Drugs
Fig. 1 Total number of differ-ent agents in Phase I trials per
year (#{149}),and number of thoseagents eventually approved (El).
methyl-GAG, and NSC-32946. The structure of mitoguazone is
shown in Fig. 3. One of the intriguing aspects of mitoguazone is
that it has a unique mechanism of action as an inhibitor of
S-adenosylmethionine decarboxylase (3, 4). Some investigators
have believed that it is also a mitochondrial interactive agent (5).
It should be remembered that compounds with new and uniquemechanisms of action have a better chance for non-cross-resis-
tance with the antineoplastics already in clinical use.
Mitoguazone was first given to patients in the early l960s,
usually on a chronic (e.g., daily X 30 days) schedule (6, 7).
Antitumor activity was noted in patients with leukemia and
lymphoma (6- 8). However, the major toxicity of mucositis (and
other severe toxicities) caused all clinical trials with the agent tobe stopped. Therefore, despite clinical activity, all clinical trials
with mitoguazone were discontinued (the drug underwent phar-
macoptosis).
The key to the rebirth of mitoguazone was new scientificknowledge from pharmacokinetic studies performed in the late
1970s and early l980s (and reconfirmed recently) that mitogua-
zone had an extremely long plasma half-life of >5 days (9-1 1).With this new understanding, it was clear that the schedule of
daily administration of mitoguazone was causing an accumula-
tion of mitoguazone that could have been responsible for the
side-effect profile (severe mucositis and myelosuppression sec-
ondary to toxicity on rapidly turning over cells in the mucous
membranes). Based on this new knowledge, weekly and every
other week schedules of administration were devised (12, 13).
Using these dosing schedules, mucositis became an unusual side
effect with the agent, and antitumor activity was noted in pa-
tients with Hodgkin’s disease, non-Hodgkin’s lymphoma, head
and neck cancer, endometrial cancer, prostate cancer, and
Table 1 Agents in which severe toxicities halted their developmentand methods used to circumvent those toxicities (and reverse
pharmacoptosis)
Agent Toxicity
Method tocircumvent
toxicity
Doxorubicin/ Congestive heart Dose limitation,Daunomycin failure dexrazoxane
Vincristine Neurotoxicity Dose limitationCisplatin Renal toxicity,
ototoxicity emesisHydration,
antiemeticsPaclitaxel Anaphylactic shock Steroids and other
premedicationDocetaxel Fluid retention SteroidsCPT-l 1 Diarrhea AntidiarrhealsFludarabine Blindness Dose limitation
phosphate
esophageal cancer (12-16). Despite that activity, there was still
little interest in mitoguazone until the emergence of non-Hodgkin’s lymphoma in patients with AIDS.
AIDS-related non-Hodgkin’s lymphoma was first reportedin 1982. It is usually a disease with aggressive histologies
(large-cell immunoblastic and small-cell non-cleaved), has a
high incidence of central nervous system involvement, accounts
for 12-16% of all AIDS-related deaths, and there is no standard
treatment for patients who have progressed on first-line therapy
(17, 18). The average survival for patients who have progressed
on first-line therapy is 63 days (19). Polyamines are elevated in
patients with lymphoma 4.0-5.3-fold compared with normal
volunteers (20, 21); therefore, an agent like mitoguazone, which
Research. on June 13, 2021. © 1998 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
http://clincancerres.aacrjournals.org/
-
Discovery(Birth)
Rediscovery(Rebirth)
Fig. 3 Structure of mitoguazone.
Fig. 2 The life cycle of a new anticancer drug (reprinted with permis-
sion from Annals of Oncology).
100w
�g 750-�:
250-
25 30 35 40 45
Fig. 4 Antitumor activity of the hypothetical new agent difungomuc-
tane ( #{149}) in an in vivo tumor growth system compared with controlgrowth (0) of the tumor.
Clinical Cancer Research 1081
Clinicaltesting
�ll�p�s� Crisis 1’� I Ssv#{149}retoxicity
‘ No activity
Programmeddrug death
(pharmacoptosis)
inhibits the synthesis of polyamines, was an attractive agent to
try against the disease. In addition, mitoguazone was interesting
to try because: (a) the agent already had documented activity in
patients with non-AIDS-related non-Hodgkin’s lymphoma (12,
13); (b) the agent causes minimal myelosuppression and other
toxicities on the new schedule (12, 13); and (c) mitoguazone
crosses the blood brain barrier, as evidenced by high concen-
trations in brain tumors (22).
Recently, Levine et a!. (23) reported the results of a Phase
II trial of mitoguazone given at a dose of 600 mg/m2 on days I,8, and then every 2 weeks, in patients with AIDS-related lym-phoma. In that study, she treated 35 patients with AIDS-related
lymphoma, all of whom had failed on prior regimen and 17 of
whom had failed 2-6 prior regimens. Twenty-four (80%) of the
patients had a CD4 count of < lOO/dl. There were three patients
who attained a complete response, plus three who attained a
partial response for an overall response rate of 6 of 35 or 17%
(95% confidence interval, 4.7-29.6%; Ref. 23). A second study
in a similar population has confirmed that response rate (24).
The toxicities associated with the use of mitoguazone on the
days 1 , 8, and then every 2 weeks schedule were minimal, with
only an 8% incidence of grades 3 and 4 mucositis. The lesson
learned from the further development of mitoguazone is that
new knowledge regarding the basis for toxicity with a corn-
pound (in this case, its clinical pharmacology) can lead to adifferent way of using that compound that avoids the toxicities
and allows its antitumor activities to be exploited.
After the presentation of this lectureship, the Phase II dataon the activity of mitoguazone in patients with AIDS-related
non-Hodgkin’s lymphorna was presented to the Food and Drug
Administration Advisory Committee, and the Committee rec-
ommended against its approval based only on Phase II data. Arandomized Phase HI trial to further support its efficacy wasrecommended by the Food and Drug Administration, and plans
are being made to conduct that study. The drug has not yet
emerged from pharmacoptosis.
Why Compounds Don’t Make It-Lack of Efficacy
A second reason new agents don’t make it into routine
clinical use is no efficacy (or a supposed lack of efficacy). Many
NH2
\/\ /\H2N\,NV N II N NH2 2HCI . H20
H2N
0 5 10 15 20
Days
investigators would argue that the reason we have so many
compounds that make it into clinical trials that have poor to no
activity in the clinic is the poor (nonpredictive) preclinical
systems we have for selecting those agents (25, 26). Our thesis
in San Antonio is that in our clinical trial designs, we are asking
the drug to do more clinically than it did preclinically. For
example, take the hypothetical new drug “difungomuctane” (nota real drug), which inhibits tumor growth to the degree noted in
Fig. 4. As can be seen in Fig. 4, difungomuctane clearly causes
tumor growth delay. However, can we really expect this drug to
shrink tumors in patients when all it does in animal systems is
slow down tumor growth? Again, the theory is that with our
current clinical trial designs, we are asking our drugs to do more
clinically than they do in preclinical systems. Perhaps new
clinical trial end points besides shrinkage of patients’ tumors are
needed to assess the antitumor activity of some new agents.
An Example of Recovery from Pharmacoptosis Becauseof Supposed Lack of Antitumor Activity. Gemcitabine (LY
18801 1, difluorodeoxycytidine; dFdC) is a nucleoside with the
structure shown in Fig. 5. The compound was synthesized by
Hertel et a!. (27) with tumor growth delay documented in
several animal models and human tumor xenograft models
including X5563 myeloma; 6C3HED lymphosarcoma; Ll 210,
P388, and P1543J leukemia; and the CX-l and LX-1 xenografts
(28 -30). In addition, the agent had a broad spectrum of antitu-
mor activity against human tumor colony-forming units taken
directly from patients and growing in soft agar (31). Activity
was particularly impressive against breast, non-small cell lung,
ovarian, and pancreatic cancer colony-forming units.
Phase I trials were conducted with gemcitabine beginning
Research. on June 13, 2021. © 1998 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
http://clincancerres.aacrjournals.org/
-
< Gemcitabine(1000 mg/m2/week)
5.Fluorouracil(600 mg/m2/week)
F
Fig. 5 Structure of gemcitabine.
1082 There Are No Bad Anticancer Drugs
‘ The abbreviations used are: ‘fl’P, time to tumor progression; DFMO,difluoromethylornithine (eflornithine); ODC, ornithine decarboxylase.
in 1987 (32, 33). The daily X 5, 30-mm infusion schedule gave
a maximally tolerated dose of only 9 mg/rn2, with hypertension
and flu-like symptoms being the dose-limiting toxicities. On the
every 2 weeks schedule, over 0.5-4 h, the maximally tolerated
dose was 3600 mg/rn2. Dose-limiting toxicities included neu-
tropenia and flu-like symptoms. In that study, we noted one
patient with pancreatic cancer with clinical improvements in-
cluding less pain, an improved performance status, and weight
gain. Based on that observation as well as observations from
other groups, a Phase II trial of gemcitabine was conducted by
Casper et a!. (34, 35). In that study, 43 patients with advanced
pancreatic cancer with no prior chemotherapy received doses of
gemcitabine ranging from 800-1250 mg/m2 weekly X 3 re-
peated every 28 days. Five partial responses were noted for an
overall response rate of 13%. Toxicities were mild. What was
striking was that, despite the overall low incidence of partial
response (only 13%), there was a far greater percentage of
patients who reported improvements in their pancreatic cancer-
related symptoms, i.e., pain, weakness, and weight loss. Because
virtually no chemotherapy has been shown to impact the symp-
torns of pancreatic cancer, it was decided to design a clinical
trial with gemcitabine to determine, in a controlled setting,
whether that agent decreased the symptoms associated with
advanced pancreatic cancer. This thinking was a departure from
the usual oncology clinical trial designs, where typical end
points are either tumor shrinkage (complete or partial responses)
or survival. However, in other areas of drug development, other
end points are always used in clinical trials. For example, the
end point for a clinical trial for a new antiarrhythmic agent is
control of arrhythmias, not survival. The end point for the
clinical trials of a new antihypertensive is control of blood
pressure, not survival. New end points in clinical trials in
patients with advanced cancer may also be in order. One of
those end points could be whether the new therapy improves the
Patients with advanced,symptomatic pancreaticcancer
- No prior chemotherapy
- Pain, intensity score of � 20
(out of 100 on MPAC*). Analgesic consumption
� 10 morphine sulfateequivalents
. Karnofsky PerformanceStatus � SO but � 80
*MPAC = Memorial Pain Assessment Card
Fig. 6 Design for definitive trial of gemcitabine in patients with ad-
vanced symptomatic pancreatic cancer.
tumor-induced problems that bother the patient. In the case of
pancreatic cancer, the things that bother the patient include pain,
weakness, and a declining performance status and weight loss
(36, 37). If a new treatment could improve those symptoms for
a patient, we postulated that the new treatment would give them
clinical benefit and should be made available to them.
To determine whether gemcitabine gave clinical benefit to
patients with advanced pancreatic cancer, the clinical trial out-
lined in Fig. 6 was designed. As can be seen in that figure, to be
eligible for the study, patients had to have advanced pancreatic
cancer which was symptomatic (e.g., patients with pain of >20
of 100 on the Memorial Pain Assessment Card (38); patients
required more than 10 mg of morphine sulfate or equivalent to
control their pain; and patients had to have a Karnofsky Per-
formance Status of >50 but
-
Table 2 Summary of results of randomized trial of gemcitabineversus 5-fluorouracil (5-FU) for treatment of patients with advanced
pancreatic cancer”
Fig. 7 Preclinical-clinical uncoupling in drug development.
Clinical Cancer Research 1083
Parameter 5-FU arm Gemcitabine arm P
Clinical benefit 4.8% 23.8% 0.0022Response rate 0% 5.4% 0.077TTP (mo) 1.0 3.2 0.0002Survival
Median (mo) 4.41 5.65 0.0025One year 2%
“Bums et al. (42).
18%
had clinical benefit versus 5% of the patients who received
5-fluorouracil (P 0.0022). This has now been reported in
other trials (43, 44). The TTP (which corresponds best to the
tumor growth delay in animals) was also significantly better. Of
additional note and importance is that median survival and
1-year survival were significantly better for the gemcitabine-
treated patients. This finding helped validate the use of our
clinical benefit parameter as a clinically meaningful end point.
Also importantly, the side-effects profile of gemcitabine was not
significantly different from the profile of 5-fluorouracil, except
that more transfusions were required in the patients receiving
gemcitabine, and neutropenia was also more common. How-
ever, the gemcitabine patients also lived longer to obtain those
transfusions or to develop neutropenia.
The gemcitabine example is provided to document that if a
new agent produces tumor growth delay in a preclinical model,
we should strive to use a tumor growth delay end point (e.g.,
‘FTP) in our clinical trials. Above all, we should strive to match
our clinical trial end points to what has been found in preclinical
models. Only then will there be a true and fair clinical test for
that new agent. This will be particularly important as our new
molecular biology gives us new agents that are growth-modu-
lating agents rather than cytotoxic agents that shrink established
tumors (see below).
Why Compounds Don’t Make It-Attention Is Not
Paid to the Mechanisms of Action of the Compound
When attention is not paid to the mechanism of action of a
new compound, our clinical trials often do not exploit that
mechanism of action, which usually leads to failure of the
compound in the clinic. As noted in Fig. 7, I refer to this as
“preclinical-clinical uncoupling.” Table 3 gives some examples
of preclinical-clinical uncoupling. As can be seen in that table,
regardless of the mechanism of action of the agent, the only
parameter we have used to determine whether to continue de-
velopment of a new agent has been to assess whether the
compound causes complete tumor shrinkage (100% disappear-
ance of the tumor lasting at least 1 month) or a partial response
(�50% disappearance of tumor lasting at least 1 month). This
nearly total neglect of the mechanism of action probably has led
to discontinuing the development of many new agents that, had
their mechanism of action been remembered and our trials
designed to measure that mechanism, may have been successful
in the clinic.
An Example of an Agent Recovering from Pharmacop-tosis Because of Attention Paid to Its Mechanism of Action.
DFMO has the structure shown in Fig. 8. The compound has a
unique mechanism of action, i.e., the irreversible inhibition of
ODC (45, 46). ODC is becoming a more interesting target in
both the areas of cancer treatment and in cancer etiology and,
hence, in cancer prevention. More specifically, ODC has been
found to be up-regulated in the tissues of several premalignant
conditions, including cervical intraepithelial neoplasia (47), co-
lon polyps (48), and others (49). Manni et a!. (50) have found
that ODC is up-regulated in patient breast cancer tissue, and
elevation of the expression of ODC in that tissue is of greater
significance for a poor prognosis than is the estrogen receptor
status or the number of positive lymph nodes. Finally, of great
interest, is that Auvinen et al. (51) have reported that ODC is an
oncogene. Transfection of the ODC gene into NIH 3T3 cells
causes transformation of the cells.
As noted above, DFMO is an irreversible inhibitor of ODC.
DFMO blocks the promotion step of carcinogenesis in several
animal models including breast, colon, bladder, skin, and pros-
tate models (52-55). Unfortunately, the initial clinical trials with
DFMO developed the agent as a standard anticancer agent with
escalation to the maximally tolerated dose of the agent. Toxic-
ities noted with doses of �20 g/m2/day of DFMO included
thrombocytopenia, nausea and vomiting, leukopenia, anemia,
diarrhea, high frequency hearing loss, and in some cases, met-
abolic acidosis (56, 57). After those studies, DFMO was aban-
doned (underwent pharmacoptosis) by most investigators.
A breakthrough in the use of DFMO was a study by Love
et al. (58) who performed a clinical trial to attempt to exploit the
mechanism of action of DFMO, the inhibition of ornithine
decarboxylase. They conducted a study to determine the small-
est dose of DFMO that could be used to inhibit ornithine
decarboxylase by carefully performing skin biopsies on the
patients. They found that a dose of 0.5 g/m2/day (one-twentieth
the daily dose used in prior cytotoxic approach studies) inhibited
ornithine decarboxylase. Also, importantly, they noted that no
toxicity was seen at that dose level. This finding fostered addi-
tional clinical trials because DFMO could now be used as a
chemoprevention agent.
The field of chemoprevention has changed substantially
over the past few years because there are now histological
changes that are regarded as premalignant. These premalignant
changes include cervical intraepithelial neoplasia (CIN III),
actimc keratosis, colon adenoma, ductal carcinoma in situ of the
breast, previously resected superficial bladder cancer, atypical
squamous metaplasia/dysplasia in the lung, oral dysplastic leu-
koplakia, and prostatic intraepithelial neoplasia. The goal in
chemoprevention now is to reverse those premalignant changes
Research. on June 13, 2021. © 1998 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
http://clincancerres.aacrjournals.org/
-
CHF2
H2N - CH2 - CH2 - CH2- C - COOH
NH2
Fig. 8 Structure of DFMO.
Period A
I (Standard �Therapy)
Period B
(New Agent) �1
- Ratio Period B = Growth Modulation IndexPeriod A-Anything > 1.33 (33% improvement) is
considered excellent (and is unexpected forsecond-line therapy)
Fig. 9 A method for early determination as to whether a new agent ishaving a modulating effect on tumor growth.
/7� Standard agent + Placebo
Standard agent + new agent
1084 There Are No Bad Anticancer Drugs
with end points of ‘FTP and 1-year survival.
Table 3 Examples of preclinical-clinical uncoupling
What is seen preclinically What we are measuring clinically
1. The agent inhibits angiogenesis Partial or complete tumor shrinkage
2.
3
The agent prevents metastases
The agent induces apoptosis
Partial or complete tumor shrinkage
Partial or complete tumor shrinkage
(use them as an intermediate marker for an effect of the agent
being studied; Ref. 59).
To date, three clinical trials have correctly used low doses
of DFMO to inhibit ODC. Mitchell et a!. (47) have recently
published their work demonstrating that DFMO could substan-
tially reduce (in 52% of patients) cervical intraepithelial neo-
plasia (CIN-Ill, a substantial risk factor for the development of
cervical cancer) after only one month of use. Meyskens et al.
(60) have clearly documented that low doses of DFMO can
suppress polyamine content in colorectal biopsy specimens as
preparation for their randomized trial of a nonsteroidal, anti-
inflammatory agent plus DFMO versus the nonsteroidal agent
alone in patients with a history of recurrent colorectal polyps.
The end point for this study will be the number of polyps that
recur per unit of time.
Last, but not least, is a trial published by Alberts et a!. (61),
who randomized 44 patients with actinic keratoses (a premalig-
nant skin condition) to receive either DFMO cream or a placebo
cream. This randomized, double-blinded trial showed a signif-
icant decrease in actinic keratoses with three months of use of
the DFMO cream, whereas there was no decrease with the
placebo cream (61). There was a reduced spermidine:spermine
ratio in the skin biopsies from the patients receiving DFMO,
indicating that DFMO was inhibiting the ODC. This is a proof
of principle that inhibition of the ODC by DFMO resulted in a
clinically measurable decrease in the number of actinic kerato-
ses after only a short period of application.
How Can We Measure a Clinical Effect of a New
Compound That Is Not a Cytotoxic Agent?
The developments in molecular biology and the identi-
fication of new targets provide challenges to both the bench
and the clinical investigator (62). There are new antisense
molecules, farnesyl transferase inhibitors, telomerase inhib-
itors, kinase inhibitors, angiogenesis inhibitors, and many
other molecules that are likely to be growth-modulating and
that will not cause immediate tumor shrinkage. One way to
test a growth-modulating agent is to conduct a randomized
trial, as was done for gemcitabine, with end points of time-
to-tumor progression and 1-year survival. However, is there
a way we can tell early in its development whether an agent
has any promise (before conducting the randomized Phase III
trial)? One way that might be used to obtain an early lead as
to whether a growth-modulating agent is having an effect is
outlined in Fig. 9. Our hypothesis is that if the new agent has
an antitumor effect, it will change the natural history of the
disease. Therefore, we will use the patient as his/her own
control. As can be seen in Fig. 9, period B is the TTP on the
new agent. Period A is the TTP on the treatment the patients
received just before the new agent was started. Given the
natural history of cancer, it would be shorter that the TTP on
the new agent would be shorter than the TTP on a prior
regimen. Therefore, if the TTP on the new agent (period B)
is greater than the TTP on the prior therapy (period A) it is
likely the new agent is having an effect on the natural history
of that patient’s tumor. In Fig. 9, we suggested a 33%
improvement; however, that value is arbitrary. Although not
yet tested prospectively, the use of a patient as his/her control
may be a way to determine whether a growth-modulating
agent is having a clinical effect before the randomized Phase
III trial is launched.
Summary of Lessons Learned
What lessons have we learned to increase the percentage of
our agents that make it in the clinic? The first lesson is to avoid
pharmacoptosis (programmed drug death). If the new agent
shows toxicities in the clinic but has hints of antitumor activity,
be persistent and find a way to circumvent that toxicity.
The second lesson is for the correct clinical trial design to
mimic the preclinical activity of the agent. For a growth-mod-
ulatory agent, a suggested clinical trial design would be:
Research. on June 13, 2021. © 1998 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
http://clincancerres.aacrjournals.org/
-
Clinical Cancer Research 1085
The third lesson is to take advantage of the mechanism of
action of the new agent when designing a clinical trial. For an
angiogenesis inhibitor, use a minimal residual disease model
such as:
. . Angiogenesis inhibitor
put patient in bestremission (minimal
residual disease) Placebo
with end points of ‘FTP or recurrence-free survival.
The fourth lesson (for the preclinical scientist) is to demand
better trial designs for your molecule! The fifth lesson for my
clinical colleagues and myself is to take it upon ourselves to
personally give new agents the best mechanism-of-action-based
trial design we possibly can.
Conclusion
In conclusion, I would like to thank all of the patients,
mentors, and colleagues who have made this work possible. San
Antonio is the birthplace of Texas freedom, and we always say,
“Remember the Alamo!” From this lectureship, I would like you
to remember that both bench investigations and clinical inves-
tigations are difficult endeavors and that we need to appreciate
both of them as important sciences. Remembering that will keep
our clinical trial designs optimal and provide the best clinical
development of a new idea or a new agent from the bench.
Acknowledgments
I gratefully acknowledge the expert assistance of Laida Garcia,Meredith Sterling, and Judy Turner in preparing the manuscript for
publication.
References
1. Von Hoff, D. D. MGBG: teaching an old drug new tricks. Ann.
Oncol., 5: 487-493, 1994.
2. Thiele, J., and Dralle, F. Zur Kenntnis des amino gunidins. I. Con-densations-produkte de amidoguanidins mit aldehyden and ketonen derfettreihe. Justus Liebigs Ann. Chem., 302: 275-299, 1898.
3. Corti, A., Dave, C., Williamson-Ashman, H. G., Mihich, E., and
Schenone, A. Specific inhibition of the enzymatic decarboxylation ofS-adenosylmethionine by methylglyoxal bis(guanylhydrazone) and re-lated substances. Biochem. J., 139: 351-357, 1974.
4. Pleshkewych, A., Kramer, D. L., Kelly, E., and Porter, C. W. Inde-
pendence of drug action on mitochondria and polyamines in L1210
leukemia cells treated with methylglyoxal bis(guanylhydrazone). Can-
cer Res., 40: 4533-4540, 1980.
5. Pine, M. J., and DiPaolo, J. The antimitochondrial actions of2-chloro-4’,4”-bis(2-imidazolin 2-yl)-terephthalanilide and methyl-
glyoxal bis(guanylhydrazone). Cancer Res., 26: 18-25, 1966.
6. Freireich, E., Frei, E., III, and Karon, M. Methylglyoxal bis(guanyl-hydrazone): a new agent active against acute myelocytic leukemia.Cancer Chemother. Rep., 16: 183-186, 1962.
7. Regelson, W., and Holland, J. F. Clinical experience with methyl-
glyoxal bis(guanylhydrazone) dihydrochloride: a new agent with clini-cal activity in acute myelocytic leukemia and the lymphomas. CancerChemother. Rep., 27: 15-26, 1963.
8. Levin, R. H., Henderson, E., Karon, M., and Freireich, E. J. Treat-
ment of acute leukemia with methylglyoxal bis(guanylhydrazone)
(methyl-GAG). Clin. Pharmacol. Ther., 6: 31-42, 1964.
9. Marsh, K. C., Liesmann, J., Patton, T. F., Fabian, C. J., and Stemson,L. A. Plasma levels and urinary excretion of methyl-GAG following iv.infusion in man. Cancer Treat. Rep., 65: 253-257, 1981.
10. Hart, R. D., Roboz, J., Wu, K., Bruckner, T., Ohnuma, T., andHolland, J. F. Clinical and pharmacologic studies with weekly andbiweekly methylglyoxal bis-guanylhydrazone (methyl-G). Proc. Am.
Assoc. Cancer Res., 21: 181, 1980.
11. Rizzo, J., Levine, A. M., Weiss, G. R., Pearce, T., Kraynak, M.,
Mueck, R., Smith, S., Von Hoff, D. D., and Kuhn, J. G. Pharmacoki-netics profile of mitoguazone (MGBG) in patients with AIDS-related
non-Hodgkin’s lymphoma. Invest. New Drugs, 14: 227-234, 1996.
12. Knight, W. A., III, Fabian, C., Costanzi, J. J., Jones, S. E., and
Coltman, C. A., Jr. Methylglyoxal bis-guanylhydrazone (methyl GAG,MGBG) in lymphoma and Hodgkin’s disease. Invest. New Drugs, 1:
225-237, 1983.
13. Warrell, R. P., Lee, B. J., Kempin, S. J., Lacher, M. J., Straus, D. J.,
and Young, C. W. Effectiveness of methyl-GAG (methylglyoxal-bis-[guanylhydrazone]) in patients with advanced malignant lymphoma.Blood,57: 1011-1014, 1981.
14. Warrell, R. P., and Burchenal, J. H. Methylglyoxal- bis(guanylhy-
drazone) (methyl GAG): current studies and future prospects. J. Clin.
Oncol., 1: 52-65, 1983.
15. Luedke, D. W., Maddox, W., Birch, R., Velez-Garcia, E., andSchleuter, J. Mitoguazone in advanced squamous cell carcinoma of head
and neck origin: a Phase II trial of the Southeastern Cancer Study Group.Cancer Treat. Rep., 70: 529-530, 1986.
16. Scher, H. I., Yagoda, A., Ahmed, T., and Watson, R. C. Methyl-glyoxal-bis(guanylhydrazone) in hormone-resistant adenocarcinoma of
the prostate. J. Clin. Oncol., 3: 224-228, 1985.
17. Karp, J. E., and Broders, S. Acquired immunodeficiency syndrome
and non-Hodgkin’s lymphoma. Cancer Res., 51: 4743-4756, 1991.
18. Levine, A. Acquired immunodeficiency syndrome related lym-
phoma. Blood, 80: 8-20, 1992.
19. Kaplan, L. D., Straus, D. J., Testa, M. A., Van Roenn, J., Dezube,B. J., Cooley, T. P., Hemdier, B., Northfelt, D. W., Huang, J., Tulpule,
A., and Levine, A. M. Low-dose compared with standard-dosemBACOD chemotherapy for non-Hodgkin’s lymphoma associated with
human immunodeficiency virus infection. N. Engl. J. Med., 336: 1641-1648, 1997.
20. Russell, D. H., Levy, C. C., Schimpft, S. C., and Hawk, I. A.
Urinary polyamine in cancer patients. Cancer Res., 31: 1555-1558,
1971.
21. Russell, D. H. Increased polyamine concentrations in the urine of
human cancer patients. Nature (Lond.), 233: 144-145, 1971.
22. Rosenblum, M. G., Stewart, D. J., Yap, B. S., Leavens, M., Ben-
jamin, R. S., and Loo, T. L. Penetration of methylglyoxal bis(guanyl-hydrazone) into intracerebral tumors in humans. Cancer Res., 41: 459-
462, 1981.
23. Levine, A. M., Tulpule, A., Tessman, D., Kaplan, L., Giles, F.,
Luskey, B. D., Scadden, D. T., Northfelt, D., Silverberg, I., Wernz, J.,Espina, B., and Von Hoff, D. D. Mitoguazone therapy in patients with
refractory or relapsed AIDS-related lymphoma: results from a multi-
cancer Phase II trial. J. Clin. Oncol., 15: 1094-1 103, 1997.
24. Levine, A. M., Giles, F., Kaplan, L, Sparano, J., Scadden, D.,
Luskey, B., Northfelt, D., Thommas, J., Anderson, L. J. J., Espina, B.,Tessman, D., and Von Hoff, D. Two sequential prospective, multi-institutional Phase II clinical trials of mitoguazone (MGBG) in refrac-tory or relapsed AIDS-related lymphoma. Blood, 86 (Suppl 1): 381a,1995.
25. Staquet, M. J., Byar, D. P., Green, S. P., and Rozenwey, M. Clinicalproductivity of transplantable tumor system in the selection of new
drugs for solid tumors: rationale for a three-stage strategy. Cancer Treat.
Rep., 67: 753-765, 1983.
26. Twentyman, P. Drug screening in xenografts: the European multi-
center project. Ann. Oncol., 5: 394-396, 1994.
27. Hertel, L. W., Boder, G. B., Kroin, J. S., Rinzei, S. M., Poore, G. A.,
Todd, G. C., and Grindey, G. B. Evaluation of the antitumor activity ofgemcitabine (2’,2’-difluoro-2’-deoxycytidine). Cancer Res., 50: 4417-4422, 1990.
Research. on June 13, 2021. © 1998 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
http://clincancerres.aacrjournals.org/
-
1086 There Are No Bad Anticancer Drugs
28. Heinemann, V., Hertel, L., Grindey, G. B., and Plunkett. W. Com-
parison of the cellular pharmacokinetics and toxicity of 2’,2’-difluoro-deoxycytidine and 1 .43-D-arabinofuranosyl cytosine. Cancer Res., 48:
4024-4031, 1988.
29. Ghandi, V., and Piunkett, W. Modulatory activity of 2’,2’-difluoro-
deoxycytidine on the phosphorylation and cytotoxicity of arabinosyl
nucleosides. Cancer Res., 50: 3675-3680, 1990.
30. Huang. P., Chubb, S., Hertel, L. W., Gnndey, G. B., and Plunkett,W. Action of 2’,2’-difluorodeoxycytidine on DNA synthesis. Cancer
Res.,5/:6l10-6l17, 1991.
31. Hanauske, A., Degen, D., Marshall, M. H., Hilsenbeck, S. G.,
Grindey, G. B., and Von Hoff, D. D. Activity of 2’,2’-difluorodeoxy-
cytidine (gemcitabine) against human tumor colony forming units. An-
ticancer Drugs, 3: 143-146, 1992.
32. O’Rourke, T. J,. Brown, T. D., Havlin, K., Kuhn, J. G., Craig, J. B.,
Burns, H. A., Satterlee, P. G., Tarassoff, F., and Von Hoff, D. D. Phase
I clinical trial of gemcitabine given as an intravenous bolus on 5
consecutive days. Eur. J. Cancer, 30A: 417-418, 1994.
33. Brown, T., O’Rourke, T., Bums, H., Kuhn, J., Tarassoff, P., Cag-
nola, J., Rodriguez, G., and Von Hoff, D. A Phase I trial of gemcitabine
(LY18801 1) administered intravenously every two weeks. Proc. Am.Soc. Clin. Oncol. Annu. Meet., 10: 115, 1991.
34. Casper, E. S., Green, M. R., Brown, T. D., Keisen, D. P., Kresek, T.,
Trochanowski, B., and Tarassoff, P. G. Phase II trial of gemcitabine
(2’.2’-difluorodeoxycytidine) in patients with pancreatic cancer. Proc.
Am. Soc. Clin. Oncol. Annu. Meet., 10: 143, 1991.
35. Casper, E. S., Green, M. R., Kelsen, D. P., Heelan, R. T., Brown,T. D., Flombaum, C. D., Trochanowski, B., and Tarassoff, P. G. PhaseII trial of gemcitabine (2’,2’-difluorodeoxycytidine) in patients with
adenocarcinoma of the pancreas. Invest. New Drugs, 12: 29-34, 1994.
36. Warshaw, A. L., and Fernandez-del Castillo, C. Pancreatic carci-
noma. N. Engi. J. Med., 326: 455-465, 1992.
37. Kelsen, D. P., Portenoy, R. K., Thaler, H. T., Niedzwiecki, D.,
Passik, S. D., Tao, Y., Banks, W., Brennan, M. F., and Foley, F. M. Pain
and depression in patients with newly diagnosed pancreas cancer.J. Clin. Oncol.. 13: 748-755, 1995.
38. Fishman, B., Pastemak, S., and Wallenstein, S. L. The Memorial
Pain Assessment Card: a valid instrument for the evaluation of cancerpain. Cancer (Phila.), 60: 1 151-1 158, 1987.
39. Andersen, J. S., Burns, H. A., Casper, E., Clayman, M., Green, M.,
Nelson, R. L., Portenoy, R., Rothenberg, M., Tarassoff, P. G., and Von
Hoff, D. D. Development of a new system for assessing clinical benefitfor patients with advanced pancreatic cancer. Proc. Am. Soc. Clin.
Oncol. Annu. Meet., 13: 461, 1994.
40. Green, S., and Weiss, G. R. Southwest Oncology Group standard
response criteria, endpoint definitions, and toxicity criteria. Invest. New
Drugs, 10: 239-253, 1992.
41. Moore, M., Andersen, J., Burns, H., Tarassoff, P., Green, M.,
Casper, E., Portenoy, R., Modiano, M., Cripps, C., Nelson, R., Stor-niolo, A., and Von Hoff, D. D. A randomized trial of gemcitabine
(GEM) versus 5FU as first-line therapy in advanced pancreatic cancer.Proc. Am. Soc. Ciin. Oncol. Annu. Meet., 14: 199, 1995.
42. Burns, H. A., III, Moore, M. J., Andersen, J., Green, M. R.,
Rothenberg, M. L., Modiano, M. R., Cripps, M. C., Portenoy. R. K.,
Storniolo, A. M., Tarassoff, P., Nelson, R., and Von Hoff, D. D.
Improvements in survival and clinical benefit with gemcitabine as
first-line therapy for patients with advanced pancreas cancer: a random-
ized trial. J. Clin. Oncol., 15: 2403-2413, 1997.
43. Rothenberg, M. L., Moore, M. J., Cripps, M. C., Andersen, J. S.,
Portenoy, R. K., Burns, H. A., III, Green, M. R., Tarassoff, P. G.,
Brown, T. D., Casper, E. S., Stomiolo, A-M., and Von Hoff, D. D. A
Phase II trial of gemcitabine in patients with 5-Hi-refractory pancreas
cancer. Ann. Oncol., 7: 347-353, 1996.
44. Carmichael, J., Fink, U., Russell, R. C., Spittle, M. F., Harris, A.,Spiessel, G., and Blatter, J. Phase II study of gemcitabine in patients
with advanced pancreatic cancer. Br. J. Cancer, 73: 101-105, 1995.
45. Bey, P.. Danzin, C., Von Dorsselaer, V., Mamont, P., Jung, M., andTardif, C. Analogues of ornithine as inhibitors of omithine decarboxy-
lase: new deductions conceming the topography of the enzyme’s active
site. J. Med. Chem., 21: 50-55, 1978.
46. Pegg, A. E. Polyamine metabolism and its importance in neoplastic
growth and as a target for chemotherapy. Cancer Res., 48: 759-784,
1988.
47. Mitchell, M. F., Tortolero-Luna, G., Lee, J. J., Hittelman, W. N.,
Lotan, R., Wharton, J. T., Hong, W. K., and Nishioka, K. Phase Idose-escalation trial of a-difluoromethylomithine in patients with cer-
vical intraepithelial neoplasia grade 3. Clin. Cancer Res., 4: 303-3 10,
1998.
48. Porter, C., Herrera-Omelas, L., Pera, P., Retrelli, N. F., and Mittel-
man, A. Polyamine biosynthetic activity in normal and neoplastic hu-man colorectal tissue. Cancer (Phila.), 60: 1275-1281, 1987.
49. Garewal, H. S., Sampliner, R., Gemer, E., Steinbronn, K., Alberts,D., and Kendall, D. Omithine decarboxylase activity in Barrett’s esoph-
agus: a potential marker for dysplasia. Gastroenterology, 94: 8 19-821,
1988.
50. Manni, A., Mauger, D., Gimotty, P., and Badges, B. Prognostic
influence on survival of increased omithine decarboxylase activity in
human breast cancer. Clin. Cancer Res., 2: 1401-1406, 1996.
51. Auvinen, M., Paasinen, A., Anderson, L. C., and Holtta, E. Omi-
thine decarboxylase activity is critical for cell transformation. Nature
(Lond.), 360: 355-358, 1992.
52. Thompson, H. J., Meeker, L. D., Herbst, E. J., Ronan, A. M., and
Minocha, R. Effect of concentration of DL-a-difluoromethylomithine onmurine mammary carcinogenesis. Cancer Res., 45: 1170-1173, 1985.
53. Reddy, B. S., Nayini, J., Tokumo, K., Rigotty, J., Zang, E., and
Kelloff, G. Chemoprevention of colon carcinogenesis by concurrentadministration of piroxicam, a nonsteroidal anti-inflammatory drug with
DL-a-difluoromethylornithine on ornithine decarboxylase inhibitor.
Cancer Res., 50: 2562-2568, 1990.
54. Homma, Y., Kakizoe, T., Samma, S., and Oyasu, R. Inhibition of
N-butyl-N (hydroxybutyl) nitrosamine-induced rat urinary bladder
carcinogenesis by a-difluoromethylornithine. Cancer Res., 47: 6176-
6179, 1987.
55. Weeks, C. E., Herrimann, A. L., Nelson, F. R.. and Sloga. T. J.Difluoromethylomithine, an irreversible inhibitor of ornithine decarbox-
ylase, inhibits tumor promoter-induced polyamine accumulation and
carcinogenesis in the mouse skin. Proc. NatI. Acad. Sci. USA, 74:
6028-6032, 1982.
56. Abeloff, M. D., Slavik, M., Luk, G. D., Griffen, C. A., Hermann, J.,
Blanc, 0.. Sjoerdsman, A., and Baylin, S. B. Phase I and pharmacoki-netic studies of a-difluoromethylornithine-an inhibitor of polyamine
biosynthesis. J. Clin. Oncol., 2: 124-130, 1984.
57. Griffin, D. A., Slavik, M., Chien, S. C., Hermann, J., Thompson, G.,
Blanc, 0., Luk, G. D., Baylin, S. B., and Abeloff, M. D. Phase I trial andpharmacokinetic study of intravenous and oral a-difluoromethylomi-thine. Invest. New Drugs, 5: 177-186, 1987.
58. Love, R. R., Carbone, P. P., Vermia, A. K., Gilmore, D., Carey, P.,
Tutsch, K. D., Pomplun, M., and Wilding, G. Randomized Phase I
chemoprevention dose-seeking study of a-difluoromethylornithine.
J. Natl. Cancer Inst., 85: 732-737, 1993.
59. Kelloff, G. J., Boone, C. W., Crowell, J. A., Steele, V. E., Lubet, R.,
and Doody, L. A. Surrogate endpoint biomarkers for Phase II cancer
chemoprevention trials. J. Cell. Biochem., 19: 1-9, 1994.
60. Meyskens, F. L., Jr., Emerson, D. S., Pelot, D., Meshkinpour, H.,Shassetz, L. R., Einspahr, J., Alberts, D. S., and Gemer, E. W. Dose
de-escalation chemoprevention trial of a-difluoromethylornithine in pa-
tients with colon polyps. J. Natl. Cancer Inst., 86: 1 122-1130, 1994.
61. Alberts, D. S., Dorr, R. T., Aikin, M., Quinn, J., Goldman, R.,Nelson, M., Xu, M. J., Peng, Y. M., Bozzo, P., and Gensler, H. Positive
randomized, double-blinded, placebo-controlled study of topical diflu-
oromethylomithine (DFMO) in the chemoprevention of skin cancer.
Proc. Am. Soc. Clin. Oncol. Annu. Meet., 15: 172, 1996.
62. Schipper, H., Goh, C. R., and Wang, T. L. Shifting the cancer
paradigm: must we kill to cure? J. Clin. Oncol., 13: 801-807, 1994.
Research. on June 13, 2021. © 1998 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
http://clincancerres.aacrjournals.org/
-
1998;4:1079-1086. Clin Cancer Res D D Von Hoff Award Lecture.designs--twenty-first Richard and Hinda Rosenthal Foundation There are no bad anticancer agents, only bad clinical trial
Updated version
http://clincancerres.aacrjournals.org/content/4/5/1079
Access the most recent version of this article at:
E-mail alerts related to this article or journal.Sign up to receive free email-alerts
Subscriptions
Reprints and
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Permissions
Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)
.http://clincancerres.aacrjournals.org/content/4/5/1079To request permission to re-use all or part of this article, use this link
Research. on June 13, 2021. © 1998 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
http://clincancerres.aacrjournals.org/content/4/5/1079http://clincancerres.aacrjournals.org/cgi/alertsmailto:[email protected]://clincancerres.aacrjournals.org/content/4/5/1079http://clincancerres.aacrjournals.org/