Cancer Chemother Pharmacol (1998) 41: 173±185
`
`Ó Springer-Verlag 1998
`
`O R I G I N A L A R T I C L E
`
`Joseph J. DeGeorge á Chang-Ho Ahn
`Paul A. Andrews á Margaret E. Brower
`Diana W. Giorgio á M. Anwar Goheer
`Doo Y. Lee-Ham á W. David McGuinn
`Wendelyn Schmidt á C. Joseph Sun
`Satish C. Tripathi
`Regulatory considerations for preclinical development
`of anticancer drugs
`
`Received: 19 December 1996 / Accepted: 2 June 1997
`
`Abstract The entry of new anticancer treatments into
`phase I clinical trials is ordinarily based on relatively
`modest preclinical data. This report de®nes the battery
`of preclinical tests important for assessing safety under
`an Investigational New Drug application (IND) and
`outlines a basis for extrapolating starting doses of in-
`vestigational anticancer drugs in phase I clinical trials
`from animal toxicity studies. Types of preclinical studies
`for the support of marketing of a new anticancer drug
`are also discussed. This report addresses di€erences and
`similarities in the preclinical development of cytotoxic
`drugs (including photosensitizers and targeted delivery
`products), drugs used chronically (chemopreventive
`drugs, hormonal drugs, immunomodulators), and drugs
`intended to enhance the ecacy (MDR-reversing agents
`and radiation/chemotherapy sensitizers) or diminish the
`toxicity of currently used anticancer therapies. Factors
`to consider in the design of preclinical studies of com-
`bination therapies, alternative therapies, and adjuvant
`therapies in the treatment of cancer, and to support
`changes in clinical formulations or route of adminis-
`tration, are also discussed.
`
`Key words Antineoplastic agents á Toxicity tests á
`Toxicology á Guidelines á Phase I clinical trials
`
`This article is not an ocial FDA guidance or policy statement. No
`ocial support or endorsement by the Food and Drug Adminis-
`tration is intended or should be inferred
`
`J.J. DeGeorge á C-H. Ahn á P.A. Andrews (&)
`M.E. Brower á D.W. Giorgio á M.A. Goheer á
`D.Y. Lee-Ham á W.D. McGuinn á W. Schmidt á
`C.J. Sun á S.C. Tripathi
`Division of Oncology Drug Products,
`Center for Drug Evaluation and Research,
`Food and Drug Administration, 5600 Fishers Lane HFD-150,
`Rockville, Maryland 20857, USA
`Tel. (301)-594-5782; Fax: (301)-594-0499;
`E.mail: andrewsp@cder.fda.gov
`
`Introduction
`
`Malignant, nonresectable cancers are life-threatening,
`and aggressive measures are used in treating them.
`Antineoplastic
`therapies
`frequently
`include
`toxic
`chemicals or biological products that are designed to
`destroy tumor tissue or halt cell replication. Despite
`the serious toxicities of many anticancer drugs, careful
`dosing, clinical monitoring and prompt treatment of
`toxicity makes the side e€ects less threatening to a
`patient than their disease. Since it is recognized that
`doses of anticancer drugs high enough to kill cancer
`cells usually induce serious side e€ects in patients, the
`preclinical testing of oncology drugs di€ers from test-
`ing of nononcology drugs. The Division of Oncology
`Drug Products within the Center for Drug Evaluation
`and Research (CDER) at the US Food and Drug
`Administration (FDA) recognizes the urgency of de-
`velopment of new anticancer drugs and the need to
`rapidly move promising agents into clinical studies.
`This report o€ers a regulatory perspective on the pre-
`clinical development of new anticancer drugs that is
`intended to clarify the di€erences from the preclinical
`testing of nononcology drugs and to describe the data
`that are important to support human testing and even-
`tual marketing.
`The types of preclinical studies expected for support
`of clinical trials and then marketing of a new drug de-
`pend on both the intended use of the drug and the
`population of patients being studied and treated. In
`situations where potential bene®ts are greatest (ad-
`vanced, life-threatening disease), greater risks of treat-
`ment
`toxicity can be accepted and the required
`preclinical testing can be minimal. In cases where the
`patient population is free of known disease (e.g. adju-
`vant therapy or chemoprevention) the acceptable risks
`are much less and preclinical evaluation should be
`more extensive [32]. The toxicities of many modulating
`agents intended to enhance the ecacy or diminish the
`toxicity of anticancer agents are more similar to those of
`Genentech 2133
`Celltrion v. Genentech
`IPR2017-01122
`
`

`

`174
`
`these modulating
`nononcology therapies. However,
`agents could enhance the toxicity or diminish the activity
`of cytotoxic drugs by altering their toxicodynamics,
`pharmacodynamics, and pharmacokinetics. Thus, toxi-
`cological evaluation in combination with the modulated
`cytotoxic drug is an important part of preclinical de-
`velopment.
`The following considerations are o€ered in an e€ort
`to balance the risks to be borne by the proposed patient
`population and the realities of drug testing in humans.
`The di€erences in preclinical testing between cytotoxic,
`chronic (i.e. adjuvant therapy, chemopreventive drugs,
`hormonal drugs, and immunomodulators), and modu-
`lating therapies are emphasized. Issues of chemistry and
`manufacturing controls, clinical study design, and de-
`velopment of biologic agents for cancer treatment are
`beyond the scope of this report. If the appropriate
`preclinical development strategy remains uncertain after
`contemplating
`the
`following
`considerations,
`then
`sponsors are encouraged to initiate pre-IND discussions
`with Division sta€ regarding their preclinical study
`plan.
`
`General considerations for anticancer drug development
`
`Preclinical studies of anticancer agents
`
`The safety of ®rst-time use in humans is assessed
`through preclinical
`studies of pharmacodynamics,
`pharmacokinetics (toxicokinetics), toxicity, and their
`relationships. The purposes of these safety studies are:
`(a) to determine a starting dose for clinical trials that is
`both reasonably safe and allows for possible clinical
`bene®t for the patient, (b) to identify potential end-or-
`gan toxicities and determine their reversibility, and (c) to
`assist in the design of human dosing regimens and es-
`calation schemes for clinical
`trials. Animal
`toxicity
`studies most e€ectively accomplish these objectives when
`performed using schedules, durations, formulations, and
`routes comparable to those proposed in clinical studies.
`Use of longer duration preclinical studies may lead to
`underestimates of the appropriate clinical dose, while
`shorter studies may not identify cumulative dosing tox-
`icities. The toxicity studies should generally conform to
`the protocols recommended by the National Cancer
`Institute
`for
`toxicology assessment
`for anticancer
`agents1 and are expected to be conducted in accordance
`with Good Laboratory Practices (GLP) [16, 17]. When
`studies are not performed according to GLP, deviations
`should be documented and the potential impact of these
`deviations on study outcome and credibility should be
`described [16, 17].
`Typically, only two toxicology studies are essential to
`support initial phase I clinical trials in patients with
`
`1 The Developmental Therapeutics Program; Division of Cancer
`Treatment, Diagnosis, and Centers; National Cancer Institute
`(Rockville, MD USA) may be contacted for protocol details
`
`advanced cancers (Table 1). The ®rst of these is usually a
`study in rodents that identi®es doses that produce life-
`threatening and non-life-threatening toxicity. The sec-
`ond study should determine whether doses identi®ed as
`tolerable in rodents produce life-threatening toxicity in a
`non-rodent species. At least one of these studies should
`assess clinical signs, body weight, food consumption,
`clinical pathology, and gross pathology over a range of
`doses from nontoxic to toxic and should include an ex-
`amination of histopathology at doses that cause toxicity
`(or at the highest dose tested). Genotoxicity tests are not
`generally needed for cancer chemotherapies to support
`testing in phase I clinical studies unless healthy volun-
`teers will be entered into the study.
`While not essential, information on the pharmaco-
`dynamics and pharmacokinetics of drugs is extremely
`valuable for supporting the safety pro®le and can sig-
`ni®cantly contribute to the eciency of drug develop-
`ment. A phase I study may be conducted with no in vitro
`or in vivo preclinical pharmacodynamic information,
`but preclinical studies on biological activity and ecacy
`can substantially aid in clinical study design. Such
`studies help estimate e€ective dosages, dosing schedules,
`and optimal plasma concentrations. This information is
`likely to be particularly useful when developing noncy-
`totoxic agents. It may be desirable to develop such
`agents (e.g. immunomodulators) by escalating the hu-
`man dose to a pharmacodynamically active range rather
`than to the maximum tolerated dose (MTD). Pharma-
`cokinetic data can be gathered as a part of pharmacol-
`ogy or toxicity studies and do not usually need to be
`collected separately. Single- and multiple-dose pharma-
`cokinetic studies in the most appropriate species are best
`performed using dosing schedules, durations, and routes
`comparable to those that will be used in clinical studies
`[15]. The pharmacokinetic information obtained assists
`the evaluation of animal toxicity and ecacy, and may
`suggest modi®cations in the intended dose, route or
`schedule for the clinical trial. The importance of the
`parameters being measured will vary depending on the
`clinical trial design and therapeutic classes as discussed
`in the subsections below. In combination with pharma-
`codynamic data, this information can be used to help
`calculate initial doses in humans that have a greater
`likelihood of activity without adversely a€ecting safety,
`and can contribute to optimal dose escalation in early
`clinical studies.
`The proposed therapeutic indication, the outcome of
`early clinical development, the nature of toxicities seen
`in animals and in humans, and the projected duration
`of clinical treatment all determine the preclinical studies
`necessary to support a New Drug Application (NDA).
`In general, for oncology drugs, sponsors should con-
`duct toxicity studies using the same schedule and du-
`ration of administration as
`the
`intended clinical
`treatment cycle (Tables 1±3). Cytotoxic drugs used to
`treat advanced disease rarely need studies with more
`than 28 days of dosing submitted with the NDA (Ta-
`ble 1). In contrast, for drugs intended for continuous
`
`

`

`Table 1 Preclinical studies for cytotoxic oncology drugs
`
`Stage
`
`Category
`
`Issues to be addressed
`
`Studies considered importanta
`
`IND
`
`All cytotoxics
`
`Modi®cations for
`Special Categories
`Photosensitizer
`
`Starting dose,
`end-organ toxicities
`Genetic toxicity
`E€ective concentrations,
`schedule
`
`Systemic toxicity
`
`Phototoxicity
`
`Plasma t1/2
`
`Rodentb and nonrodentc
`toxicologyd
`Genetic toxicity panele
`
`Toxicology studies
`in subdued light
`
`Antibody conjugate
`
`Stability
`Toxicity of drug alone
`Speci®city
`
`Stability in plasma
`Toxicology in one species
`Human tissue screen
`
`Pharmacokinetics
`
`Liposomal delivery
`
`Drug product toxicity
`
`Depots
`
`Pharmacokinetics versus
`free drug
`Drug product toxicity
`
`Include free drug and blank
`liposomes in toxicity testing
`Pharmacokinetics
`
`Include free drug and
`empty depot in toxicity testing
`
`Toxicity to contacted tissues
`
`Histopathology of depot site
`
`175
`
`Studies considered
`useful
`
`Pharmacokinetics,
`pharmacodynamics
`
`In vivo study with
`illuminated skin
`Pharmacokinetics
`
`Activity in cell lines
`‹ target antigen
`
`Pharmacokinetics
`
`NDA All cytotoxics
`
`Rodent and nonrodent
`toxicologya,g,
`genetic toxicity,
`stage C-D teratogenicityf
`in rodents and nonrodents
`
`Targeted special
`toxicity
`
`a In general, the schedule and duration of administration in the toxicology study should mimic the clinical trial
`b Should determine the dose severely toxic to 10% of the animals (STD10)
`c Should determine toxicity of one-tenth the rodent STD10 on a mg/m2 basis
`d One study should include histopathology
`e Only for phase I testing in normal volunteers or patients believed to be disease-free
`f Should be submitted during development
`g Studies with more than 28 days of dosing are rarely needed
`
`daily administration such as for chemoprevention, ad-
`juvant therapy, or long-term hormonal or immuno-
`modulation
`therapy,
`chronic
`studies
`should
`be
`conducted up to a maximum of 6 months in rodent and
`12 months in nonrodent species (Table 2). International
`Conference on Harmonization (ICH) stage C-D2 re-
`productive toxicity studies in a rodent and a non-rodent
`species are important components of the preclinical
`evaluation of anticancer drugs and should be submitted
`early in development [14].
`Carcinogenicity studies are not required for cytotoxic
`drugs used to treat advanced systemic disease, but can
`
`2 ICH stage A-B, C-D, and C-F reproduction toxicity studies
`correspond to the previously designated segment I, II, and III
`studies which are de®ned by daily administration of drug,
`respectively, during the period from premating to implantation,
`implantation to birth (period of organogenesis), and implantation
`to sexual maturity [14]
`
`be important in the assessment of drugs intended for
`chronic use for chemoprevention, adjuvant, or hor-
`monal therapy when patients are likely to have a long
`survival [18]. The current standard is the 2-year rodent
`bioassay [47], although alternatives may be suitable [20].
`Depending upon the nature of toxicities seen with the
`drug or drug class in animals and in humans, targeted
`special toxicity studies to support NDA ®ling may also
`be needed. For example, in the development of anthra-
`cyclines and platinum drugs, which are known to have
`cardiotoxic and ototoxic potential, respectively, addi-
`tional preclinical cardiotoxicity and ototoxicity studies
`have been useful [11, 28, 36, 46]. In addition, neonatal
`reproductive toxicology and DNA adducting studies
`have been useful in the development of antiestrogenic
`agents [5, 30, 31, 37, 44, 45]. A discussion with FDA sta€
`on the preclinical studies needed for marketing approval
`for a particular drug is recommended at or before the
`end of phase II clinical studies.
`
`

`

`176
`
`Table 2 Preclinical studies for noncytotoxic, chronically administered oncology drugs
`
`Category
`
`Studies considered important
`
`Studies considered useful
`
`Stage
`
`IND
`
`All noncytotoxic chronic
`therapy
`
`Modi®cations for special
`categories
`Adjuvant therapy
`
`Chemopreventive
`
`Hormonal
`
`Immunomodulator
`
`Rodenta and nonrodentb
`toxicologyc,d
`
`Pharmacokinetics,
`-dynamics
`
`Ecacy studies
`
`Carcinogenicitye
`Stage A-B reproductive toxicity
`Stage C-D teratogenicitye
`
`Genetic toxicity panele
`
`Toxicology studies should
`also de®ne NOAEL,
`Genetic toxicity panel
`
`28-day toxicology studies usually suce for
`limited phase I/II testing in advanced cancer,
`genetic toxicity panele
`
`28-day toxicology studies usually suce for
`limited phase I/II testing in advanced cancer,
`genetic toxicity panele, de®ne dose versus
`immunologic response curve to identify shape
`(bell-shaped?) and surrogate markers
`
`Toxicology studies of equivalent duration to
`labeled use up to 6 months in rodents and
`12 months in nonrodents, genetic toxicity panel,
`carcinogenicityf, stage C-D teratogenicity in
`rodents and non-rodents
`
`NDA
`
`All non-cytotoxic chronic
`therapy
`
`Additional for hormonal
`
`Stage A-B reproductive toxicity
`
`Stage C-F reproductive toxicity,
`neonatal reproductive
`tract toxicity, DNA
`adducting (drug speci®c)
`
`Additional for
`chemopreventive
`
`Stage A-B and C-F reproductive toxicity
`carcinogenicity (always)
`
`a Should determine the dose severely toxic to 10% of the animals (STD10)
`b Should determine toxicity of one-tenth the rodent STD10 on a mg/m2 basis
`c In general, the schedule of administration in the toxicology study should mimic the clinical trial with a duration as long as the intended
`clinical study up to 6 months in rodents and 12 months in non-rodents
`d One study should include histopathology
`e Expected prior to clinical testing in patients with low risk of cancer recurrence, or testing in healthy volunteers
`f May be unnecessary depending on intended patient population [18]
`
`Starting doses and dose escalation
`
`As described above, one of the primary goals of preclin-
`ical studies is to estimate a safe starting dose for the ini-
`tiation of phase I trials in humans. The starting dose for
`clinical trials with cytotoxic drugs for oncology indica-
`tions has traditionally been one-tenth the dose lethal to
`10% of rodents on a body surface area basis (milligrams
`per meter squared) [23, 29, 35]. Studies that actually
`measure death as an endpoint, however, are not required
`so long as the dose range studied includes doses that cause
`severe, life-threatening toxicity. Thus, the starting dose is
`generally now chosen as one-tenth of the dose that causes
`severe toxicity (or death) in 10% of the rodents (STD10)
`
`3 This calculation is the same as taking one-third of the toxic dose
`low (TDL) [29, 35]. We believe the current expression of ``one-sixth
`the highest non-severely toxic dose'' is simpler and can be applied
`to the data more universally than taking, in practice, ``one-third the
`dose which causes toxicity but when doubled does not kill the non-
`rodents''. Frequently, the TDL cannot be technically de®ned in
`many studies
`
`on a milligrams per meter squared basis, provided that
`this starting dose, i.e. one-tenth the STD10, does not cause
`serious irreversible toxicity in a nonrodent species [29,
`35]. If irreversible toxicities are produced at the proposed
`starting dose in nonrodents (usually dogs) or if the non-
`rodent is known to be the more appropriate animal
`model, then the starting dose would generally be one-
`sixth of the highest dose tested in nonrodents that does
`not cause severe, irreversible toxicity3. In some cases,
`rodents or dogs may not be appropriate species because
`they do not model the relevant human biochemical or
`metabolic processes. For example, folate pools in rodents
`greatly exceed those in humans [4], so that rodents are
`generally inappropriate species for testing antifolates.
`Also, dogs poorly predict the toxicity of some platinum
`analogues, and an alternate animal model might be pre-
`ferred [34]. Knowledge of relevant physiological, bio-
`chemical, and pharmacokinetic di€erences between
`humans and animal models can help determine the most
`appropriate species to be used for selecting a starting
`dose. Whenever feasible, these starting doses should be
`
`

`

`Table 3 Preclinical studies for modulators of oncology drugs
`
`Stage
`
`Category
`
`Issues to be addressed
`
`Studies considered importanta
`
`IND
`
`All modulators
`
`Starting dose,
`end-organ toxicities
`
`Genetic toxicity
`E€ective concentrations,
`schedule
`
`Rodentb and
`non-rodentc
`toxicologyd
`genetic toxicity panele
`
`Additional studies for
`special categories
`MDR modulator
`
`Combination toxicity
`
`Chemosensitizer
`
`Combination toxicity
`
`Pharmacokinetic perturbations
`
`One species at minimally
`and signi®cantly toxic
`doses of cytotoxic
`Pharmacokinetics
`
`One species at minimally and
`signi®cantly toxic doses of
`cytotoxic
`
`Radiation sensitizer
`
`Delayed toxicity to normal
`tissues
`
`Chemoprotection
`
`Combination toxicity, tumor
`protection
`
`In vivo ecacy of combination
`with histopathology
`
`NDA All modulators
`
`Toxicology studies of equivalent
`duration to labeled use up to
`6 months in rodents and 12 months
`in non-rodents, genetic toxicity,
`stage C-D teratogenicity in rodents
`and non-rodents
`
`a In general, the schedule and duration of administration in the toxicology study should mimic the clinical trial
`b Should determine the dose severely toxic to 10% of the animals (STD10)
`c Should determine toxicity of one-tenth the rodent STD10 on a mg/m2 basis
`d One study should include histopathology
`e Only for phase I testing in normal volunteers or patients believed to be disease-free
`
`177
`
`Studies considered
`useful
`
`Pharmacokinetics
`
`In vivo ecacy of
`combination
`
`Skin/leg contracture
`
`Targeted special
`studies
`
`calculated from studies using the proposed clinical route,
`schedule, and duration.
`The dose escalation scheme for phase I clinical studies
`often follows the standard or modi®ed Fibonacci proce-
`dure [10]. Examples of other common and acceptable
`approaches include modi®ed continual reassessment
`methods [13, 39] and pharmacokinetically guided dose
`escalation strategies [8]. These alternatives often necessi-
`tate a more extensive preclinical evaluation. For example,
`pharmacokinetic guidance of dose escalation is most ef-
`fectively applied when: (a) linear pharmacokinetics are
`observed at drug concentrations spanning the pharma-
`cological and toxicological e€ects, (b) the area under the
`drug concentration versus time curve (AUC) at the mouse
`STD10 can be de®ned, (c) protein binding in mouse and
`human plasma has been quanti®ed, and (d) it is known
`whether metabolites contribute to the toxic e€ects [7, 8,
`27, 40]. Although preclinical studies are used to determine
`the starting dose for phase I clinical trials, the highest
`doses for oncology drugs are rarely restricted by the doses
`used in preclinical toxicology studies as long as the tox-
`icities of the new anticancer drug can be readily moni-
`tored, are reversible, and suciently precede lethality in
`animals. Instead, the maximum dose is restricted by the
`toxicity observed in the clinical trial, judged most often
`using NCI/DCTDC Common Toxicity Criteria [38].
`
`Considerations for speci®c cytotoxic therapies
`
`Combinations of cytotoxic agents
`
`The evaluation of cytotoxic agent combinations has
`traditionally been conducted in the clinical setting using
`an empirical approach. This has generally been suc-
`cessful, but may not be optimal. Preclinical studies
`provide an opportunity to explore a variety of doses,
`dose ratios, and schedules to optimize bene®t and min-
`imize toxicity. Nonetheless, unless there is reason to
`believe that synergistic interactions occur that would
`substantially increase the toxicity of the combination,
`preclinical testing is not considered essential provided
`that each agent has been fully evaluated in humans.
`When synergistic e€ects may be anticipated such as
`when one agent interferes with the metabolism or elim-
`ination of the other agent or both cytotoxic agents target
`the same metabolic pathway or cellular function, pre-
`clinical testing of the combination is desirable.
`
`Photosensitizers
`
`One class of cancer chemotherapeutic drugs is thera-
`peutically inactive until
`irradiated with light. These
`
`

`

`178
`
`photosensitizers or phototherapy agents usually form
`radicals after absorbing light energy that are ultimately
`responsible for tumor destruction. In photosensitizer
`therapy, tumor tissues are typically irradiated with laser
`light. When there is a choice, longer wavelengths of the
`irradiating light are preferred because they cause less
`direct tissue damage and because they penetrate more
`deeply into tumor tissue than shorter wavelengths.
`Selective damage to tumor tissue is obtained by di-
`recting the activating light to the tumor. In addition,
`most phototherapy compounds concentrate in tumor
`tissues more than in surrounding normal tissue when
`given systemically. This increased concentration of
`photosensitizer combined with localized irradiation can
`kill tumor cells with great selectivity. Nevertheless, when
`these compounds are given systemically they commonly
`distribute in appreciable concentrations in all tissues and
`this provides the potential for toxicity. When these drugs
`accumulate in the eye or skin, patients may su€er irre-
`versible retinal damage or severe phototoxicity similar to
`sunburn when exposed to ambient light [12]. Thus, it is
`important to know the plasma elimination half-life (and,
`if possible, tissue elimination half-lives) in preclinical
`studies so that the length of time a patient should protect
`themselves from light can be estimated.
`Standard toxicity studies with multiple dose levels
`should be conducted in subdued illumination to clearly
`de®ne the systemic toxicities of the photosensitizer. Sub-
`dued lighting allows systemic toxicities to be more clearly
`distinguished from phototoxicities. In addition to these
`standard toxicity studies, it is bene®cial to assess photo-
`toxicity before phase I clinical investigation begins be-
`cause these drugs can cause prolonged photosensitivity.
`Acceptable models for these photosensitivity tests are
`either hairless or appropriately shaved species. The pho-
`tosensitivity assessment should include toxicity testing as
`a function of both light dose (total energy) and drug dose
`and should ideally determine the duration of sensitivity in
`relation to plasma levels of the photosensitizer. Since a
`primary concern for the patient is the toxicity related to
`sunlight exposure, the light source for these tests should
`have a spectral distribution that approximates sunlight.
`Frequently, doses that are well below the no observable
`adverse e€ect limit (NOAEL) when the animal is housed
`in subdued light are lethal when the animal is brie¯y ir-
`radiated. Even though the photodynamic e€ect is ex-
`pected to a€ect only tissues that are exposed to the light
`source, there is concern that photodegradation products
`could cause distant toxicities. Therefore, these photo-
`toxicity tests usually include standard assessments of
`clinical signs, clinical pathology, gross pathology, histo-
`pathology of major organs, and the reversibility of tox-
`icities. Clinical photodynamic therapy does not routinely
`involve repeated doses, and thus preclinical studies using
`daily irradiation during repeat dose testing may not be
`relevant to clinical safety concerns.
`Without light these photosensitizers may not cause
`genotoxicity in standard tests, but subsequent irradia-
`tion may cause considerable damage to the DNA of cells
`
`exposed to the compound. Thus, genotoxicity tests are
`best done with and without light. The assessment of
`clastogenicity and mutagenicity should be done with
`increasing compound concentrations at a high light
`dose, and with increasing light dose (total energy) using
`broad-spectrum light at high compound concentration.
`The highest doses of drug of each series of tests should
`be consistent with international standards [19, 22].
`In many cases an e€ective dose of drugs in this class is
`nontoxic in subdued light and the starting dose can be
`chosen based on ecacy studies rather than toxicity
`studies. This pertains only if the projected ecacious
`starting dose is lower than the safe dose estimated from
`the toxicity studies.
`
`Specialized drug delivery
`
`Administration of anticancer drugs as depots, attached
`to carriers, or in specialized encapsulated forms has the
`potential for signi®cantly improving ecacy. Advan-
`tages of specialized drug delivery may include: (a) spe-
`ci®c targeting of the drug to the tumor, (b) minimization
`of toxic side e€ects, (c) prolongation of therapeutic drug
`concentrations, (d) improved delivery of hydrophilic
`drugs to tumor cytoplasm, and (e) practical adminis-
`tration of very lipophilic drugs. Examples of delivery
`systems include copolymer implants, human albumin
`microspheres, monoclonal antibody±drug conjugates,
`and liposomal encapsulation. Development of antican-
`cer drugs administered via carriers or in depots may
`necessitate additional preclinical evaluation beyond that
`of conventional cytotoxic drugs.
`For antibody±drug conjugates, the two main safety
`concerns are the potential for toxicity from abrupt re-
`lease of the drug and the potential for the antibody±drug
`conjugate to cause unexpected, speci®c toxicity in nor-
`mal human tissues. Studies of the stability of the con-
`jugate in human plasma as a function of the proposed
`release mechanism (e.g. pH if hydrolytic, glutathione
`concentration if reductive) help determine the necessity
`of conducting additional toxicology studies [21]. When
`additional studies are indicated, using the form of the
`drug released from the conjugate (i.e. including linker
`groups) may identify clinically important
`toxicities.
`Testing the reactivity of the conjugate with a complete
`panel of human tissues from at least three di€erent
`sources is suggested [21]. When the target antigen is not
`expressed in the tissues of the standard preclinical ani-
`mal models, a tolerance study in Pongidae apes at a dose
`that is at least double the planned human starting dose
`should also be considered. Both the reactivity screen
`and the tolerance study may reveal sites of potential
`tissue-speci®c toxicity, while the standard toxicology
`studies may de®ne nonspeci®c toxicities. Speci®city
`studies of binding or cytotoxicity in cell lines with and
`without an expressed target antigen also help to assess
`whether there is a signi®cant di€erential between the
`toxicity to a targeted and nontargeted tissue. If feasible,
`
`

`

`pharmacokinetic studies that distinguish between con-
`jugate, free antibody, and free drug are also highly
`desirable for interpreting toxicology ®ndings and sup-
`porting interspecies comparisons. Selection of a starting
`dose for clinical study should consider not only the
`results of the toxicity studies with the conjugate, but
`also the stability of the conjugate and the potential
`toxicity of released drug.
`With liposomal drugs, standard preclinical toxicol-
`ogy studies of the delivery system, free drug, and the
`®nal formulation are important for evaluating a drug
`product's potential for toxicity. Liposomal formula-
`tions usually dramatically prolong systemic exposure.
`Thus, when repeated doses are to be used clinically, it is
`especially important to study a similar schedule pre-
`clinically because of the potential for drug accumula-
`tion. When the delivery system is designed to a€ect
`drug absorption, distribution, biotransformation, ex-
`cretion or target organ accumulation, small changes in
`the design of the delivery system may have substantial
`e€ects on overall
`toxicity. Conducting the toxicity
`studies with the ®nal formulation can avoid concerns
`about
`such e€ects. Comparative pharmacokinetic
`studies of the ®nal formulation versus free drug can be
`very helpful
`in suggesting schedules and interpreting
`changes in the spectrum and severity of toxicities. Oc-
`casionally, studies of the empty liposomes plus free
`drug in combination may also be useful for under-
`standing alterations in ecacy seen with the liposomal
`preparation. For example, blank liposomes may alter
`the pharmacokinetics of the free drug in a fashion
`sucient for therapeutic gain [33].
`Preclinical development of depot formulations gen-
`erally follows that of liposomal formulations. Addi-
`tionally, a study of the toxicity of the depot in the tissue
`or compartment intended to be used clinically should be
`conducted which includes a histopathologic examination
`of the adjacent tissues. Initial clinical doses similar to the
`total dose of the drug previously investigated in humans
`may be used in the absence of signi®cant changes in
`toxicity pro®le for the depot formulation.
`
`Alternative therapies
`
```Alternative'' therapies include both single agents and
`multicomponent entities derived from plants or animals.
`Herbal products and tissue or ¯uid extracts from animal
`sources intended for the treatment or prevention of
`cancer or precancerous conditions belong in this cate-
`gory. The identity of the active ingredient of these en-
`tities is frequently uncertain. Consistency in taxonomic
`identi®cation, collection, storage, and processing may
`pose additional diculties. A useful initial step is to
`prepare a batch of the drug product large enough to be
`sucient for both initial preclinical and clinical studies.
`The usual battery of toxicology studies for anticancer
`agents should be conducted unless there is adequate
`human safety experience. Since it is dicult to correlate
`
`179
`
`speci®c drug product components with pharmacologic
`action, attempts should be made early in the develop-
`ment scheme to control the manufacturing processes to
`produce consistent batches for subsequent preclinical
`and clinical study. Further e€orts should be made in the
`later stages of development to identify biologic assays
`which can be used to assure activity and as release
`speci®cations for the marketed product.
`Herbal products represent a specialized subset of al-
`ternative therapies, as there is often signi®cant human
`experience with their use. If there is a documented his-
`tory of use of the herbals or if these preparations are
`freely marketed in the United States, then no preclinical
`pharmacology or toxicology is required for initial trials
`using the marketed product. Submission of data on the
`traditional use, preparation of the product, and safety
`pro®le of any known components of the herbal prepa-
`ration for the IND is encouraged. When a product dif-
`ferent from the marketed version is intended for the
`clinical trial,
`information on the preparation of the
`product to be tested is import