Cancer Chemother. Pharmacol. 4, 5--10 (1980) ancer hemotherapy and harmacology © Springer-Verlag 1980 General Reviews The Clinical Evaluation of Analogs - III. Anthraeyclines S. K. Carter Director, Northern California Cancer Program, 180l Page Mill Road, Bldg. B, Suite 200, Palo Ako, Calif. 94304, USA Adriamycin is one of the few examples in the short his- tory of cancer chemotherapy of an analog that is supe- rior to its parent compound [2, 4]. The parent com- pound in this case was daunomycin (daunorubicin), which had been shown to be active in acute leukemia but had also been shown to have a slim therapeutic index with severe hematologic toxicity that precluded its successful use against the more common solid tumors. Adriamycin has been shown to have a broader spectrum of activity and a superior therapeutic index than dau- nomycin. In the original clinical studies, daunomycin mainte- nance therapy had been shown to be associated with a fatal cardiomyopathy, which appeared to be dose-re- lated as it occurred after the total doses administered exceeded 20-30 mg/kg. As adriamycin's clinical activ- ity was observed, many responding patients were placed on maintenance treatment and again dose-related car- diomyopathy occurred, which limited the duration of successful maintenance that could be adminiistered. Be- cause the range of adriamycin's clinical activities was so much wider than that of daunomycin, the discussions of dose-limiting cardiomyopathy with adriamycin caused a flurry of investigation into the mechanism of the cardiac damage and into ways to predict it, evaluate it, and ame- liorate it. Analog development work increased in many countries. Therefore, analogs have entered clinical trial from sources as varied as the Soviet Union (carminomy- tin), Japan (aclacinomycin), Spain (quelamycin), the United States (AD-32), Italy (4' epi-adriamycin), and France (rubidazone and 14 DEA daunorubicin). Only some of these drugs have shown either significant supe- riority of antitumor activity in experimental tumors or diminished cardiac toxicity in animal model systems compared with adriamycin. Adriamycin has been tested clinically in the United States since 1969, and a massive number of patients have been treated with the drug alone or in combination with other chemotherapeutic agents. One of the most impressive aspects of the antitumor effect of adriamycin is its broad spectrum of activity (Table 1). In 14 tumor categories adriamycin has established anticancer activ- Table 1. Antitumor activity spectrum of adriamycin Established activity Possible activity Unresponsive Breast adenocarcinoma Soft-tissue and bone sarcomas Bladder adenocarcinoma Bronchogenic carcinoma Testicular carcinoma Prostatic adenocarcinoma Thyroid carcinoma Pediatric solid tumors Malignant lymphomas Acute leukemias Stomach adenocarcinoma Hepatoma Ovarian adenocarcinoma Endometrial carcinoma Squamous-cell carcinoma of cervix Squamous-cell carcinoma (head and neck) Multiple myeloma Pancreatic adenocarcinoma Large-bowel adenocarcinoma Malignant melanoma Renal cancer Malignant gliomas Squamous-cell carcinoma of the esophagus 0344-5704/80/0004/0005/$ 01.00
`
`Ex. 1037-0001
`
`

`
`ity. These encompass a broad range of solid tumors that in the past have been relatively insensitive to chemother- apy, especially the soft-tissue and bone sarcomas and bladder cancer. In four additional tumor categories the data available indicate some degree of adriamycin activ- ity, although a def'mitive statement cannot be made yet. This group also includes some classically unresponsive tumor types. The toxic effects of adriamycin are dose-related, pre- dictable, and for the most part reversible, The major toxicities are doseqimiting myelosuppression in approxi- mately 60%-80% of patients, stomatitis in as many as 80%, nausea and/or vomiting in 20%-55%, and alope- cia in virtually all cases. Leukopenia is the predominant hematologic manifestation of toxicity, and the severity depends on the adriamycin dose and the regenerative capacity of the bone marrow. Drug-induced stomatitis typically begins as a burn- ing sensation with erythema of the oral mucosa, and in 2-3 days it may produce frank ulceration, particularly in the sublingual and lateral tongue margins. Alopecia involving the scalp, axillary, and pubic hair occurs in almost all patients. Growth of hair usually resumes on cessation of the drug. Gastrointestinal toxicity evi- denced by nausea and occasional vomiting is associated with the drug, but rarely limits clinical use. Extravasa- tion during IV administration can produce local tissue necrosis, but normal precautions can prevent this toxic effect. Cardiac toxicity is the one harmful effect of adria- mycin that causes the greatest problem in long-term ad- ministration. This toxicity may involve transient electro- cardiogram (ECG) abnormalities, definitive cardiomyo- pathy, or both. The chronic toxicity of drug-induced cardiomyopathy produces both morbidity and mortality to a significant degree [7-9]. This 'pump' failure is dose-dependent, but shows no apparent relationship to pre-existing heart disease. The clinical presentation and pathophysiology of cardiac damage by adriamycin are indistinguishable from other known cardiomyopathies. Although the speed of the clinical course varies, it is usually a rapidly progressing syndrome of congestive heart failure and cardiorespiratory decompensation in- cluding dilation of the heart, pleural effusion, and ve- nous congestion. Reversibility of the heart failure does not appear to be a function of the therapeutic inven- tion. The overall incidence of congestive heart failure caused by drug-induced cardiomyopathy is 1%, al- though this is deceptive since the toxicity is related to the total dose administered, if the total dose is kept below 450 mg/m a, cardiomyopathy is rarely observed. Unfortunately, this limits the amount and duration of drug therapy. The frequency of cardiomyopathy is markedly increased at total doses above 550 mg/m 2, so S. K. Carter: The Clinical Evaluation of Analogs that a clinician who exceeds these dose levels must be aware of the high risk and balance it against the risk of discontinuing therapy in rapidly developing malig- nancy. Von Holt [10] has analyzed 4,018 patients treated with adriamycin in the United States cooperative groups between March 1970 and March 1977. A range of vari- ables were recorded for all patients. These totalled 67 and included age, sex, race, performance status, tumor type, prior cardiac disease, prior anticancer treatment, and concomitant treatment with other drugs. Ten spe- cific parameters were looked at in relation to the devel- opment of congestive heart failure (CHF) caused by adriamycin. These consisted of total dose and schedule of drug administration, concomitant chemotherapy, prior radiotherapy to the mediastinum, prior cardiac dis- ease, age, sex, race, type of tumor, and performance status. In this analysis, adriamycin-induced CHF occurred in 88 cases (2.2%). This was observed at intervals rang- ing from 0 to 231 days, with a median of 23 days, after the last administration of the drug. The mean and me- dian total dosages of the anthracycline received by the patient with CHF were 364 and 390 mg/m 2, respective- ly. Death Occurred within 70 days after the diagnosis of CHF in 63 of the 88 patients. In only 38 of the 63 was death attributed to the CHF. In the others it was attri- buted to progressive disease. In these latter 25 cases, the CHF was stable but unresolved in 12, partially resolved in 8, and totally resolved in 5. In this analysis, the total dose of adriamycin was related strongly to the development of CHF. The cumu- lative probability of developing drug-induced CHF was 0.3 at 400 mg/m 2, 0.7 at 550 mg/m z, and 0.18 at 700 mg/m z. When the schedule was examined, the weekly schedule had the lowest incidence of CHF at 0.8% (8/967), the incidence with a single dose every 3 weeks was 2.9% (66/2262), and the three consecutive daily doses repeated every 3 weeks involved an incidence of 2.5% (14/576). The importance of adriamyein as a clinically useful compound has led to a variety of approaches to improv- ing the therapeutic index by diminishing the cardiac tox- icity. These have included: (1) the search for new ana- logs; (2) schedule manipulation; (3) the use of cardiac toxicity blocking agents. A major strategic difficulty in clinical evaluation of any of these approaches is defining a feasible and appro- priate end-point for protocol studies. The discovery that adriamycin caused cardiomyopathy was empirical. As responding patients were maintained on the drug, the clinical picture was observed. Retrospective analysis determined that it was a dose-related phenomenon. The utilization of clinically manifest cardiomyopathy as a prospective end point is fraught with ethical and logistic
`
`Ex. 1037-0002
`
`

`
`S. K. Carter: The Clinical Evaluation of Analogs difficulties. Unfortunately no noninvasive diagnostic techniques reliable enough to be utilized in this manner have yet been discovered. The endomyocardial biopsy technique, initially described by Daniels et al. [1, 3] at Stanford, currently offers the only reliable test for pro- spective use. The cardiac biopsy data previously re- ported indicate that the pathologic insult to the heart is nearly universal when total doses in excess of 200-250 mg/m 2 are administered. This pathologic insult can be reproducibly graded after electron microscopy evalua- tion. It offers the possibility of measuring the cardiac damage occurring with an analog, new schedule, or blocking agent at a time when the pathologic insult is subclinical. It is therefore an ethically feasible ap- proach. There are four possible ways in which a new anthra- cycline analog of adriamycin could show superiority to its parent compound: 1. Increased efficacy in tumors responsive to adriamy- ein, as measured by (a) increased complete response rate; (b) increased overall response rate (CR + PR); 7 (c) increased duration of remission; (d) increased sur- vival. 2. Efficacy in tumors unresponsive to adriamycin, e.g., adenocarcinoma of the large bowel, malignant melano- ma, renal carcinoma. 3. Diminished acute toxicity as measured by (a) de- creased leukopenia and/or thrombocytopenia; (b) de- creased stomatitis; (c) decreased alopecia: (d) decreased gastrointestinal toxicity. 4. Diminished cardiac (chronic) toxicity as measured by (a) less cardiac functional damage, as measured by non- invasive techniques; (b) less pathologic damage obvious on endomyocardial biopsy; (c) diminished incidence of clinical cardiomyopathy. The clinical evaluation for a new anthracycline has to be tailored to fit which ever possibility of improving the therapeutic index is of highest priority. An attempt at a rough combination of the varying possibilities in an I New acute toxicity observed not seen with adriamycin, e.g., renal, pulmonary, CNS, etc. 1 Drop New anthracycline 1 Phase I study Acute toxicity pattern established at MTD I Acute toxicity spectrum similar to adriamycin I Phase II study I Acute toxicity spectrum less than with adriamycin I I Cardiac toxicity observed D!op I Ad ' ' namycm-responsive tumors I I No activity 1 Drop I No potential toxicologic superiority D!op I Activity potentially comparable to that of adriamycin I Activity potentially superior to that o I adriamycin Potential toxicologic ] superiority, e.g., card~ J Phase Ill study¢, Any activity I Adriamycin-unresponsive tumors I I No activity D!op Fig. 1. Hypothetical phase I and II decision flow for new anthracycline
`
`Ex. 1037-0003
`
`

`
`8 initial schema is shown in Fig. 1. This is just a skeleton, on which the actual criteria would need to be fleshed out in greater detail. Phase I Evaluation An appropriate initial schedule should be chosen on the basis of careful evaluation of the preclinical data. Among the preclinical data that should be evaluated are the following: a) Optimal schedule in the experimental tumors that are responsive to the drug b) Cell-cycle specificity or the lack of this c) Toxicology of the drug in rodents and large ani- mals d) Pharmacology of the drug in experimental ani- mals and correlation with all the above e) Appropriate biochemical data. An appropriate starting dose level should be chosen from analysis of the animal toxicology data. The most commonly used approach is to take a third of the toxic dose low (minimally toxic dose) of the most sensitive large animal species, worked out in milligrams per square meter of body surface area. The safety of this approach has been demonstrated in a wide range of drugs, including anthracyclines, by the retrospective analysis of Goldsmith et al. [6]. An appropriate dose escalation procedure should be designed, based on analysis of the preclinical data, in- eluding the steepness of the toxicologic dose-response curve. It has been shown by the members of the Nation- al Cancer Institute's Phase I working groups that, with all drugs analyzed to date, a completely nontoxic dose can safely be doubled. This will allow a more rapid es- calation than could be accomplished with a modified Fibinocci search scheme approach, as originally de- scribed by Selawry [10]. Doses should not be escalated in the same patient, so the potential for cumulative toxicity can be evaluated at each dose level, unless this is not considered in the best interests of the patient. Ideal- ly, pharmacologic determinations should be made at each dose level, and if it is deemed appropriate on the basis of the data obtained, the initial schedule can be modified. The end-points of the study would be as follows: a) The establishment of a maximally tolerated dose on the schedule tested and a recommended dose level for Phase II evaluation b) Elucidation of the acute toxicity pattern at all dose levels c) Elucidation of the pharmakokinetics of the drug in man, so as to make correlations with the pharmaco- kinetics in animals S. K. Carter: The Clinical Evaluation of Analogs d) Early indication of the cardiac toxic potential of the drug e) Early indication of the therapeutic efficacy of the drug, recognizing that patients who are placed on Phase I study often do not have measurable disease and may be extensively pretreated with other modalities and/or drugs. With a new anthracycline, there are four possibilities regarding acute toxicity in comparison to adriamycin. One possibility is that the analog will have a similar pattern, i.e., drug-limiting myelosuppression, stomatitis, alopecia, ulceration on extravasation, and mild gastroin- testinal side effects. A second possibility is that the ana- log may have a more favorable pattern, with some of the toxic manifestations less severe than those seen with adriamycin. A third possibility would be the observation of a new qualitative toxicity with the analog. One exam- ple of this is the jaw pain and postural hypotension ob- served with cyclocytidine and not seen with its parent drug arabinosyl cytosine. If a new anthracycline demon- strated renal toxicity, hepatic toxicity, neurologic toxic- ity, or pulmonary toxicity, this might be a sufficient rea- son to drop the drug after the phase I study. Any an- thracycline that demonstrated acute cardiac toxicity would obviously not be a prime candidate for further evaluation. The fourth possibility is some mixture of all of the above, and then the trade-offs would have to be carefully examined in the decision-making process. Phase II Evaluation Phase II evaluation becomes a disease-oriented process, and so with a cancer drug a variety of phase II strategies appropriate to the many diseases which constitute the cancer spectrum needs to be developed. With an analog, the evaluation must always involve a comparison with the known available data on the parent drug in each disease studied. In terms of efficacy, there are three pos- sibilities for an analog that can be tested for in a phase II design. The first is that the analog will be more active against a responsive tumor. The second is that the ana- log will have activity in a tumor that is unresponsive to the parent drug. The third possibility is that the analog will not be cross-resistant with the parent drug. In this last case, activity with the analog would be demon- strated after progressive disease was seen on the parent drug. Vincristine and vinblastine are examples of close structural analogs that are not cross-resistant. The phase II strategy for a new anthracycline would very much depend on which possibility was being evaluated. If increased efficacy were being sought, then a clinical situation would have to be developed where the analog- would be given as a single agent without prior exposure
`
`Ex. 1037-0004
`
`

`
`s. K. Carter: The Clinical Evaluation of Analogs tO the parent structure. Where the parent structure was part of a highly active combination, this would create strategic difficulties. The end-point of such a study would be a response rate that would have to be com- pared to the response rate of adriamycin in the same clinical situation. It would be unfair to compare the re- sponse rate of an analog used as tertiary drug treatment in breast cancer to that of adriamycin used in previously untreated patients. For each adriamycin-responsive tumor it will be necessary to determine (1) when an ana- log should have phase II evaluation; and (2) what re- sponse rate would indicate (a) little possibility of supe- riority to adriamycin in phase III; (b) probable compa- rable activities to adriamycin; or (c) probable superior activity to adriamycin. The other two possibilities for an analog are much more simple to test for: if the desired end-point is activ- ity in an adriamycin-unresponsive tumor, then a phase II study can be undertaken in previously un- treated colorectal cancer, renal cancer, and malignant melanoma. Any level of meaningful activity would indi- cate the advisability of phase III trials. If the possible analog effect is lack of cross-resistance, the strategy just mentioned also includes this concept. Another approach is to take tumors responsive to adriamycin, after relapse on the drug, and to utilize the analog. Again, any mean- ingful response rate would be a positive result. Phase III trials for new analogs have to be designed according to the effect being sought. One effect would be increased efficacy at a comparable level of toxicity. The second would be diminished cardiac toxicity at least at a comparable level of efficacy. The first effect re- quires the usual phase III approaches utilized with cyto- toxic chemotherapy. Ideally this would involve some controlled comparison of the analog, either alone or in combination, with adriamycin. The potential for dimin- ished cardiac toxicity requires a more innovative ap- proach, which should utilize the endomyocardial biopsy technique. With an anthracycline analog, the crucial end-point will not be the total dose of drug alone, but the total Table 3. Hypothetical end-points of phase II studies Toxicity a t Efficacy b Acute t Maybe No No Cardiac t Acute t Yes Maybe No Cardiac -- Acute t Yes Yes Yes C ardiac'~ Acute "~ Maybe No No Cardiac t Acute "~ Yes Yes Maybe Cardiac - Acute x~ Yes Maybe No Cardiac t Acute - Maybe No No Cardiac t Acute - Yes Maybe No Cardiac - Acute - Yes Yes Maybe Cardiac t a Toxicity Evidence that toxicity could be less than observed for adriamy- cin - Evidence that toxicity is comparable to that of adriamycin or there are not enough data to allow a meaningful decision t Evidence that toxicity could be more pronounced than observed for adriamycin b Efficacy Evidence that activity is superior to that observed with adriamy- cin - Evidence that activity is comparable Evidence that activity is less pronounced than that observed with adriamycin number of courses that can be administered. The new American drug AD-32 has a recommended dose sched- ule for phase II trials of 600 mg/m z every 3 weeks. If cardiotoxicity were observed at a total dose of 1800 mg/m 2, it could be said that AD-32 is less toxic to the heart than adriamycin, since more than three times the Table 2. Cardiac toxin comparisons possible between adriamycin and analog Adriamycin Analog No. of Total dose No. of Possible biopsy courses (mg/m 2) courses changes sought 4 240 4 8 480 8 12 720 12 16 960 16 > 50% with normal biopsies > 25% with normal biopsies < 75% with 3+ changes _< 50% with 3+ changes
`
`Ex. 1037-0005
`
`

`
`10 TaMe 4. Adriamycin analogs for which clinical data are available Drug Country of origin Daunomycin Italy 4'-epiAdriamycin Italy Rubidazone France Dubiromycin France 14-DEA Daunomycin France Quelamycin Spain Carminomycin USSR Adriamycin-DNA Belgium Daunomycin-DNA Belgium Aclacinomycin-A Japan AD-32 USA dose can be administered before cardiac damage is ob- served. It should be obvious, however, that being able to give only three courses makes this treatment distinctly inferior to the treatment with nine courses of adriamyein which can be administered before a total dose of 540 mg/m 2 is reached. The design strategy with the biopsy, for an analog, could have two aspects (Table 2). One ap- proach would be to biopsy at a fixed number of courses and compare the biopsy changes to those seen with adriamycin at a comparable number of courses. For example, biopsies could be taken after four or five tri- weekly courses of the analog, which would be equivalent to a total dose of 240 or 300 mg/m 2 of the parent com- pound. A second approach would be to do serial biop- sies after 4, 8, 12, and 16 courses of the new analog, if feasible. With at least three such points, a curve could be calculated that expressed biopsy damage as a manifesta- tion of the total dose administered. The slope of this curve could be compared with the slopes of similar curves developed for adriamycin. When all the data are finally analyzed, there are 27 hypothetical end-points for each individual tumor type, based on the possibilities of efficacy, acute toxicity, and cardiotoxicity (Table 3). If ten tumor types were clini- cally evaluated this would make 270 potential clinical end-points that could be observed. The reality of oneologie research today is that there are more ideas that could be clinically tested than avail- able clinical research resources. This is particularly true s. K. Carter: The Clinical Evaluation of Analogs with the anthracycline antitumor antibiotics. There are more than ten analogs of adriamycin in clinical trial today throughout the world (Table 4), and many more analogs are on the drawing board and in various stages of preclinical work-up. Possibilities exist for schedule manipulation studies, and several approaches have been postulated for blocking the toxicity. Unless some type of accepted clinical trial strategy is developed, chaos could result. There is a danger that the literature will become glutted with clinical results that fail to answer the critical comparative questions that have to be asked. An even more disturbing possibility is that the cancer patient, who can benefit from anthracycline therapy will fail to receive the maximal benefit possible from these drugs. References 1, Billingham, M. E., Bristow, M. G., Glatstein, E., et al.: Adriamy- cin cardiotoxicity: Endomyocardial biopsy of enhancement by irradiation. Am. J. Surg. Pathol. 1, 17-23 (1977) 2. Blum, R. M., Carter, S. K.: Adriamycin. A new anticancer drug with significant clinical activity. Ann. Intern. Med. 80, 249-259 (1974) 3. Bristow, M. R., Masur, J. W., BiUingham, M. E., et al.: Doxoru- bicin cardiomyopathy: Evaluation by phonocardiography, en- domyocardial biopsy and cardiac catheterization. Ann. Intern. Med. 88, 168-175 (1978) 4. Carter, S. K.: Adriamycin - A review. J. Natl. Cancer Inst. 55 (6), 1265-1274 (1975) 5. Carter, S. K., Selawry, O. S,, Slavik, M.: Phase I clinical trials. Natl. Cancer Inst. Monogr. 45, 75-81 (1977) 6. Goldsmith, M., Slavik, M., Carter, S. K.: Quantitative prediction of and toxicity in humans from toxicology in small and large animals. Cancer Res. 35, 1354 (1975) 7. Lenaz, L., Page, J. A.: Cardiotoxicity of adriamycin and related anthracyclines. Cancer Treat. Rev. 3, 111-120 (1976) 8. Minow, R. A., Benjamin, R. S., Gottlieb, J. A.: Adriamycin (NSG 123127) cardiomyopathy: An overview with determining of risk factor. Cancer Chemother. Rep. 6 (2), 195-201 (1975a) 9. Minow, R. A., Benjamin, R. S., Lee, E. T.: Adriamycin cardio- myopathy: An overview with determining of risk factor. Cancer Chemother. Rep. 6 (2), 195--201 (1975b) 10. Von Hoff, D. D., Layard, D. D., Basa, P., et al.: Risk factors for adriamycin-induced congestive heart failure (in press, 1979) Received June 29/Accepted July 31, 1979
`
`Ex. 1037-0006