COMMENTARY Phase II Trial Design Considerations for Small-Cell Lung Cancer Timothy D. Moore, Edward L. Korn* investigational therapies end up being tested in a potentially unfavorable setting. This observation is supported by the fact that two of the most active agents used for the initial treatment of this disease (etoposide and carboplatin) give response rates under 20% in previously treated patients (14) compared with 5O%-80% in patients who have not previously had chemotherapy (i.e., chemotherapy-naive patients) (75). Recent similar experiences with new promising agents such as ifosfamide and teniposide (75) have provided an impetus to explore alternative trial designs.
Possible Trial Designs History Small-cell lung cancer has been shown to be biologically distinct from other histologic subtypes of lung cancer in both clinical and laboratory studies conducted since the 1960s (5,6). Untreated, it exhibits a very aggressive natural history; patients presenting with extensive disease have a median survival of only 6 weeks (7). This relentless clinical course is consistent with the results of thymidine-labeling studies, which demonstrate a high proliferative index (8). In addition, autopsy evidence indicates that the disease is frequently disseminated even in patients whose disease clinically appears to be confined to the chest (9). Fortunately, small-cell lung cancer was shown in many early studies of cytotoxic therapy to be the most chemosensitive type of lung cancer (7,10). By 1977, 15 agents had demonstrated activity (objective response rate of >20%) in this disease (10). Subsequently, combination chemotherapy (consisting of three or four drugs) was found to be superior to single agents or twodrug combinations in randomized and nonrandomized trials (7012). By 1980, combination chemotherapy was considered to be the standard treatment (75). While this strategy results in responses of meaningful duration in the majority of patients, it remains uncommon for a patient with extensive disease to survive beyond 2 years (73). In addition, with the acceptance of polychemotherapy as standard treatment, investigational drugs have not commonly been administered until a patient has been exposed to several cytotoxic agents having different mechanisms of action. As a result,
150
Designs for Pretreated Patients Lower the target response rate to 10%. Conventional phase II trials in previously treated small-cell lung cancer patients, as well as in patients with other relatively chemoresistant tumors, usually are designed to differentiate between a drug having a 20% level of activity and one having a 5% level of activity. Specifically, the design of these trials is such that an agent with a true 20% level of activity will be declared uninteresting no more than 10% of the time (i.e., a false-negative rate ^10%) and that an agent with a true 5% level of activity will be declared interesting no more than 10% of the time (i.e., a false-positive rate ^10%). Designs that both satisfy these criteria and accrue patients in two stages are popular because they can be stopped early if it is unlikely that the agent is truly active (76); see design A in Table 1 for an example. Suppose the threshold for activity of interest is lowered to 10%, as has been suggested (2), in order to avoid missing active
Received June 26, 1991; revised September 24, 1991; accepted October 22, 1991. T. t>. Moore (Clinical Investigations Branch), E. L. Korn (Biometrics Research Branch), Cancer Therapy Evaluation Program, Division of Cancer Treatment, National Cancer Institute, Bethesda, Md. Present address: T. D. Moore, M.D., Medical Oncology Associates, 3471 Fifth Ave., Pittsburgh, PA 15213. 'Correspondence to: Edward L. Kom, Ph.D., Biometrics Research Branch, Cancer Therapy Evaluation Program, National Cancer Institute, National Institutes of Health, EPN-739, Bethesda, MD 20892.
Journal of the National Cancer Institute
Downloaded from http://jnci.oxfordjournals.org/ at Karolinska Institutet on May 25, 2015
Therapeutic options for the systemic treatment of small-cell lung cancer have not improved substantially in the past 15 years. Recognition of effective new chemotherapeutic agents is crucial, since this cancer is generally considered to be a disseminated disease at presentation. Recently, several publications have implied that many phase II studies have not been designed to evaluate fairly an investigational agent's efficacy in the treatment of this disease (1-4). Several different strategies have been proposed for testing new agents. This commentary critiques these options and presents recommendations for the design of future phase II investigational drug trials in small-cell lung cancer.
Table 1. Possible two-stage designs*
Design
No. of patients treated at stage 1 2
A
15
Add 20
B
29
Add 181
C
29
Add 41
D
29
Add 21
Declare drug interesting iff S4/35 responses £15/210 responses £4/70 responses 22/50 responses
1%
.005 (.25) .076 (.25)
2%
True response ratet 3% 5%
.047 (.44) .23 (.44)
.09 (.54) .10 (.77) .43 (.77) .94 (.95)
.14 (.59) .39 (.59)
10%
20% .92 (96)
.90 (.95) .90 (.95)
*In each design, we proceed to stage 2 if there is at least one response seen in stage 1. tResults are expressed as No. of responses/total No. of patients treated. ^Numbers are probabilities of declaring the drug interesting. Numbers in parentheses are the probabilities a second stage will be required.
Since the desire here is to determine a level of disinterest, only the inactive and marginally active drugs will be considered. Depending on which of these agents is viewed as truly inactive, the overall observed response rate for inactive agents could be as low as 0 of 573 or as high as 16 of 912 (1.8%). The corresponding upper 95% confidence limits for the true response rate are 0.5% and 2.7%. These observed response rates and confidence limits suggest that the setting of a level of disinterest of 2% and the use of design C in Table 1 would probably yield an acceptable false-positive rate. However, even if these limitations are accepted, 44% of the time a trial of a new drug with a true response rate of 2% would not be terminated at 29 patients but would require 70 patients. In addition to setting the level of disinterest, another important issue is the appropriateness of treating patients when only a 10% response rate is targeted. It is questionable whether such Vol. 84, No. 3, February 5,1992
critically ill patients should be subjected to potentially toxic chemotherapy when there is such a low expectation of response even to active investigational agents. This observation is true even though these patients would not respond well to standard salvage chemotherapy. At a minimum, a frank discussion with the patient of the limited goals of the trial is essential. Use of relapsed patients, keeping the target rate at 20%. Phase n studies of pretreated patients are usually composed of a heterogeneous population. Some patients have refractory disease in which the tumor grows, or is at best stable, on induction therapy. However, other patients respond to treatment and then relapse after a period of freedom from chemotherapeutic treatments. This distinction has important consequences, both in small-cell and other malignancies, because a higher proportion of relapsed patients may respond to second-line treatment than those whose disease is truly refractory to chemotherapy (21-23). These studies indicate that relapsed patients may respond not only to second-line regimens, but also to cytotoxic agents to which they were previously exposed. In selected relapsed smallcell lung cancer patients, therefore, it may be reasonable to have higher expectations for systemic treatments than in the relapsed, refractory patient population taken as a whole. This statement translates into the ability to realistically target a higher response rate (e.g., 20%) without being overly concerned that an active drug will be missed. This statement, in turn, implies that a smaller sample size is required (e.g., design A). Three uncertainties exist with this approach. Considering the appreciable response rates observed in relapsed patients with reinduction-salvage regimens, physicians might be reluctant to advocate investigational treatments, despite the relatively short remissions observed with exposure to previously active drugs. Secondly, the general feasibility of this approach regarding the number of available patients and their willingness to participate in clinical investigations at this stage in their treatment is unknown, although certain examples exist of successful studies carried out in this manner (2425). In particular, long-term toxic effects of the initial therapy overlapping with the study drug may restrict the numbers of patients eligible. Finally, it is unclear what the minimal treatment-free interval should be before a patient is considered a good risk to be re-treated with conventional cytotoxic chemotherapy. This question has not been proCOMMENTARY 151
Downloaded from http://jnci.oxfordjournals.org/ at Karolinska Institutet on May 25, 2015
agents that may not achieve a 20% response rate in this pretreated population. If the level of disinterest remains at 5%, a large number of patients will be required, e.g., design B in Table 1. The large number of patients is required to keep the false-positive rate low, i.e., to keep from declaring an agent interesting when the true response rate is 5%. If one lowers the level of disinterest from 5% to 2%, then design C in Table 1 shows that 70 patients would be sufficient to ensure that the false-positive rate is less than 10%. If a 1% level of disinterest were appropriate, then 50 patients would be sufficient (design D in Table 1). Determining the level of disinterest for small-cell lung cancer is not a trivial task. One consideration is the measurement error inherent in determining whether a patient has had a response (77). Another consideration is the occasional true response seen with inactive agents. Theoretically, both of these factors are taken into account by averaging the observed response rates from a number of trials of inactive agents. However, in practice, it is not known which agents are truly inactive; we have only the observed response rates by which to be guided. This difficulty is illustrated if one considers the observed response rates of previously treated small-cell lung cancer patients from singleagent, phase II trials. Table 2 contains the published results of such trials obtained through a MEDLINE search, an issue of Cancer Treatment Symposia (78), and three review articles {15,1920). (A complete list of references is available from the authors.)
Table 2. Response rates in phase II trials of previously treated small-cell lung cancer patients* No. of patients responding/total No. of patients
%
Drug
9/72 27/159 4/14 42/388 3/26 6/41 18/117 5/24 17/100
13 17 29 11 12 15 15 21 17
Carboplatin Cisplaiin Doxorubicin Etoposide Fluorouracil Ifosfamide Teniposide Vincristine Vindesine
"Marginally active drug" 9
3 4 7 3 10 4 2 3 6
ACNUt Cytarabine Eflomithine Epirubicin Iproplatin Lonidamine N-Methylformamide Mitoguazone Mitoxantrone PCNUt
"Inactive drug" 0/29 0/127 0/17 0/18 0/10 0/43 0/11 0/15 0/23 0/15 0/17 0/30 0/19 0/21 0/24 0/29 0/41 0/20 0/14 0/34 0/16
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Aclarubicin Amsacrine Bisantrene Chlorozotocin Dianhydrogalactitol Diaziquone Fludaratrine phosphate Interferon-a Maytansine 6-Mercaptopurine Methotrexate Mitomycin-C PALA§ Prednimustine Procarbazine Spirogermanium Streptozocin TGUH Tiazofurin Vinblastine Zinostatin
*Only dmgs that were evaluated in at least 10 patients were included. They have been classified into three groups based on their response rates: "active" for response rates (complete response or partial response) >10%, "marginally active" for agents with 30%) can be required for a drug to be considered worthy of further testing, since a patient population with a relatively favorable risk is studied (28). A major concern with this strategy in the treatment of smallcell lung cancer relates to the biologic aggressiveness of the disease. Reluctance on the part of the investigator to advocate initial treatment with an untested therapy centers on the fear that delaying effective multidrug chemotherapy will result in an appreciable decrease in survival (331). This reluctance may be mitigated to some extent by switching to salvage therapy after one cycle if the patient has not responded in some well-defined manner. For example, evidence (32) suggests that absence of a decrease in tumor size is predictive that the patient will not achieve at least a partial response in subsequent cycles. Further research would be helpful for evaluating strategies that predict, after one cycle of therapy, whether or not a patient will achieve at least a partial response. The apprehension of delaying effective therapy is most relevant for trials in which a drug later proven to be inactive is tested. Survival inferior to historical controls has been observed when an inactive agent (oral idarubicin) was tested using the window-of-opportunity design (57). The Eastern Cooperative Oncology Group recently reported results of a 90-patient, randomized trial in which patients receiving an inactive drug (menogaril), with cisplatin and etoposide salvage (EP) for patients having progressive or stable disease, were compared with those receiving standard therapy consisting of cyclophosphamide, doxorubicin, and vincristine (CAV) from the outset (29). No adverse effect on survival was claimed for menogaril-EP (median survival, 38.3 weeks), compared with the CAV arm (median survival, 41.7 weeks), despite the fact that only two partial responses (5%) were observed in patients receiving menogaril. It is questionable whether that study has adequately proven that the window-of-opportunity approach is equivalent to standard therapy when an inactive drug is tested. Although the observed ratio of median survivals (92% = 38.3/41.7) is close to 100%, one cannot be very confident of this estimate because the sample size is limited. For example, even assuming that all the patients have been followed to their death, an approximate 90% confidence interval of the ratio of median survivals would be broad: 64%-131%. While it might be argued that wide confidence intervals are acceptable, considering the dismal prognosis for patients with extensive-stage small-cell lung cancer, it is doubtful whether the risk of a 36% decrement in median survival would be considered acceptable by many investigators. An equivalency trial would be required in order to prove rigorously that an unacceptable decrement in survival does not Journal of the National Cancer Institute
Downloaded from http://jnci.oxfordjournals.org/ at Karolinska Institutet on May 25, 2015
3/34 1/36 1/23 1/14 1/34 3/30 1/25 1/46 2/63 2/34
Designs for Patients Without Prior Exposure to Cytotoxic Therapy
Limit to elderly patients or patients with very poor prognoses. It has been suggested that elderly patients or patients with especially poor prognoses (e.g., patients with two or more metastatic sites or bone marrow metastases) can be treated with investigational agents without prior treatment (34-37). The rationale behind this suggestion appears to be that these patients may not be able to tolerate combination chemotherapy or may expect relatively fewer benefits from this aggressive approach. This strategy has been used to both support (34,36) and rule out (55) a drug's efficacy in the treatment of this disease.
Table 3. Sample sizes per group required to have 90% power for detecting the alternative that the hazard ratio is as specified*
Hazard ratio 0.50 0.70 0.80 0.90 *a = .05, one-sided test.
Vol. 84, No. 3, February 5, 1992
Required sample size per group 36 135
345 1540
Some problems with this approach, however, can be highlighted by asking two questions: 1) Is the response rate to new agents expected to be similar in this group of patients to that in small-cell lung cancer patients at large? 2) Is there an effective way to treat these patients with standard agents? To arrive at an overall appraisal, we need to consider individually the two groups defined above. For the group of elderly patients, limited experience with carboplatin (34) and teniposide (36) suggests that the response rates may be similar to those expected in nonelderly patients, which answers question 1. However, with regard to question 2, there are ways to treat these patients with standard agents that will result in meaningful response rates. Even if combination therapy is not advisable because of coexisting medical conditions, one of the established drugs used in this disease would likely have an acceptable toxicity profile that would allow its use as a single agent. Thus, with these criteria, elderly patients appear similar to untreated patients in general. Therefore, it is not clear why they would not be subject to the same concerns mentioned in the previous section. The other category of patients presumed to have low expectations of benefiting from standard chemotherapy consists of those patients with poor prognostic features, such as a poor performance status or bone marrow metastases. Although these patients will occasionally respond to standard therapy (38), the benefits of such therapy are very limited (59), thus answering question 2. However, is the response rate to new agents expected to be similar in this group to that in small-cell lung cancer patients at large? We would intuitively expect these response rates to be lower. In fact, such has been the case in a group of patients with poor performance status treated with etoposide (40). The problem in trial design for these patients, therefore, is no different from that in trial design for previously treated patients in which lower response rates are expected (see section "Lower the target response rate to 10%")- An additional problem in treating patients with poor performance status is that they may be subject to excessive toxic effects (40,41).
Recommendations We suggest the following guidelines for the study of investigational drugs in the treatment of small-cell lung cancer. We recommend that a phase II design, targeting at least a 30% response rate, should be used for previously untreated patients, for the following classes of agents: (a) Agents having novel or unknown mechanisms of activity and which show clinical activity in other solid tumors (e.g., taxol). (b) Agents showing specific activity for small-cell lung cancer in preclinical drug screens. (c) Analogues of active drugs currently used in this disease, which are being developed primarily because of a different toxicity profile (e.g., carboplatin). These studies should be limited to patients with extensivestage disease, Kamofsky performance status of 70% or more, no superior vena cava syndrome, and no significant visceral organ COMMENTARY 153
Downloaded from http://jnci.oxfordjournals.org/ at Karolinska Institutet on May 25, 2015
occur. This hypothetical trial would compare a series of drugs, later proven to be inactive, given by the window-of-opportunity method, with conventional therapy. Depending on the hazard ratio (i.e., the ratio of median survival times) that would be considered tolerable, however, large sample sizes could be required (Table 3). A modification of the window-of-opportunity approach has been described in which the overall response rate of experimental treatment followed by standard treatment is compared with the historical response rate of standard therapy (4,33). Using this design, the trial is stopped early if the response rate to the sequential therapy is too low. For example, taking 67% to be the expected response rate to standard therapy for extensive smallcell lung cancer (4), we might stop the trial of the sequential treatment if five or more nonresponders were observed out of the first seven patients treated. There are three problems with this approach. The first is that this approach offers only very limited protection for the patient. Continuing with the above example, if the true response rate for the sequential therapy is 40%, then the probability of getting five or more nonresponders out of the first seven patients treated and stopping early is only 42%. Furthermore, even with equal response rates, there is a potential for a decrement in median survival. Secondly, this approach does not adequately address the question of whether an investigational drug has activity against small-cell lung cancer. For instance, with the use of the above example, it is possible for a response to the investigational drug to be observed and still have the study terminated because of a lack of drug activity. Finally, this approach assumes that the true response rate to standard therapy is known for the patients about to be treated with the sequential therapy. With the heterogeneity of patients entered in phase II trials, this response rate may be difficult to estimate. Therefore, this modification of the window-of-opportunity approach is not satisfactory.
References (/) ETTINGER DS: Evaluation of new drugs in untreated patients with smallcell lung cancer Its time has come. J Clin Oncol 8:374-377, 1990 (2) KRBTJANSEN PEG, HANSEN HH: Management of small cell lung cancer A summary of the Third International Association for the Study of Lung Cancer Workshop on Small Cell Lung Cancer. J Natl Cancer Inst 82:263-266, 1990 (3) AISNER J: Identification of new drugs in small cell lung cancer: Phase II agents first? Cancer Treat Rep 71:1131-1133, 1987 (4) CULLEN M: The design of phase II trials. Lung Cancer 5:214-220, 1989 (5) COHEN MH, MATTHEWS MJ: Small cell bronchogenic carcinoma: A distinct clinicopathologic entity. Semin Oncol 5:234-243, 1978
sessment of response to anticancer chemotherapy: Proposal for new criteria of tumor response. J Clin Oncol 2:1040-1046, 1984 (18) WrrTES RE, ED: Cancer Treatment Symposia, vol 4. Washington, DC: US Govt Print Off, 1985 (19) Joss RA, CAVALLJ F, GOLDHIRSCH A, ET AL: New drugs in small-cell lung cancer. Cancer Treat Rev 13:157-176, 1986 (20) AISNER J, ABRAMS J: Cisplatin for small-cell lung cancer. Semin Oncol 16:2-9,1989 (21) BATIST G, IHDE DC, ZABELL A, ET AL: Small-cell carcinoma of lung:
Reinduction therapy after late relapse. Ann Intern Med 98:472-474, 1983 (22) GIACONNE G, DONADIO M, BONARDI G, ET AL: Teniposide in the treatment
of small-cell lung cancer The influence of prior chemotherapy. J Clin Oncol 6:1264-1270,1988 (23) JOHNSON DH, STRUPP J, GRECO FA, ETAL: Prolonged administration of oral
etoposide in previously treated small cell lung cancer (SCLC) patients: A phase II trial. Proc ASCO 9:227, 1990 (24) GIACONNE G, FERRATI P, DONADIO M, ET AL: Reinduction chemotherapy in
small cell lung cancer. Eur J Clin Oncol 23:1697-1699, 1987 (25) POSTMUS PE, BERENDSEN HH, VAN ZANDWUK N, ET AL: Retreatment with
the induction regimen in small cell lung cancer relapsing after an initial response to short term chemotherapy. Eur J Cancer Clin Oncol 23:14091411,1987 (26) GIACCONE G: Second line chemotherapy in small cell lung cancer. Lung Cancer5:207-213, 1989 (27) BLACKSTEIN M, EISENHAUER EA, WIERZBICKI, ET AL: Epirubicin in exten-
sive small-cell lung cancer: A phase II study in previously untreated patients: A National Cancer Institute of Canada Clinical Trials Group Study. J Clin Oncol 8:385-389, 1990 (28) EVANS WK, EISENHAUER EA, CORMIER Y, ETAL: Phase II study of amona-
fide: Results of treatment and lessons learned from the study of an investigational agent in previously untreated patients with extensive small-cell lung cancer. J Clin Oncol 8:390-395, 1990 (29) ETTINGER DS, FINKELSTEIN DM, ABELOFF MD, ET AL: Justification for
evaluating new anti-cancer drugs in selected untreated patients with a chemotherapy-sensitive advanced cancer: An ECOG randomized study. Proc ASCO 9:224, 1990 (30) SMITH IE: New treatments for small cell lung cancer. When to test. Chest 96:59S-61S, 1989 (31) CULLEN MH, SMITH SR, BENFIELD GFA, ET AL: Testing new drugs in un-
treated small cell lung cancer may prejudice the results of standard treatment: A phase II study of oral idarubicin in extensive disease. Cancer Treat Rep 71:1227-1230, 1987 (32) WOLF M, HAVEMANN K, HOLLE R, ET AL: The prognostic value of response
to the first cycle of chemotherapy in small cell lung cancer. Results of a multicenter German trial. Eur J Cancer Clin Oncol 23:1197-1205, 1987 (33) CULLEN MH, HILTON C, STUART NSA: Evaluating new drugs as first treatment in patients with small-cell carcinoma: Guidelines for an ethical approach with implications for other chemotherapy-sensitive tumors. J Clin Oncol 6:1356-1357,1988
(6) CARNEY DN, BRODER L, EDELSTEIN M, ET AL: Experimental studies of the
(34) SMITH IE, HARLAND SJ, ROBINSON BA, ET AL: Carboplatin: A very active
biology of human small cell lung cancer. Cancer Treat Rep 67:27-35,1983
new cisplatin analog in the treatment of small cell lung cancer. Cancer Treat Rep 69:43-46, 1985
(7) GREEN RA, HUMPHREY E, CLOSE H, ETAL: Alkylating agents in bronchoge-
nic carcinoma. Am J Med 46:516-525, 1969 (8) SHACKNEY SE, STRAUS MJ, BUNN PA JR: The growth characteristics of
small cell carcinoma of the lung. In Small Cell Lung Cancer (Greco AF, Oldham RK, Bunn PA Jr, eds). New York: Grune, 1981, pp 225-234 (9) MATTHEWS MJ, KANHOUWA S, PICKREN J, ET AL: Frequency of residual and
metastatic tumor in patients undergoing curative surgical resection for lung cancer. Cancer Chemother Rep 4:63-67, 1973 (10) BRODER LE, COHEN MH, SELAWRY OS: Treatment of bronchogenic carci-
noma. II. Small cell. Cancer Treat Rev 4:219-260, 1977 (//) BUNN PA, COHEN MH, IHDE DC, ETAL: Advances in small cell bronchoge-
(12) (75) (14) (15) (16)
154
(17) WARR D, MCKJNNEY S, TANNOCK I: Influence of measurement error on as-
nic carcinoma. Cancer Treat Rep 61:333-342, 1977 GRECO FA, EINHORN LH, RICHARDSON RL, ET AL: Small cell lung cancer Progress and perspectives. Semin Oncol 5:323-335, 1978 MORSTYN G, IHDE DC, LICHTER AS, ET AL: Small cell lung cancer 19731983: Early progress and recent obstacles. Int J Radial Oncol Biol Phys 10:515-539,1984 ETNHORN LH: Cisplatin plus VP-16 in small cell lung cancer. Semin Oncol 13:3-4,1986 JOHNSON DH: New drugs in the management of small cell lung cancer. LungCancer5:221-23l, 1989 FLEMING TR: One-sample multiple testing procedures for phase II trials. Biometrics 38:143-151, 1982
(35) LUND B, HANSEN F, HANSEN M, ET AL: Phase II study of 1,2,4-
triglycidylurazol (TGU) in previously untreated and treated patients with small cell lung cancer. Eur J Cancer Clin Oncol 23:1031-1033, 1987 (36) BORK E, HANSEN M, DOMBERNOWSKY P, ET AL: Teniposide (VM-26), an
overlooked highly active agent in small-cell lung cancer. Results of a phase II trial in untreated patients. J Clin Oncol 4:524-527, 1986 (37) AISNER J, ALBERTO P, BITRAN J, ETAL: Role of chemotherapy in small cell
lung cancer: A consensus report of the International Association for the Study of Lung Cancer Workshop. Cancer Treat Rep 67:37-43, 1983 (38) IHDE DC, MAXUCH RW, CARNEY DN, ET AL: Prognostic implications of
stage of disease and sites of metastases in patients with small cell carcinoma of the lung treated with intensive combination chemotherapy. Am Rev RespirDis 123:500-507, 1981 (39) OSTERLAND K: Prognostic factors in small cell lung cancer An analysis of 874 consecutive patients. In Lung Cancer Basic and Clinical Aspects (Hansen HH, ed). Boston: Martinus-Nijhoff, 1986, pp 139-152 (40) POSTMUS PE, SMTT EF, BERENDSEN HH, ET AL: A feasibility study of test-
ing new drugs for small-cell lung cancer in patients with a poor performance status. Cancer Chemother Pharmacol 27:490-491,1991 (41) CERNY T, PEDRAZZINI A, JOSS RA, ET AL: Unexpected high toxicity in a
phase II study of teniposide (VM-26) in elderly patients with untreated small cell lung cancer (SCLQ. Eur J Clin Oncol 24:1791-1794, 1988
Journal of the National Cancer Institute
Downloaded from http://jnci.oxfordjournals.org/ at Karolinska Institutet on May 25, 2015
dysfunction. Unless there is prior evidence that multiple courses will be required to observe a response, the window-of-opportunity strategy should be used. In addition, every effort should be made to choose a salvage regimen that has as few as possible toxic effects in common with the investigational agent. For agents not in drug classifications a-c, we recommend that they be studied in previously treated patients who have had disease recurrence after a disease-free and treatment-free interval sufficient to exclude those patients having predominantly cytotoxic agent-refractory disease. These agents should be tested using a conventional phase II design targeting a 20% response rate. We recommend a 3-month treatment-free interval after a complete response, although further research validating this time frame would be helpful. These recommendations are meant to provide guidelines for the initial evaluation of investigational drugs in small-cell lung cancer. Additionally, we are not addressing the development of analogues that are being investigated primarily because of crossresistant properties to their parent compound, e.g., tetraplatin. These agents would be first tested in pretreated patients. To place these guidelines in their proper perspective, it is essential that one does not equate clinical investigation with clinical care. It has not been the intention of this commentary to suggest therapy for patients ineligible for conventional investigational drug trials. Rather it is hoped that focusing investigational drug development in the manner described will result in improved treatment options for all small-cell lung cancer patients.
Possible Trial Designs History Small-cell lung cancer has been shown to be biologically distinct from other histologic subtypes of lung cancer in both clinical and laboratory studies conducted since the 1960s (5,6). Untreated, it exhibits a very aggressive natural history; patients presenting with extensive disease have a median survival of only 6 weeks (7). This relentless clinical course is consistent with the results of thymidine-labeling studies, which demonstrate a high proliferative index (8). In addition, autopsy evidence indicates that the disease is frequently disseminated even in patients whose disease clinically appears to be confined to the chest (9). Fortunately, small-cell lung cancer was shown in many early studies of cytotoxic therapy to be the most chemosensitive type of lung cancer (7,10). By 1977, 15 agents had demonstrated activity (objective response rate of >20%) in this disease (10). Subsequently, combination chemotherapy (consisting of three or four drugs) was found to be superior to single agents or twodrug combinations in randomized and nonrandomized trials (7012). By 1980, combination chemotherapy was considered to be the standard treatment (75). While this strategy results in responses of meaningful duration in the majority of patients, it remains uncommon for a patient with extensive disease to survive beyond 2 years (73). In addition, with the acceptance of polychemotherapy as standard treatment, investigational drugs have not commonly been administered until a patient has been exposed to several cytotoxic agents having different mechanisms of action. As a result,
150
Designs for Pretreated Patients Lower the target response rate to 10%. Conventional phase II trials in previously treated small-cell lung cancer patients, as well as in patients with other relatively chemoresistant tumors, usually are designed to differentiate between a drug having a 20% level of activity and one having a 5% level of activity. Specifically, the design of these trials is such that an agent with a true 20% level of activity will be declared uninteresting no more than 10% of the time (i.e., a false-negative rate ^10%) and that an agent with a true 5% level of activity will be declared interesting no more than 10% of the time (i.e., a false-positive rate ^10%). Designs that both satisfy these criteria and accrue patients in two stages are popular because they can be stopped early if it is unlikely that the agent is truly active (76); see design A in Table 1 for an example. Suppose the threshold for activity of interest is lowered to 10%, as has been suggested (2), in order to avoid missing active
Received June 26, 1991; revised September 24, 1991; accepted October 22, 1991. T. t>. Moore (Clinical Investigations Branch), E. L. Korn (Biometrics Research Branch), Cancer Therapy Evaluation Program, Division of Cancer Treatment, National Cancer Institute, Bethesda, Md. Present address: T. D. Moore, M.D., Medical Oncology Associates, 3471 Fifth Ave., Pittsburgh, PA 15213. 'Correspondence to: Edward L. Kom, Ph.D., Biometrics Research Branch, Cancer Therapy Evaluation Program, National Cancer Institute, National Institutes of Health, EPN-739, Bethesda, MD 20892.
Journal of the National Cancer Institute
Downloaded from http://jnci.oxfordjournals.org/ at Karolinska Institutet on May 25, 2015
Therapeutic options for the systemic treatment of small-cell lung cancer have not improved substantially in the past 15 years. Recognition of effective new chemotherapeutic agents is crucial, since this cancer is generally considered to be a disseminated disease at presentation. Recently, several publications have implied that many phase II studies have not been designed to evaluate fairly an investigational agent's efficacy in the treatment of this disease (1-4). Several different strategies have been proposed for testing new agents. This commentary critiques these options and presents recommendations for the design of future phase II investigational drug trials in small-cell lung cancer.
Table 1. Possible two-stage designs*
Design
No. of patients treated at stage 1 2
A
15
Add 20
B
29
Add 181
C
29
Add 41
D
29
Add 21
Declare drug interesting iff S4/35 responses £15/210 responses £4/70 responses 22/50 responses
1%
.005 (.25) .076 (.25)
2%
True response ratet 3% 5%
.047 (.44) .23 (.44)
.09 (.54) .10 (.77) .43 (.77) .94 (.95)
.14 (.59) .39 (.59)
10%
20% .92 (96)
.90 (.95) .90 (.95)
*In each design, we proceed to stage 2 if there is at least one response seen in stage 1. tResults are expressed as No. of responses/total No. of patients treated. ^Numbers are probabilities of declaring the drug interesting. Numbers in parentheses are the probabilities a second stage will be required.
Since the desire here is to determine a level of disinterest, only the inactive and marginally active drugs will be considered. Depending on which of these agents is viewed as truly inactive, the overall observed response rate for inactive agents could be as low as 0 of 573 or as high as 16 of 912 (1.8%). The corresponding upper 95% confidence limits for the true response rate are 0.5% and 2.7%. These observed response rates and confidence limits suggest that the setting of a level of disinterest of 2% and the use of design C in Table 1 would probably yield an acceptable false-positive rate. However, even if these limitations are accepted, 44% of the time a trial of a new drug with a true response rate of 2% would not be terminated at 29 patients but would require 70 patients. In addition to setting the level of disinterest, another important issue is the appropriateness of treating patients when only a 10% response rate is targeted. It is questionable whether such Vol. 84, No. 3, February 5,1992
critically ill patients should be subjected to potentially toxic chemotherapy when there is such a low expectation of response even to active investigational agents. This observation is true even though these patients would not respond well to standard salvage chemotherapy. At a minimum, a frank discussion with the patient of the limited goals of the trial is essential. Use of relapsed patients, keeping the target rate at 20%. Phase n studies of pretreated patients are usually composed of a heterogeneous population. Some patients have refractory disease in which the tumor grows, or is at best stable, on induction therapy. However, other patients respond to treatment and then relapse after a period of freedom from chemotherapeutic treatments. This distinction has important consequences, both in small-cell and other malignancies, because a higher proportion of relapsed patients may respond to second-line treatment than those whose disease is truly refractory to chemotherapy (21-23). These studies indicate that relapsed patients may respond not only to second-line regimens, but also to cytotoxic agents to which they were previously exposed. In selected relapsed smallcell lung cancer patients, therefore, it may be reasonable to have higher expectations for systemic treatments than in the relapsed, refractory patient population taken as a whole. This statement translates into the ability to realistically target a higher response rate (e.g., 20%) without being overly concerned that an active drug will be missed. This statement, in turn, implies that a smaller sample size is required (e.g., design A). Three uncertainties exist with this approach. Considering the appreciable response rates observed in relapsed patients with reinduction-salvage regimens, physicians might be reluctant to advocate investigational treatments, despite the relatively short remissions observed with exposure to previously active drugs. Secondly, the general feasibility of this approach regarding the number of available patients and their willingness to participate in clinical investigations at this stage in their treatment is unknown, although certain examples exist of successful studies carried out in this manner (2425). In particular, long-term toxic effects of the initial therapy overlapping with the study drug may restrict the numbers of patients eligible. Finally, it is unclear what the minimal treatment-free interval should be before a patient is considered a good risk to be re-treated with conventional cytotoxic chemotherapy. This question has not been proCOMMENTARY 151
Downloaded from http://jnci.oxfordjournals.org/ at Karolinska Institutet on May 25, 2015
agents that may not achieve a 20% response rate in this pretreated population. If the level of disinterest remains at 5%, a large number of patients will be required, e.g., design B in Table 1. The large number of patients is required to keep the false-positive rate low, i.e., to keep from declaring an agent interesting when the true response rate is 5%. If one lowers the level of disinterest from 5% to 2%, then design C in Table 1 shows that 70 patients would be sufficient to ensure that the false-positive rate is less than 10%. If a 1% level of disinterest were appropriate, then 50 patients would be sufficient (design D in Table 1). Determining the level of disinterest for small-cell lung cancer is not a trivial task. One consideration is the measurement error inherent in determining whether a patient has had a response (77). Another consideration is the occasional true response seen with inactive agents. Theoretically, both of these factors are taken into account by averaging the observed response rates from a number of trials of inactive agents. However, in practice, it is not known which agents are truly inactive; we have only the observed response rates by which to be guided. This difficulty is illustrated if one considers the observed response rates of previously treated small-cell lung cancer patients from singleagent, phase II trials. Table 2 contains the published results of such trials obtained through a MEDLINE search, an issue of Cancer Treatment Symposia (78), and three review articles {15,1920). (A complete list of references is available from the authors.)
Table 2. Response rates in phase II trials of previously treated small-cell lung cancer patients* No. of patients responding/total No. of patients
%
Drug
9/72 27/159 4/14 42/388 3/26 6/41 18/117 5/24 17/100
13 17 29 11 12 15 15 21 17
Carboplatin Cisplaiin Doxorubicin Etoposide Fluorouracil Ifosfamide Teniposide Vincristine Vindesine
"Marginally active drug" 9
3 4 7 3 10 4 2 3 6
ACNUt Cytarabine Eflomithine Epirubicin Iproplatin Lonidamine N-Methylformamide Mitoguazone Mitoxantrone PCNUt
"Inactive drug" 0/29 0/127 0/17 0/18 0/10 0/43 0/11 0/15 0/23 0/15 0/17 0/30 0/19 0/21 0/24 0/29 0/41 0/20 0/14 0/34 0/16
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Aclarubicin Amsacrine Bisantrene Chlorozotocin Dianhydrogalactitol Diaziquone Fludaratrine phosphate Interferon-a Maytansine 6-Mercaptopurine Methotrexate Mitomycin-C PALA§ Prednimustine Procarbazine Spirogermanium Streptozocin TGUH Tiazofurin Vinblastine Zinostatin
*Only dmgs that were evaluated in at least 10 patients were included. They have been classified into three groups based on their response rates: "active" for response rates (complete response or partial response) >10%, "marginally active" for agents with 30%) can be required for a drug to be considered worthy of further testing, since a patient population with a relatively favorable risk is studied (28). A major concern with this strategy in the treatment of smallcell lung cancer relates to the biologic aggressiveness of the disease. Reluctance on the part of the investigator to advocate initial treatment with an untested therapy centers on the fear that delaying effective multidrug chemotherapy will result in an appreciable decrease in survival (331). This reluctance may be mitigated to some extent by switching to salvage therapy after one cycle if the patient has not responded in some well-defined manner. For example, evidence (32) suggests that absence of a decrease in tumor size is predictive that the patient will not achieve at least a partial response in subsequent cycles. Further research would be helpful for evaluating strategies that predict, after one cycle of therapy, whether or not a patient will achieve at least a partial response. The apprehension of delaying effective therapy is most relevant for trials in which a drug later proven to be inactive is tested. Survival inferior to historical controls has been observed when an inactive agent (oral idarubicin) was tested using the window-of-opportunity design (57). The Eastern Cooperative Oncology Group recently reported results of a 90-patient, randomized trial in which patients receiving an inactive drug (menogaril), with cisplatin and etoposide salvage (EP) for patients having progressive or stable disease, were compared with those receiving standard therapy consisting of cyclophosphamide, doxorubicin, and vincristine (CAV) from the outset (29). No adverse effect on survival was claimed for menogaril-EP (median survival, 38.3 weeks), compared with the CAV arm (median survival, 41.7 weeks), despite the fact that only two partial responses (5%) were observed in patients receiving menogaril. It is questionable whether that study has adequately proven that the window-of-opportunity approach is equivalent to standard therapy when an inactive drug is tested. Although the observed ratio of median survivals (92% = 38.3/41.7) is close to 100%, one cannot be very confident of this estimate because the sample size is limited. For example, even assuming that all the patients have been followed to their death, an approximate 90% confidence interval of the ratio of median survivals would be broad: 64%-131%. While it might be argued that wide confidence intervals are acceptable, considering the dismal prognosis for patients with extensive-stage small-cell lung cancer, it is doubtful whether the risk of a 36% decrement in median survival would be considered acceptable by many investigators. An equivalency trial would be required in order to prove rigorously that an unacceptable decrement in survival does not Journal of the National Cancer Institute
Downloaded from http://jnci.oxfordjournals.org/ at Karolinska Institutet on May 25, 2015
3/34 1/36 1/23 1/14 1/34 3/30 1/25 1/46 2/63 2/34
Designs for Patients Without Prior Exposure to Cytotoxic Therapy
Limit to elderly patients or patients with very poor prognoses. It has been suggested that elderly patients or patients with especially poor prognoses (e.g., patients with two or more metastatic sites or bone marrow metastases) can be treated with investigational agents without prior treatment (34-37). The rationale behind this suggestion appears to be that these patients may not be able to tolerate combination chemotherapy or may expect relatively fewer benefits from this aggressive approach. This strategy has been used to both support (34,36) and rule out (55) a drug's efficacy in the treatment of this disease.
Table 3. Sample sizes per group required to have 90% power for detecting the alternative that the hazard ratio is as specified*
Hazard ratio 0.50 0.70 0.80 0.90 *a = .05, one-sided test.
Vol. 84, No. 3, February 5, 1992
Required sample size per group 36 135
345 1540
Some problems with this approach, however, can be highlighted by asking two questions: 1) Is the response rate to new agents expected to be similar in this group of patients to that in small-cell lung cancer patients at large? 2) Is there an effective way to treat these patients with standard agents? To arrive at an overall appraisal, we need to consider individually the two groups defined above. For the group of elderly patients, limited experience with carboplatin (34) and teniposide (36) suggests that the response rates may be similar to those expected in nonelderly patients, which answers question 1. However, with regard to question 2, there are ways to treat these patients with standard agents that will result in meaningful response rates. Even if combination therapy is not advisable because of coexisting medical conditions, one of the established drugs used in this disease would likely have an acceptable toxicity profile that would allow its use as a single agent. Thus, with these criteria, elderly patients appear similar to untreated patients in general. Therefore, it is not clear why they would not be subject to the same concerns mentioned in the previous section. The other category of patients presumed to have low expectations of benefiting from standard chemotherapy consists of those patients with poor prognostic features, such as a poor performance status or bone marrow metastases. Although these patients will occasionally respond to standard therapy (38), the benefits of such therapy are very limited (59), thus answering question 2. However, is the response rate to new agents expected to be similar in this group to that in small-cell lung cancer patients at large? We would intuitively expect these response rates to be lower. In fact, such has been the case in a group of patients with poor performance status treated with etoposide (40). The problem in trial design for these patients, therefore, is no different from that in trial design for previously treated patients in which lower response rates are expected (see section "Lower the target response rate to 10%")- An additional problem in treating patients with poor performance status is that they may be subject to excessive toxic effects (40,41).
Recommendations We suggest the following guidelines for the study of investigational drugs in the treatment of small-cell lung cancer. We recommend that a phase II design, targeting at least a 30% response rate, should be used for previously untreated patients, for the following classes of agents: (a) Agents having novel or unknown mechanisms of activity and which show clinical activity in other solid tumors (e.g., taxol). (b) Agents showing specific activity for small-cell lung cancer in preclinical drug screens. (c) Analogues of active drugs currently used in this disease, which are being developed primarily because of a different toxicity profile (e.g., carboplatin). These studies should be limited to patients with extensivestage disease, Kamofsky performance status of 70% or more, no superior vena cava syndrome, and no significant visceral organ COMMENTARY 153
Downloaded from http://jnci.oxfordjournals.org/ at Karolinska Institutet on May 25, 2015
occur. This hypothetical trial would compare a series of drugs, later proven to be inactive, given by the window-of-opportunity method, with conventional therapy. Depending on the hazard ratio (i.e., the ratio of median survival times) that would be considered tolerable, however, large sample sizes could be required (Table 3). A modification of the window-of-opportunity approach has been described in which the overall response rate of experimental treatment followed by standard treatment is compared with the historical response rate of standard therapy (4,33). Using this design, the trial is stopped early if the response rate to the sequential therapy is too low. For example, taking 67% to be the expected response rate to standard therapy for extensive smallcell lung cancer (4), we might stop the trial of the sequential treatment if five or more nonresponders were observed out of the first seven patients treated. There are three problems with this approach. The first is that this approach offers only very limited protection for the patient. Continuing with the above example, if the true response rate for the sequential therapy is 40%, then the probability of getting five or more nonresponders out of the first seven patients treated and stopping early is only 42%. Furthermore, even with equal response rates, there is a potential for a decrement in median survival. Secondly, this approach does not adequately address the question of whether an investigational drug has activity against small-cell lung cancer. For instance, with the use of the above example, it is possible for a response to the investigational drug to be observed and still have the study terminated because of a lack of drug activity. Finally, this approach assumes that the true response rate to standard therapy is known for the patients about to be treated with the sequential therapy. With the heterogeneity of patients entered in phase II trials, this response rate may be difficult to estimate. Therefore, this modification of the window-of-opportunity approach is not satisfactory.
References (/) ETTINGER DS: Evaluation of new drugs in untreated patients with smallcell lung cancer Its time has come. J Clin Oncol 8:374-377, 1990 (2) KRBTJANSEN PEG, HANSEN HH: Management of small cell lung cancer A summary of the Third International Association for the Study of Lung Cancer Workshop on Small Cell Lung Cancer. J Natl Cancer Inst 82:263-266, 1990 (3) AISNER J: Identification of new drugs in small cell lung cancer: Phase II agents first? Cancer Treat Rep 71:1131-1133, 1987 (4) CULLEN M: The design of phase II trials. Lung Cancer 5:214-220, 1989 (5) COHEN MH, MATTHEWS MJ: Small cell bronchogenic carcinoma: A distinct clinicopathologic entity. Semin Oncol 5:234-243, 1978
sessment of response to anticancer chemotherapy: Proposal for new criteria of tumor response. J Clin Oncol 2:1040-1046, 1984 (18) WrrTES RE, ED: Cancer Treatment Symposia, vol 4. Washington, DC: US Govt Print Off, 1985 (19) Joss RA, CAVALLJ F, GOLDHIRSCH A, ET AL: New drugs in small-cell lung cancer. Cancer Treat Rev 13:157-176, 1986 (20) AISNER J, ABRAMS J: Cisplatin for small-cell lung cancer. Semin Oncol 16:2-9,1989 (21) BATIST G, IHDE DC, ZABELL A, ET AL: Small-cell carcinoma of lung:
Reinduction therapy after late relapse. Ann Intern Med 98:472-474, 1983 (22) GIACONNE G, DONADIO M, BONARDI G, ET AL: Teniposide in the treatment
of small-cell lung cancer The influence of prior chemotherapy. J Clin Oncol 6:1264-1270,1988 (23) JOHNSON DH, STRUPP J, GRECO FA, ETAL: Prolonged administration of oral
etoposide in previously treated small cell lung cancer (SCLC) patients: A phase II trial. Proc ASCO 9:227, 1990 (24) GIACONNE G, FERRATI P, DONADIO M, ET AL: Reinduction chemotherapy in
small cell lung cancer. Eur J Clin Oncol 23:1697-1699, 1987 (25) POSTMUS PE, BERENDSEN HH, VAN ZANDWUK N, ET AL: Retreatment with
the induction regimen in small cell lung cancer relapsing after an initial response to short term chemotherapy. Eur J Cancer Clin Oncol 23:14091411,1987 (26) GIACCONE G: Second line chemotherapy in small cell lung cancer. Lung Cancer5:207-213, 1989 (27) BLACKSTEIN M, EISENHAUER EA, WIERZBICKI, ET AL: Epirubicin in exten-
sive small-cell lung cancer: A phase II study in previously untreated patients: A National Cancer Institute of Canada Clinical Trials Group Study. J Clin Oncol 8:385-389, 1990 (28) EVANS WK, EISENHAUER EA, CORMIER Y, ETAL: Phase II study of amona-
fide: Results of treatment and lessons learned from the study of an investigational agent in previously untreated patients with extensive small-cell lung cancer. J Clin Oncol 8:390-395, 1990 (29) ETTINGER DS, FINKELSTEIN DM, ABELOFF MD, ET AL: Justification for
evaluating new anti-cancer drugs in selected untreated patients with a chemotherapy-sensitive advanced cancer: An ECOG randomized study. Proc ASCO 9:224, 1990 (30) SMITH IE: New treatments for small cell lung cancer. When to test. Chest 96:59S-61S, 1989 (31) CULLEN MH, SMITH SR, BENFIELD GFA, ET AL: Testing new drugs in un-
treated small cell lung cancer may prejudice the results of standard treatment: A phase II study of oral idarubicin in extensive disease. Cancer Treat Rep 71:1227-1230, 1987 (32) WOLF M, HAVEMANN K, HOLLE R, ET AL: The prognostic value of response
to the first cycle of chemotherapy in small cell lung cancer. Results of a multicenter German trial. Eur J Cancer Clin Oncol 23:1197-1205, 1987 (33) CULLEN MH, HILTON C, STUART NSA: Evaluating new drugs as first treatment in patients with small-cell carcinoma: Guidelines for an ethical approach with implications for other chemotherapy-sensitive tumors. J Clin Oncol 6:1356-1357,1988
(6) CARNEY DN, BRODER L, EDELSTEIN M, ET AL: Experimental studies of the
(34) SMITH IE, HARLAND SJ, ROBINSON BA, ET AL: Carboplatin: A very active
biology of human small cell lung cancer. Cancer Treat Rep 67:27-35,1983
new cisplatin analog in the treatment of small cell lung cancer. Cancer Treat Rep 69:43-46, 1985
(7) GREEN RA, HUMPHREY E, CLOSE H, ETAL: Alkylating agents in bronchoge-
nic carcinoma. Am J Med 46:516-525, 1969 (8) SHACKNEY SE, STRAUS MJ, BUNN PA JR: The growth characteristics of
small cell carcinoma of the lung. In Small Cell Lung Cancer (Greco AF, Oldham RK, Bunn PA Jr, eds). New York: Grune, 1981, pp 225-234 (9) MATTHEWS MJ, KANHOUWA S, PICKREN J, ET AL: Frequency of residual and
metastatic tumor in patients undergoing curative surgical resection for lung cancer. Cancer Chemother Rep 4:63-67, 1973 (10) BRODER LE, COHEN MH, SELAWRY OS: Treatment of bronchogenic carci-
noma. II. Small cell. Cancer Treat Rev 4:219-260, 1977 (//) BUNN PA, COHEN MH, IHDE DC, ETAL: Advances in small cell bronchoge-
(12) (75) (14) (15) (16)
154
(17) WARR D, MCKJNNEY S, TANNOCK I: Influence of measurement error on as-
nic carcinoma. Cancer Treat Rep 61:333-342, 1977 GRECO FA, EINHORN LH, RICHARDSON RL, ET AL: Small cell lung cancer Progress and perspectives. Semin Oncol 5:323-335, 1978 MORSTYN G, IHDE DC, LICHTER AS, ET AL: Small cell lung cancer 19731983: Early progress and recent obstacles. Int J Radial Oncol Biol Phys 10:515-539,1984 ETNHORN LH: Cisplatin plus VP-16 in small cell lung cancer. Semin Oncol 13:3-4,1986 JOHNSON DH: New drugs in the management of small cell lung cancer. LungCancer5:221-23l, 1989 FLEMING TR: One-sample multiple testing procedures for phase II trials. Biometrics 38:143-151, 1982
(35) LUND B, HANSEN F, HANSEN M, ET AL: Phase II study of 1,2,4-
triglycidylurazol (TGU) in previously untreated and treated patients with small cell lung cancer. Eur J Cancer Clin Oncol 23:1031-1033, 1987 (36) BORK E, HANSEN M, DOMBERNOWSKY P, ET AL: Teniposide (VM-26), an
overlooked highly active agent in small-cell lung cancer. Results of a phase II trial in untreated patients. J Clin Oncol 4:524-527, 1986 (37) AISNER J, ALBERTO P, BITRAN J, ETAL: Role of chemotherapy in small cell
lung cancer: A consensus report of the International Association for the Study of Lung Cancer Workshop. Cancer Treat Rep 67:37-43, 1983 (38) IHDE DC, MAXUCH RW, CARNEY DN, ET AL: Prognostic implications of
stage of disease and sites of metastases in patients with small cell carcinoma of the lung treated with intensive combination chemotherapy. Am Rev RespirDis 123:500-507, 1981 (39) OSTERLAND K: Prognostic factors in small cell lung cancer An analysis of 874 consecutive patients. In Lung Cancer Basic and Clinical Aspects (Hansen HH, ed). Boston: Martinus-Nijhoff, 1986, pp 139-152 (40) POSTMUS PE, SMTT EF, BERENDSEN HH, ET AL: A feasibility study of test-
ing new drugs for small-cell lung cancer in patients with a poor performance status. Cancer Chemother Pharmacol 27:490-491,1991 (41) CERNY T, PEDRAZZINI A, JOSS RA, ET AL: Unexpected high toxicity in a
phase II study of teniposide (VM-26) in elderly patients with untreated small cell lung cancer (SCLQ. Eur J Clin Oncol 24:1791-1794, 1988
Journal of the National Cancer Institute
Downloaded from http://jnci.oxfordjournals.org/ at Karolinska Institutet on May 25, 2015
dysfunction. Unless there is prior evidence that multiple courses will be required to observe a response, the window-of-opportunity strategy should be used. In addition, every effort should be made to choose a salvage regimen that has as few as possible toxic effects in common with the investigational agent. For agents not in drug classifications a-c, we recommend that they be studied in previously treated patients who have had disease recurrence after a disease-free and treatment-free interval sufficient to exclude those patients having predominantly cytotoxic agent-refractory disease. These agents should be tested using a conventional phase II design targeting a 20% response rate. We recommend a 3-month treatment-free interval after a complete response, although further research validating this time frame would be helpful. These recommendations are meant to provide guidelines for the initial evaluation of investigational drugs in small-cell lung cancer. Additionally, we are not addressing the development of analogues that are being investigated primarily because of crossresistant properties to their parent compound, e.g., tetraplatin. These agents would be first tested in pretreated patients. To place these guidelines in their proper perspective, it is essential that one does not equate clinical investigation with clinical care. It has not been the intention of this commentary to suggest therapy for patients ineligible for conventional investigational drug trials. Rather it is hoped that focusing investigational drug development in the manner described will result in improved treatment options for all small-cell lung cancer patients.
