Clinical Development of Anticancer A
`Perspective
`
`gents—A National Cancer Institute
`
`Silvia Marsoni* and Robert Wittes1
`
`Since the first report that a chemical could cause sig-
`nificant tumor shrinkage (1), the development of clin-
`ically useful antineoplastic drugs has grown from the
`preoccupation of a few investigators to a major interna-
`tional effort. Over the past 35 years or so, about 30
`drugs have been defined as active in one or more tumor
`types. When used alone in patients with disseminated
`mahgnancy,
`these drugs cause reduction in bulk of
`measurable neoplasm in a significant percent of cases;
`for most tumor types, however, ample evidence of re-
`51dual cancer usually persists and after a few months,
`tumor regrowth occurs. More striking successes have
`been achieved with combinations of drugs; as is well
`known, for several kinds of disseminated human can-
`cers, a high frequency of clinical complete remissions,
`with substantial long-term disease-free survival rates,
`is now possible (2—5). For other cancers which may not
`be curable by chemotherapy once they have disseminat-
`ed, combinations of drugs appear to result in a higher
`total remission rate and a greater prolongation of life
`than smgle drugs (6,7). Perhaps more significant in the
`long run is the apparent effect of chemotherapy when
`(1;
`used as part of a planned multimodality effort
`13y some perverse quirk of fate, chemotherapy seems to
`,
`chiefly exert a major impact in rare tumors, while the
`common
`epithelial
`neoplasms
`of
`adulthood
`have
`thus far resisted satisfactory solutions. Therefore, the
`central problem of drug development, the identification
`of effective agents with reasonable therapeutic index,
`IS as pertinent for oncology now as at any time in the
`past.
`The idealized outlines of the successive steps in drug
`development are familiar to all oncologists. In phase I
`trials, we define a dose suitable for use in studies of
`the drug’s activity across a spectrum of human tumors.
`Increasing awareness of the importance of patient— and
`disease-related parameters has effectively led to the re-
`placement of the broad phase II trial with a series of
`disease-oriented activity studies. Having defined set-
`
`investigators
`tings in which the new drug is active,
`then proceed to compare the new treatment with stand-
`ard therapy (phase III) and to further explore the
`drug’s therapeutic potential in other ways, such as in
`combination with other agents or by alternate routes of
`administration.
`
`Anyone familiar with the actual workings of this
`process over the past two decades knows that despite
`its successes, it has not functioned as systematically or
`efficiently as the above description might imply. In ad-
`dition, many of the assumptions on which the process
`was based are in need of reexamination. Since there
`are
`no
`reliable
`laboratory predictors of
`efficacy
`for specific human cancers, drug development will con-
`tinue to require extensive testing in human subjects, an
`endeavor that is never without ethical dilemmas, how-
`ever thoughtfully it is carried out. In addition, because
`human cancers vary widely in sensitivity to anticancer
`drugs,
`the apparatus required to sustain the clinical
`trials effort is necessarily large and very expensive. For
`these reasons alone, another look at the drug develop-
`ment program of the National Cancer Institute (NCI)
`seems to be worthwhile.
`
`Phase I
`
`Phase I trials of antineoplastic compounds are con-
`ducted in patients with disseminated malignancies for
`whom standard treatment either does not exist or has
`proved ineffective. Drugs are given in a phase I trial
`with therapeutic intent; the main scientific goal is to
`define the qualitative and quantitative characteristics
`of the drug’s acute toxicity, and in so doing, to deter-
`mine a biologically active dose which is tolerable for ev-
`ery patient. The maximum tolerated dose (MTD)
`is
`usually higher in children than in adults (10), probably
`because of better organ function and possibly because
`of different pharmacokinetics (11,12). Also, since the
`MTD for patients with acute leukemia is often substan-
`tially higher than that for solid tumors, at least four
`
`I
`
`.
`.
`.
`.
`Cancer Therapy Evaluation Program, Dmsnon of Cancer Treatment.
`Na'tional Cancer Institute, Bethesda, MD.
`Reprint requests to: Silvia Marsoni, MD, Drug Evaluation and Re
`
`.
`porting Section, Investigational Drug Branch, Cancer Therapy Evaluation
`Program, National Cancer Institute, Landow Bldg, Rm 4C09, 7910 Wood-
`mont Ave, National Institutes of Health, Bethesda, MD 20814.
`
`Cancer Treatment Reports Vol. 68, No. 1, January 1984
`
`77
`
`This mate ria i was copied
`
`Celltrion v. Genentech
`
`IPR2017—01122
`
`Genentech Exhibit 2023
`
`

`

`trials should be conducted for each drug. In
`phase I
`practice, of course, three to four phase I trials testing
`different schedules of the drug are usually performed
`in adult patients with solid tumor alone, and trials in
`children do not start until substantial experience is ac-
`cumulated from the adult trials.
`The assumption underlying all phase I escalation pro-
`cedures is that anticancer compounds must be given at
`or near the MTD; therefore, the job of a phase I trial is
`to define the highest dose that can be safely delivered
`to a patient, since this dose will also be the one that
`has the best chance of being active. This approach re-
`flects the difficulties in establishing a clear-cut measur-
`able endpoint for drug activity against cancer. Since
`induction of response is not usually an event that is im-
`mediately recognizable, attainment of toxicity is the
`only assurance that, if a response is not obtained, at
`least a biologically active dose was delivered. Underly-
`ing this assumption is the more fundamental one that
`the dose-response curve f0r human cancers is monoton-
`ically increasing throughout
`the range of
`tolerated
`doses and is to the right of the dose-toxicity curve.
`Needless to say, the details of this assumption have not
`been generally verified for antitumor agents, chiefly
`because rigorously defining the shape of a clinical dose-
`response curve is a laborious task, requiring a large
`number of patients treated at each of several dose lev-
`els. Where the relationship between dose and response
`has been examined, however, most of the data are at
`least consistent with the conclusion that the higher the
`administered dose, the more probable an antitumor ef-
`fect (13-17) or the longer the duration of remission
`(18). On the other hand, recent trials in small cell lung
`cancer suggest that the probability of response does
`not continue to increase linearly as the dose approaches
`the MTD (19).
`For most of the clinically useful compounds, the bone
`marrow is dose-limiting. The dose-toxicity curve for
`
`myelosuppression is quite reproducible, and the status
`of marrow reserve is the major source of interpatient
`variability. Generally, treatment of six to ten patients
`at or near the MTD is sufficient to establish a safe
`phase II dose when myelosuppression is the dose-limit-
`ing toxicity.
`However, major problems may arise when other toxic
`effects which are less easily quantifiable are dose-limit-
`ing. For example, in a phase I study of escalating doses
`of carmustine with autologous bone marrow support
`(20), major organ toxicity (liver, central nervous system,
`and lung) surfaced abruptly at a dose of 1500 mg/m2.
`Because of
`the sudden appearance of
`these side
`effects in the escalation scheme and the long interval
`from the beginning of treatment to onset of toxicity
`(6-9 weeks), the overall mortality rate for patients en-
`tered at a dose of 2 1500 mg/m2 was approximately
`30%.
`
`Experience suggests that whenever a drug has dose-
`limiting side effects other than myelosuppression, its
`transition into phase II has often been compromised.
`An analysis of 31 drugs entered in phase I evaluation
`by the NCI shows that whenever the drug had myelo-
`suppression alone as the dose-limiting toxicity, it had a
`high probability of undergoing full phase II study; how-
`ever, when other organ toxicity was dose-limiting, only
`25% of the drugs proceeded to full phase II study (ta-
`ble 1). Evaluation of the remainder of drugs was re-
`stricted by either the NCI or lack of investigator inter-
`est. The main reason for these difficulties relates large-
`ly to the uncertainty regarding reversibility of acute
`major organ damage. In addition, even if organ damage
`should turn out to be reversible, medical support dur-
`ing periods of severe organ failure is either extremely
`intensive and costly (kidney, CNS) or technically unsat-
`isfactory (liver), and is not seen as feasible or justifia-
`ble by most investigators in the context of a clinical ex-
`periment.
`
`TABLE 1.——Pliase II evaluation as a function of the dose-limiting toxicity of 31 cytotoxic compounds (1975- 1982)
`
`Dose-limiting toxicitym
`Myelosuppressicn
`Organ
`and organ toxicityi"
`toxicity”:
`3
`3
`3
`4
`
`Myelosuppression *
`10
`
`MF
`
`Phase II
`
`evaluation
`ull
`
`Restricted
`Dropped
`Toxicity
`No interest
`No drug supply
`
`0
`
`0
`1
`1
`
`0
`l
`0
`
`4
`1
`0
`
`12
`Total
`7
`
`'Biszmtrene, (liaziquOne, ziclarubicin (aclacinomycin). mitoxantrone, PCNU, amsacrinc. zorubicin. ChIOI'OZOtociIL
`carboplatin, ICRI“ 187, 5-metliyltetrahydrohomofolate, and 3-(lcnzauridine.
`JrAcivicin, maytansine, anguidine, teroxirone, dihydro-5-azacitidine, indicine N-oxide, and DON.
`iPyrazofurin, L-alanosinc, homoharringtonine, 'l‘CN-I’, N-methylformamide, pentostatin, liycantbone, dichloroul-
`lyl lawsone, pyrazole, aminotbiudiazole, bruceantin, and spirogermanium.
`
`12
`
`78
`
`This material was copied
`_xx_a__ LII ll___l _,_r_1_-
`
`Cancer Treatment Reports
`
`

`

`This dilemma appears to have no easy solutions. One
`alternative might be to utilize data from pharmacologic
`studies to define the relationship between dose, plasma
`level, tissue level, and clinical activity. For example, in
`the case of pentostatin, knowledge concerning the
`amount of drug needed to abolish activity of the target
`enzyme adenosine deaminase has helped to establish a
`phase II dose independent of the attainment of clinical
`toxicity. Unfortunately, however, this is an exceptional
`situation and, under most circumstances, specific intra-
`cellular targets of drug action either have not been
`identified or are not so susceptible to analysis.
`A somewhat more empiric approach is exemplified by
`the current plans for developing N-methylformamide, a
`drug which had been introduced into the clinic in the
`19505 and was subsequently dropped while in phase I
`because of hepatotoxicity (21). Interest in the drug has
`recently been revived because of its activity against hu-
`man tumor xenografts (22) and its capacity to induce
`differentiation in vitro (23). Phase I trials in both the
`United Kingdom and the US confirm that, at a dose of
`1000 mg/mz, the reversible hepatotoxicity of N-methyl—
`formamide is dose-limiting and myelosuppression is
`completely absent.2-3 This dose has been defined, as the
`MTD, at which phase II trials have just begun. If activ-
`ity is observed in any tumor type, a repeat phase II
`study in one or more susceptible tumors will be per-
`formed at a level
`immediately below the MTD. This
`procedure will define whether the attainment of toxic-
`ity is necessary for activity.
`
`Phase II
`
`In a phase II trial, the main goal is to assess the ac-
`tivity of the drug in a variety of disseminated malig-
`nancies and to further define the patterns of acute
`toxicity in patients who are homogeneous in diagnosis
`and in better general medical condition than patients in
`phase I. Since large numbers of patients are treated
`during phase II, rarer acute toxic effects often surface
`for the first
`time (24). Also, since cumulative drug
`doses may be appreciable in responding or stable pa-
`tients, phase II provides an appropriate setting for ini-
`tial assessment of chronic toxic effects.
`Several vexing problems are inherent in this process.
`In the first place, since patient numbers and resources
`are finite, it is impossible to explore the activity of
`each drug in each tumor type. A method must be found
`to focus the effort of drug development in a way that
`will minimize the chance of overlooking active agents.
`Accordingly,
`the NCI decided to evaluate all experi-
`
`chVie JG, ten Bokkel Huinink WW, Simonetti G, et al. Phase I trial of
`N-methylformamide (NSC 3051) (NMF). Manuscript submitted to Cancer
`Trea tmcnt Reports.
`3Minutes of the Phase I Working Group Meeting, NCI, Bethesda, MD,
`July 1983.
`
`mental drugs in selected types of cancer. The NCI
`Human Tumor Panel was created in 1975 and included
`lung, colon, and breast carcinomas, and lymphoma, leu.
`kemia, and melanoma. The original intention was to
`match tumors in the human panel with those in the
`preclinical panel, thereby providing information for the
`validation of the preclinical screening program. In addi-
`tion, these classes of human cancer represent the two
`extremes of chemotherapy sensitivity and might be ex-
`pected to exhibit both high sensitivity and high selec-
`tivity. Finally,
`the inclusion of
`the most common
`causes of cancer deaths (breast, colon, and lung can-
`cers) permits the study of large numbers of patients
`and assures that results will have immediate implica-
`tions for the treatment of prevalent cancers. Needless
`to say, evaluation of individual drugs is also carried out
`in tumors other than those in the panel, particularly if
`there is a specific reason to do so. For example, dia-
`ziquone was chemically designed to cross the blood-
`brain barrier and therefore has been extensively eval-
`uated in brain tumors with encouraging results.
`How has activity in the prelinical panel correlated
`with clinical activity? Thus far, we have analyzed the
`results with 13 experimental drugs for which clinical
`and experimental data are available. The correlation of
`activity in each model tumor system with activity in
`the corresponding human cancer is shown in figure 1.
`Prediction of true-negative results (resistance) seems
`fairly reliable across most of the rodent and xenograft
`systems. On the other hand, the probability of predicting
`true-positive results
`(sensitivity)
`is very low. Be-
`cause of the small number of active drugs in humans
`for which complete data are available, no definitive
`conclusions can be drawn. However, even if the pre-
`clinical panel should not turn out to be an accurate pre-
`dictor of response in individual tumor types, overall ac-
`tivity in prelinical screening may still serve as a gen-
`eral predictor of activity in at least one human cancer.
`The aggregated data are, in fact, consistent with this
`notion. This has obviously been the general premise on
`which antitumor screening programs have operated for
`years. Its validity has been widely assumed but has not
`been subjected to direct
`test,
`since drugs are not
`brought to the clinic if screening data are not positive.
`An assessment of the validity of the assumption will be
`afforded by the use of the human tumor stem cell
`assay as a screening tool. The plans are to bring
`selected compounds which are positive in the human
`tumor stem cell assay to clinical trial, even if they are
`negative in the P388 prescreen (26). Obviously, more
`data are needed to determine the ultimate usefulness of
`the panel.
`How has the human panel fared as a predictor of
`clinical efficacy in human tumors other than those of
`the panel? Since 1971, 62 cytotoxic agents have been
`introduced into clinical trials under the sponsorship of
`
`Vol. 68. No. 1,January 1984
`
`This material was {spied
`
`79
`
`

`

`CX-lsc
`Colon xenografts
`
`I+ I-I
`c —I_ ‘ ‘I—T
`o + |
`0
`|
`o
`|
`I
`I
`I
`|
`o "I" _ "I_'I_
`n -
`I
`0
`I
`8
`I
`I
`I
`I
`—T—’_1—'T'
`
`Colon 38
`
`_L:J -I
`I
`|
`I
`c
`0
`0
`o + i
`I
`I
`I __ _____.
`o
`I
`I“_I"
`5
`n —
`I
`i
`3
`i
`‘1__I““—I_
`
`Anguidine, Pyrazofurin, Maytansine, Chlorozotocin,
`Deazauridine, Rubidazone, PALA, Amsacrine
`
`LX—lsc
`Iung xenografts
`
`|
`|
`|
`|
`1_|
`I 0|
`0
`u+l
`n_I_*_l
`|
`g
`I
`I
`|
`'IZI6 I
`~1—_I—I'_
`
`3LL
`Lewis lung carcinoma
`
`|
`|
`|
`"I
`l
`I
`l 0|
`0
`u+l
`n_|___I
`I
`g
`|
`|
`|
`‘IZISI
`T—l—I—
`
`Bis
`melanoma
`
`L1210
`leukemia
`
`—
`0
`
`I
`|
`+
`I
`m
`I
`e_|__~|
`I
`1+|
`0
`|
`I
`a
`I
`|
`nfllfl__I—T_
`o—l
`l
`I
`6
`2
`m
`|
`|
`|
`a
`
`-
`1
`
`I
`l_I__:_|
`|
`e
`I
`I
`I
`u+I
`4
`I
`|
`k__I_____I
`e
`I
`l—I—
`m-I
`3
`I
`I
`1
`1'
`I
`I
`I
`a
`I
`|
`I
`
`Anguidine, Pyrazofurin,
`Maytansine, ChIorozotocin,
`Rubidazone, PALA,
`Amsacrine, DON
`
`Thymidine, Aciacinomycin,
`Anguidine, Chlorozotocin,
`Deazauridine, Rubidazone,
`Indicine N-Oxide,
`Amsacrine, Pyrézofurin
`
`MX-lsc
`breast xenografts
`
`_
`CDSFl
`mammary carc1noma
`
`1
`
`I+| —|
`b ‘1‘ " ‘I
`I
`r + I
`0
`I
`I
`e
`I
`I
`I
`a ‘I‘ “ "I
`|
`s -
`I
`1
`|
`I
`t
`I
`I
`_I—_I—_'I_
`
`6
`
`_L_J -I
`|
`I
`I
`b
`1
`r + |
`|
`I
`e_L_J
`I
`a
`|
`|
`I
`s -
`|
`|
`I
`t
`I
`|
`I
`_I—”_T‘fi_
`
`0
`
`3
`
`5
`
`Anguidine, Pyrazofurin, Maytansine, Chlorozotocin,
`PALA, Amsacrine, DON
`
`Anguidine, Pyrazofurin, Maytansine, Bruceantin, PALA,
`Amsacrine, Mitoxantrone, ChIorozotocin
`
`FIGURE 1.—Correlation of activity of 10 antitumor agents in murine model tumor systems with activity in human cancer. Activity in murine tumors was
`judged according to NCI Decision Network 2 criteria (Goldin A, et al. Eur J Cancer 17:129-142, 1981). Activity in human tumors is defined as a 20% re-
`sponse rate in at least 1 clinical trial with 2 14 evaluable patients.
`
`NCI. Results of an interim analysis of phase II results
`are available for 13 drugs, which were studied in 180
`protocols (table 2). Although these data represent only
`a fraction of the large NCI experience, certain trends
`are apparent. First, significant activity of 2 20% was
`seen. only in the lymphomas,
`leukemia, and breast
`carcmoma (50%, 29%, and 14% of the studies, respec-
`tlvely); most of the results were in the 20%-30% range.
`No drug showed > 20% activity in colon carcinoma and
`melanoma, and only 6% of the lung cancer trials showed
`pOSItive activity. Even with only those drugs which have
`shown activity in at least one tumor type, the overall re-
`sponse rate in colon and lung cancers and melanomas is
`Stlll consistently< 10%.
`
`It is well known that colon cancer and melanoma are
`highly resistant diseases, and that these diseases which
`are consistently refractory to all therapies are of no
`value in screening. Although more data are needed, the
`results to date suggest that inclusion of colon cancer
`and melanoma in the panel may not be useful for
`screening. However, it would seem reasonable to con-
`tinue testing new agents for activity in those common
`and refractory neoplasms until truly reliable screens
`for activity have been defined. Such screens may
`emerge from further analysis of data for clinical trials
`or from advances in the use of in vitro or in vivo lab-
`oratory methods.
`Second, even in intrinsically sensitive diseases like
`
`TABLE 2.-—Outcome of phase II studies in human cancer
`
`M
`
`Disease
`"Breast cancer
`Colon cancer
`Leukemia
`Lung cancer
`Lymphoma
`Melanoma
`
`Total No
`of studies
`30
`38
`21
`47
`18
`26
`
`Response rate
`< 20%
`13 (43%)
`13 (34%)
`9 (48%)
`1 7 (36%)
`5 (28%)
`10 (38%)
`
`13 (43%)
`25 (66%)
`6 (29%)
`27 (57%)
`4 (22%)
`16 (62%)
`
`2 20%
`4 (14%)
`0
`6 (29%)
`3 (6%)
`9 (50%)
`0
`
`———————.—“__—_
`Total
`180
`91 (51%)
`67 (37%)
`22 (12%)
`
`80
`
`TI'I is material was ended
`
`Cancer Treatment Reports
`
`

`

`breast or small cell lung cancer, the number of drugs
`showing activity turned out to be very small. As seen
`in table 3, of the 11 drugs considered in breast cancer,
`only bisantrene showed an overall response rate of
`> 20%. As has been true throughout
`the history of
`medical oncology (27),
`the estimates of activity vary
`widely from trial to trial. For example, with mitoxan-
`trone, response rates ranged from 5% to 28%. This ob-
`servation suggests the well-known importance of fac-
`tors other than drug dose and schedule as major influ-
`ences on estimates of response rate. Again, for mitoxan-
`trone, the deleterious effect of prior therapy on response
`seems to be fairly clear (table 4).
`In fact, the success of the human tumor panel in pre-
`dicting (or ruling out) general patterns of efficacy for
`other human cancers will depend to a large extent on
`what kinds of patients with “panel cancers” are chosen
`for entry in the study. A negative trial of a new drug
`in 20 patients with breast cancer who have failed mul-
`tiple prior regimens tells us nothing about either the
`potential of this drug in a more favorable breast cancer
`population or drug activity in other tumors. Moreover,
`data from earlier eras of cancer chemotherapy cannot
`be used reliably to decide which tumors may be use-
`fully included in a panel without extensive consider-
`ation of how shifting patterns of practice may have al-
`tered important patient characteristics.
`The phase II effort also needs certain administrative
`refinements. Table 5 shows a breakdown by disease of
`patient accrual patterns for negative phase II studies,
`ie, trials yielding a < 10% response rate. Even allowing
`for the histologic heterogeneity of certain primary sites
`such as lung, the extent of overaccrual in some of these
`categories suggests the need for much earlier review of
`the data by investigators and a tighter system of con-
`trol by the statistical offices of cooperative groups. In-
`deed, several groups have already implemented proce-
`
`TABLE 3.—Activity of 11 NCI drugs in patients with breast cancer
`(1975-1980)
`
`
`No. of responding patients/
`Response
`
`Drug
`total evaluable
`rate (%)
`Aclarubicin
`1/48
`2
`Amsacrine
`12/173
`7
`
`Anguidine
`Acivicin
`Bisantrene
`
`Bruceantin
`
`Diaziquone
`Mitoguazone
`Mitoxantrone
`PCNU
`
`1/37
`0/ l5
`18/50
`
`0/15
`
`2/63
`4/104
`16/182
`0/4 5
`
`3
`0
`26
`
`0
`
`3
`4
`9
`0
`
`
`
`3/47Piperazinedione 6
`
`
`
`TABLE 4.—Responses to mitoxantrone in carcinoma of the breast trials
`according to previous treatment
`Response
`No. of previous
`rate (%)
`regimens
`3
`3
`10
`3
`
`Institution ll
`SWOG
`SECSG
`
`5
`
`22
`
`2 1
`
`19
`
`2.6
`
`3
`
`1
`
`1
`
`ECOG
`
`M. D. Anderson Hospital
`and Tumor Institute
`Ohio State
`University
`EORTC
`
`The Royal Marsden
`0
`28
`Hospital
`
`* SWOG = Southwest Oncology Group; SECSG = Southeastern Cancer
`Study Group; ECOG = Eastern Cooperative Oncology Group; and EORTC
`= European Organization for Research on Treatment of Cancer.
`
`dures which should minimize the chances that patients
`will be entered in treatments already shown to be inac-
`tive.
`
`Phase III
`
`Once the activity of a compound is established in one
`or more diseases, subsequent development of the drug
`proceeds along two separate lines. One of these lines is
`to establish the role of the drug in the disease for
`which activity was demonstrated. The endpoints of
`such studies, which are designed to compare the drug
`alone or in combination against standard treatment in
`a randomized fashion, are not only relative activity (eg,
`response rate), but also response duration, survival, and
`toxicity; the ultimate goal is to define the specific con-
`tribution of the drug in the treatment of a particular
`cancer. The data from such trials may be used by
`pharmaceutic firms seeking New Drug Application
`(NDA) approval from the Food and Drug Administra-
`tion (FDA) for marketing purposes.
`In this connection,
`the intense interest in chemical
`analogs of existing active agents poses special chal-
`lenges to clinical drug development. Of 31 drugs devel-
`oped by NCI since 1975, eight have been analogs of
`commercially available or experimental drugs. Until
`very recently,
`the development of analogs proceeded
`along essentially the same lines as that of novel struc-
`tures. Formal prospective comparisons of analog versus
`parent were rarely carried out (28). As a result, little
`direct comparative data exist on the relative merits of
`the various bifunctional alkylating agents, nitrosoureas,
`anthracyclines, or epipodophyllotoxins.
`Surely, if parent and analog have borderline activity
`in a certain cancer, such direct comparisons are prob-
`ably not worth undertaking. Moreover, when such com-
`parisons are worth doing, the trials need to be quite
`
`Vol. 68, No. 1,January 1984
`
`81
`
`

`

`.
`.
`TABLE 5.—Patient accrual onto negative phase II studies us1ng 13 compounds
`—______—_____________—_'__—___—_____________
`No. of studies
`Median
`< 25
`25-50
`2 50
`No. of
`Disease
`Total
`patients
`patients
`patients
`patients
`
`Breast cancer
`10
`1 (10%)
`7 (70%)
`2 (20%)
`32
`Colon cancer
`9
`2 (22%)
`G (66%)
`1 (11%)
`81
`Leukemia
`5
`2 (410%)
`2 (40%)
`1 (20%)
`35
`
`t
`
`:15
`7 (4 7%)
`6 (40%)
`2 (13%)
`15
`Lung cancer (non-small cell)
`(31)
`O
`1
`0
`l
`Lymphoma
`
`
`
`
`
`8 2 (25%) 5 (62%) 1 (12%)Melanoma 85
`
`*Aclarubicin, bisantrene, amsacrine, anguidine,
`piperazinedione, and zinostatin.
`
`acivicin, diaziquone, bruceantin, mitoxantrone, DON, mitoguazone,
`
`l’CNU,
`
`large, because multiple endpoints are involved. Also,
`since current analog programs, especially for platinum
`compounds and anthracyclines, are undertaken to ob-
`viate major organ toxic effects, design of comparative
`trials must permit the simultaneous assessment of rela-
`tive toxicity and efficacy. This task may be particularly
`difficult if one is dealing with equivalent degrees of
`efficacy in the presence of decreased toxicity. In such a
`case, the existence of reduced toxicity is medically im-
`portant only if no material
`loss of efficacy has oc-
`curred. Clearly, accrual should be large enough that
`such a conclusion can be drawn with reasonable power.
`
`Another significant problem with orderly analog de-
`velopment is the current embarrassment of riches: ana-
`logs are much easier to synthesize than to test compre-
`hensively in the clinic. Accordingly, we now have a sit
`uation in which the number of worthy platinum com-
`pounds, anthracyclines, and antifols may actually exceed
`our capacity for rigorous and comprehensive comparative
`trials.
`
`An example of the NCI’s current strategy is the de-
`velopment program for two analogs of cisplatin, CHIP
`and carboplatin. The focus of their development is in
`cancers for which the parent compound is active, but
`not curative, such as carcinomas of the head and neck,
`uterine cervix, and bladder. In these disorders, random-
`ized trials to compare the two analogs are initiated im-
`mediately upon completion of phase I; the superior ana-
`log is then to be compared to the parent compound. The
`trial design includes provisions
`for early stopping
`based on absence of efficacy. Nonrandomized phase II
`trials are implemented in diseases which are less sensi-
`tive to cisplatin, such as non-small cell lung or colorec-
`tal cancer. The results of these trials will show whether
`the spectrum of activity with the analogs is extended
`significantly compared to the parent.
`
`It is also useful to consider within phase III the mul-
`titude of exploratory studies that go on with a new
`agent after conclusion of phase II. These studies in-
`clude:
`incorporation of the new drug into combina-
`
`tions, exploration of alternate routes of administration,
`and use of the agent in high doses accompanied by “res-
`cue” procedures. Such studies, particularly the combina-
`tion trials, constitute a large proportion of clinical
`research in cancer. It would certainly be churlish not to
`acknowledge the many advances that have come from
`such studies, including many of the crucial advances in
`medical oncology (2,3,5). Protection of
`the kidneys
`from platinum-induced damage (13) and current efforts
`with intracavitary chemotherapy (29) and intra-arterial
`chemotherapy (80) are other examples of “post-phase
`II” drug development that either have been proved val—
`uable already or appear particularly promising.
`However, it is equally clear that much of the work in
`phase III, particularly studies with drug combinations,
`has been disappointing in concept, design, or execution.
`Perhaps influenced by the success in certain leukemias
`and lymphomas,
`investigators have all
`too often at-
`tempted to hit
`the home run against
`tumors which
`have repeatedly proven refractory to easy solutions.
`The vast clinical literature in most solid tumors bears
`
`witness to the inefficacy of this approach, at least with
`combinations of drugs having individually low complete
`response rates. At the end of such uncontrolled “pilot”
`studies, one is almost always left with efficacy and
`toxicity data having no clear point of reference. The
`true pilot study does exist, of course, and is an inval-
`uable tool for testing the feasibility of an approach.
`Beyond the question of feasibility, however, a clinical
`trial is always comparative in intent, simply because
`there is always another way to treat the patient. There-
`fore, the issue of a suitable control group will never go
`away, no matter how difficult or inconvenient it is to
`deal with. Whether the control group should always (ie,
`wherever possible) be selected by a random process or
`Whether under certain circumstances nonrandomized
`controls can serve nearly as well is, in a sense, less im-
`portant than the somewhat more basic notion that the
`use of an explicit control group must be much more
`pervasive and sophisticated than is now common prac-
`tice.
`
`82
`
`Th is material was copied
`
`Cancer Treatment Reports
`
`

`

`Coordination of the Process
`
`For many years, the NCI has been the single largest
`contributor to all aspects of anticancer drug develop-
`ment. Because of its multifaceted involvement (from
`organic
`synthesis
`and natural product
`screening
`through phase III clinical trials) and the sheer volume
`of its support for the total effort toward drug develop-
`ment, the NCI has functioned, in effect, as the national
`coordinator of drug development
`in the US. In the
`past,
`the pharmaceutic industry has been less heavily
`involved in the development of antineoplastics than in
`other areas. Recently, however, with the growth of
`clinical oncology as a specialty and the far wider appli-
`cation of chemotherapy in the treatment of cancer, the
`industry has dramatically increased its interest
`in
`cancer.
`
`At the outset, one must recognize that the goals of
`the NCI and pharmaceutic industry share an important
`common feature:
`to define the activity of a new agent
`and make it available to patients as expeditiously as
`possible. It is only in this way that the population at
`large will benefit maximaHy from the fruits of
`research. However, one must also recognize the consid-
`erable differences
`in emphasis
`that
`the NCI and
`pharmaceutic industry often bring to this task. In its
`capacity as a financially disinterested supporter of
`basic and clinical research,
`the NCI wishes to see all
`reasonable steps taken to ensure that the therapeutic
`potential of a new agent is explored fully. By contrast,
`a pharmaceutic firm may be somewhat less inclined to
`support exploratory trials that have no direct role in
`securing an NDA for a new drug, and will strive to get
`a drug to market by the most direct and least costly
`route possible. Once a drug has been approved by the
`FDA,
`the company can deal with the question of
`whether to expand the indications in the postmarketing
`period or, as has been more often the case with anti-
`neoplastic agents, whether to let the evolution of subse-
`quent results work its own effects on the market.
`Obviously, the multiplicity of drug sponsors creates
`both opportunities and problems. The opportunities are
`clear enough: at a time when federal funds for re-
`search are increasingly difficult to obtain, support from
`private sources is sorely needed.
`In addition, many
`pharmaceutic firms now have, in both basic and clinical
`areas, excellent professional staff with the competence
`to formulate and supervise comprehensive drug devel-
`opment programs in cancer.
`The problems deal chiefly with the optimal utilization
`of scarce resources. In recent years, the pharmaceutic
`industry has relied heavily on the same pool of clinical
`investigators who are funded by the NCI via its peer-
`review mechanism, which obviously provides private
`sponsors with an important assurance of competence.
`On the other hand, industry-funded trials may possibly
`
`lead to difficulties in the peer-review process itself, for
`both scientific and financial reasons. At
`least some
`
`trials which are “NDA directed” may not appear very
`exciting or innovative when examined for the renewal
`of an institutional or cooperative group grant. The
`availability of financial support from industry presents
`investigators with the powerful incentives to perform
`trials which may be inherently uninteresting and un-
`likely to move the field forward. The existence of mul—
`tiple sources of private support in addition to federal
`funding creates the neceSSity for very careful account-
`ing of how funds are spent, to avoid the comingling of
`private and federal funds.
`The NCI has attempted to evolve constructive rela-
`tionships with the many pharmaceutic firms involved
`in clinical cancer research. Specifically,
`this involves
`much advanced joint planning to assure the orderly
`development of new agents of mutual
`interest
`in a
`manner which satisfies
`the needs of pharmaceutic
`industry for filing of an NDA with the FDA, while
`simultaneously preserving the autonomy of the clinical
`cooperative groups. As the chief source of funding for
`the cooperative groups