絞ヶ弇离: 忑珜 > ぶ膳 > ▲還散笫雖悝◎ > 2005爛菴6ぶ > 淏恅
晤瘍:11329472
Successful Implementation of the Randomized Discontinuation Trial Design: An Application to the Study of the Putative Antiangiogenic Agent C
http://www.100md.com ▲還散笫雖悝◎
     the University of ChicagoMedical Center, Chicago, IL

    Cancer and Leukemia Group B Statistical Center, Durham

    Southeast Cancer Control Consortium Inc, CCOP, Goldsboro, NC

    University of California at San Francisco, San Francisco, CA

    University of Iowa, Iowa City, IA

    ABSTRACT

    PURPOSE: To assess the disease-stabilizing activity of carboxyaminoimidazole (CAI) in patients with metastatic renal cell cancer (RCC) using a randomized discontinuation trial (RDT) design.

    PATIENTS AND METHODS: Recruited patients had a performance status of 0 to 2, minimal neuropathy or cerebellar dysfunction, measurable disease, and normal organ function. Treatment with 250 mg/d CAI was initiated in all patients and continued until disease progression in those with an objective response. Protocol treatment was discontinued for unacceptable toxicity or progressive disease; patients with stable disease at the 16-week evaluation point were randomly assigned in a double-blind manner to continued CAI or placebo. The primary end point was the stable disease rate in the randomized groups.

    RESULTS: A total of 368 patients were accrued and received therapy. Ninety percent had a performance status of 0 or 1, 80% underwent a prior nephrectomy, and 41% had received no prior systemic therapy. Serious or life-threatening toxicity was experienced by 34%, with asthenia (15%) and neuropsychiatric difficulties (7%) being most common. At the randomization point, 51% of patients had progressed, 30% withdrew, 1% experienced a partial response, and 17% had stable disease and were randomly assigned. A Bayesian futility analysis utilizing the first 49 randomly assigned patients suggested that the probability of demonstrating a higher stable disease rate in the experimental group was less than 9% even under the most optimistic a priori assumptions, and further trial accrual was halted.

    CONCLUSION: CAI is inactive in RCC. The RDT design should be further explored for evaluating activity of putative disease stabilizing agents.

    INTRODUCTION

    Patients with metastatic renal cell carcinoma (RCC) have a poor prognosis.1 Immunotherapy with interferon alfa, interleukin-2, or combinations containing both these agents can achieve a 10% to 15% objective response rate. However, partial and nonresponders have a median survival of 6 to 8 months, and no chemotherapy program has resulted in reproducible complete responses.2 Antiangiogenic agents are of high interest as potential antitumor agents in RCC patients, in part due to its highly vascular nature. In patients with clear-cell RCC inactivation of the Von Hippel Lindau (VHL) gene leads to overexpression of HIF-1 and HIF-2,3,4 which in turn results in elevated expression of multiple proangiogenic factors including vascular endothelial growth factor (VEGF).5-7

    In nonclinical models, antiangiogenic drugs have been reported to be mostly growth inhibitory and cytostatic, without leading to significant shrinkage in established tumors. Nevertheless, successful clinical development of a minimally toxic cytostatic agent would certainly be considered an advance in the therapy of this disease. Classic paradigms for phase II evaluation of putative cytostatic agents are, however, inadequate. Traditionally, the phase II end point for cancer drugs has been tumor shrinkage as defined by arbitrary but standardized radiologic criteria.8 For most solid tumors, it is appropriate to assume that the response rate in the absence of therapy is 0%, which would be considered the "no treatment effect." Because spontaneous remissions have been reported in renal cell cancer, this value is somewhat higher in this disease, but is clearly less than 10%.9 Therefore, reproducible or clear tumor shrinkage rates of more than 10% would be considered sufficient evidence for drug activity.

    Cytostatic drugs are not expected to cause tumor shrinkage, and thus, the relevant clinical end point is duration of stable disease or time to disease progression. In this case, however, the variable natural history of renal cancer makes it extremely difficult to define a "no treatment effect" for time to progression that, if exceeded in an uncontrolled phase II study, could reliably be considered evidence of drug activity.10 This has lead a number of investigators to conclude that establishment of cytostatic activity in an oncologic clinical setting requires a randomized comparison to a concurrent control group, consistent with paradigms used for phase II clinical trials in other therapeutic areas.11

    In a classical randomization scheme, patients are assigned to either the investigational agent or placebo, and time to progression, or stable disease rate at an arbitrary time point would then be compared. More recently, we proposed an alternative randomized discontinuation trial (RDT) design.12 With this design, all patients receive the investigational drug for an initial period of time. Patients with standard radiologic tumor shrinkage within that timeframe would continue investigational therapy, while those with radiologic progression or unacceptable toxicity would discontinue therapy. All patients with radiologic stable disease after the initial therapy period are then randomized to continuing or discontinuing therapy in a double-blind placebo-controlled manner. This is thus an enrichment strategy in which patients with the end point of interest are preferentially enrolled in the randomized portion and in which the heterogeneity of the randomized population is decreased. These two factors result in an increased power for detecting a clinically relevant difference and decrease the number of patients exposed to placebo. Importantly, the enrichment is "driven" by the properties of the investigational drug as opposed to clinical prognostic factors identified in historical untreated patients or patients treated with a different class of agents. In addition, the statistical behavior of the trial is not highly dependent on investigators' assumptions regarding the "no dose effect" for time to progression or stable disease rate, and thus effectively deals with uncertainty in this variable.12 Finally, patients may find such a trial design more appealing, resulting in brisk accrual.

    To test the feasibility of this clinical trial design, we initiated a study with the putative antiangiogenic carboxyaminoimidazole (carboxyamidotriazole [CAI]; IND 38,583, NSC 609974) within the Cancer and Leukemia Group B. CAI is an oral agent whose mechanism of action is believed to be due to inhibition of nonvoltage gated calcium mediated signal transduction.13 In vitro, this results in inhibition of phosphalipase C- and phosphalipase A2 phosphorylation and subsequent blockade of arachidonic acid release.14 This leads to inhibition of tumor cell mobility and invasion, as well as endothelial cell proliferation, adhesion, motility, and tube formation.14,15 CAI is also active in a variety of xenograft models.16,17 Clinical studies have shown that CAI can be administered orally with major toxicities being asthenia, neuropsychiatric difficulties, and nausea.18-20 The investigators for these phase I trials suggested that this agent may have activity in renal cell cancer based on observations of disease stabilization and minor responses.

    Thus, we aimed to assess the feasibility of the RDT design in a multicenter study, as well as attempt to confirm the preliminary evidence for activity in metastatic renal cell carcinoma.

    PATIENTS AND METHODS

    Patients and Treatment

    Patients were recruited from cancer and Leukemia Group B (CALGB) member sites. Histologic documentation of metastatic or unresectable renal cell carcinoma and documented radiological disease progression before enrollment was required. There were no restrictions on prior therapy, but at least 4 weeks had to have elapsed since prior major surgery, radiotherapy, immunotherapy, or chemotherapy. Patients were required to have a Common Toxicity Criteria (CTC) performance status of 0 to 2, no baseline neuropathy or cerebellar dysfunction greater than grade 1, and all were required to have measurable disease by standard Response Evaluation Criteria in Solid Tumors (RECIST) criteria.8 Laboratory requirements included WBC more than 2,000/米L, platelet count 75,000/米L, creatinine 2 mg/dL, bilirubin less than upper limits of normal, and AST 2.5 x the upper limits of normal.

    All patients provided written informed consent before enrollment. All patients were initially treated with oral CAI at a dose of 250 mg/d. Radiologic disease re-evaluation occurred every 8 weeks. Patients with RECIST-documented response continued therapy until disease progression or undue toxicity. Patients with RECIST progression by 16 weeks or with intolerable toxicity were removed from the protocol. Patients with stable disease after 16 weeks of therapy were randomly assigned in a double-blind manner to continuation of CAI or placebo. Random assignment was performed in the CALGB Statistical Center, and was stratified by time from diagnosis of metastatic disease to study entry (< 24 months v 24 months). Radiologic disease re-evaluation continued every eight weeks following randomization and patients continued therapy until documentation of progressive disease. At that point, patients were unblinded, and if found to be on the placebo arm could cross over to experimental therapy at their and their physician's discretion.

    Toxicity was assessed every 4 weeks using CTC version 2.0 guidelines. Protocol defined dose modifications included decreases of the CAI dose to 200 mg/d and then to 150 mg/d for grade 2 neuropsychiatric toxicity or other grade 3/4 toxicities. Patients with significant toxicities at a dose of 150 mg/d or whose therapy was interrupted for more than 4 weeks were removed from the study.

    Statistics

    Statistical details of the RDT design as applied to this study have been previously published.12 In brief, the protocol specified two decision points. The first was an early stopping rule that was to be invoked if the rate of progression at 16 weeks was much higher or lower than expected under a Bayesian analysis performed after every 20 patients. Although the simulations performed to plan trial size did evaluate responses every 20 patients, the complexities of the cooperative group process meant that actual reassessments were performed every 3 months for Data and Safety Monitoring Board (DSMB) review. The second decision point was the primary end point in which the stable disease rate in the two randomized populations was compared. The assumptions for the statistical modeling that were used to implement the early stopping rule included a 30% stable disease rate at 16 weeks and a beta(7,3) distribution for the progressive disease risk. The stable disease rate was based on single institution observations with refractory renal cell cancer patients treated with inactive cytotoxic therapies (Stadler, unpublished). The early stopping rule specified that the trial end if, based on this a priori distribution and the accumulating data, the progressive disease rate at 16 weeks was greater than 90% or less than 20%. The data were reviewed quarterly by a CALGB DSMB. Standard O'Brien-Fleming boundary rules were also used to assess the difference in stable disease rate between the two randomized populations halfway through the planned accrual (ie, after 50 patients had been randomly assigned). The Kaplan-Meier method was used to calculating time to progression of the randomized patients. Those who went off treatment for any reason other than progression or who continued treatment at the time of data analysis were censored.

    The power of this study was determined using simulations. An additional assumption included an exponential tumor growth with growth rates distributed in a log-normal fashion. The model also assumed a CAI effect of 1 if the drug was completely effective and no further tumor growth occurred, and an effect of 0 if the drug was completely inactive. With these assumptions, the power to detect a CAI effect of 0.46 was 86% with a total of 100 randomly assigned patients. This is equivalent to a 16-week postrandomization progressive disease risk of 94% in the placebo group versus 72% in the experimental group.

    Data and Safety Monitoring

    Patient data collection was managed by the CALGB Statistical Center. Data quality was ensured by careful review of data at the CALGB Statistical Center and by the study chairperson. A DSMB reviewed data quarterly. As part of the CALGB quality assurance program, members of the Data Audit Committee visit all participating institutions at least once every 3 years to review source documents. Such on-site review of medical records was performed in a subgroup of 47 patients (13% of the enrollees).

    RESULTS

    A total of 374 patients were accrued between December 2000 and July 2002 for a rate of 19 patients per month. Six patients withdrew from the study before starting any treatment. Table 1 presents patient characteristics for the 368 patients who received treatment. Seventy percent of patients were male, the median age was 61 years, 80% had a prior nephrectomy, and only 10% had a performance status of 2. Data on other clinical prognostic factors such as time from diagnosis of metastatic disease, number of metastatic sites, LDH, hemoglobin, and calcium were not collected. Forty-one percent of patients had received no prior systemic therapy for their disease.

    Table 2 provides data on serious toxicity during the open-label, first phase of the trial. One hundred twenty-four patients (34%) experienced a serious or life-threatening toxicity, of which one was a fatal respiratory failure in a patient with concomitant COPD. As expected and previously reported, the most common serious toxicity was asthenia and neuropsychiatric difficulties, experienced by 15% and 7% of patients, respectively. Peripheral neuropathy, anemia, and nausea were also common. These events led to dose reductions or treatment delays in 101 patients, and protocol discontinuation in 35 patients.

    Table 3 provides data regarding outcome during the open-label, first phase of the trial. Of the 368 treated patients, 51% of patients had documented progressive disease by the 16-week random assignment point and were withdrawn from the study. An additional 30% of patients were withdrawn from the study at or before 16 weeks due to death, toxicity, patient choice, or protocol closure (see below). Five patients (1%) experienced an objective partial response by standard RECIST criteria and continued on open-label CAI until progression. Thus, only 17% of patients (n = 64) went on to the randomized portion of the study. Although this was somewhat lower than the 30% predicted rate it did not cross the prespecified boundaries for early stopping (see Patients and Methods). None of the patient characteristics enumerated in Table 1 was predictive of going on to the randomized portion of the study (data not shown).

    Although not prespecified in the protocol, as part of the data safety and monitoring board (DSMB) activities, a Bayesian futility analysis was performed utilizing outcome data from the first 49 randomized patients who completed 16 weeks of blinded postrandomization therapy. At that time, preliminary reports indicated that 14 of 23 patients randomized to continuing CAI and 13 of 26 randomized to placebo progressed by 16 weeks postrandomization. Under the assumption of a noninformative prior hypothesis for progressive disease risk after randomization as previously suggested,21,22 the probability of concluding that the risk of progressive disease in the CAI group would be less than in the placebo group, if the trial continued until 100 patients were randomized, is less than 1%. If an unrealistic, protocol defined, prior hypothesis is used in which the risk of progression in the CAI group is 6 times less than in the placebo group, the posterior distribution of the progressive disease risk 16 weeks after randomization is such that there is only a 9% chance that full accrual to 100 randomized patients would lead to the conclusion that the progressive disease risk in the CAI group is less than that in the placebo group. As a result of this analysis the DSMB recommended that the trial be closed. Table 4 presents the final reviewed outcome of all randomized patients, and Figure 1 shows the final progression-free survival curves of the randomized patients. In support of the DSMB decision, no difference in progressive disease rate between the two groups is discernable. The median times to progression (and 95% CI) for the CAI and placebo groups from the time of randomization are 2.8 months (2.1 to 5.4) and 4.2 months (2.4 to ), respectively.

    DISCUSSION

    This study demonstrates that the RDT design is feasible in patients with metastatic renal cell cancer and that CAI is unlikely to have a disease stabilizing effect in this disease. Importantly, this conclusion was reached very quickly as a result of brisk accrual, demonstrating that treating physicians and patients considered the RDT design acceptable. In fact, 41% of patients had no prior therapy demonstrating that patients with this disease will accept investigational therapy with such a design even before standard immunotherapy.

    It is also instructive to compare the results of this trial with a more traditional phase II trial of CAI in essentially the same patient population conducted by the Eastern Cooperative Oncology Group. That trial used a single arm, uncontrolled design in which the primary end point was the fraction of patients who were progression free at 6 months postenrollment (Gray J, personal communication, March 2003). Fifty-six patients were enrolled and similar to the results reported here, 2% experienced an objective response and 25% were progression free at the six-month point.23 Interestingly, however, per the protocol defined criteria, this observation constituted a "positive" trial justifying CAI worthy of further investigation. The results from the randomized portion of the trial reported here makes it clear that the observed stable disease rate in both trials is most likely due to the well-known slow natural history of some metastatic renal cell cancer patients and not due to the effects of CAI. This also illustrates the RDT design's ability to eliminate false-positive conclusions regarding drug activity. Single-arm designs have concentrated on minimizing false-negative conclusions, but the increasing number of putative oncologic agents and the high financial and patient resource cost of failing to confirm activity in phase III studies suggests that greater attention should be paid to the minimizing false-positive phase II results.

    The difficulty in estimating a "no treatment effect" for a stable disease rate a priori is due to multiple issues. First, renal cell cancer is highly heterogeneous and the natural history of a studied population is thus highly variable. Various prognostic classifications have been proposed to minimize this heterogeneity, but they explain the variability only in part and sufficiently broad confidence intervals on the survival curves remain even within a prognostic group.10 In addition, response to any drug is highly variable in a population and it is unlikely that prognostic classifications based on natural history will also be useful for predicting response to therapy. Finally, and as demonstrated here, patients often drop out of a trial for toxicity and other reasons without objective progression and thus the number of patients with stable disease at an arbitrary time point is likely to be heavily influenced by specifics of the therapy. In other words, in the RDT design the drug rather than the investigator's incomplete a priori knowledge "selects" the patient population putatively experiencing a positive effect, a hypothesis which is then tested in the randomized portion of the study.

    The importance of this conclusion cannot be overemphasized. CAI has shown interesting and occasional dramatic effects in nonclinical in vitro as well as in vivo models.13-17 The results from phase I studies were also encouraging and the observation of prolonged stable disease and a "minor response" in patients with renal cell cancer led several investigators to propose more formal phase II studies. Prolonged stable disease in a subset of patients in standard single arm phase II trials could then potentially encourage further study and such studies would cost invaluable financial and patient resources to come to the same conclusion that was reached in this 2-year phase II study. This is especially critical since CAI, although tolerable by usual oncologic drug criteria, can have serious toxicity including fatigue, neuropsychiatric difficulties, and neuropathy.

    It is unlikely that the conclusions reached here are a false-negative. First, the low level of objective response and the 4 month stable disease rate are not inconsistent with phase II data from multiple studies in this disease with cytotoxic agents considered to be inactive by most investigators.2 Second, evidence from patient completed surveys suggests that blinding was successful and thus biased early withdrawal of patients randomized to placebo is unlikely to have occurred. One could argue that the power of a 64 patient randomized study is minimal. However, the futility analysis demonstrated that, even under an unrealistically optimistic prior assumption, there was only a 9% chance that full accrual would lead to the conclusion that CAI maintains stable disease for 16 weeks.

    Despite the robust nature of the conclusions, the randomized discontinuation design can be criticized and one can question its utility for assessing a cytostatic agent in the oncologic setting. Specifically, this design will fail to detect a cytostatic drug whose duration of activity is short. Although technically correct, it is highly unlikely that a cytostatic that is active for less than 4 months will lead to a clinically significant patient benefit. Second, the randomized discontinuation design will lead to a false-negative result if there is a significant carry-over effect in the placebo group after drug discontinuation. However, there is no nonclinical evidence for such an effect and if present, one would expect that fewer than the observed 51% of patients would have progressed within 16 weeks after randomization (Table 3). Nevertheless, these two factors of the design suggest that it is not appropriate for evaluating an agent whose hypothesized antitumor activity is due to a short-term cytotoxic effect, as is the case with cisplatin in germ cell tumors.

    It has also noteworthy that the overall number of patients enrolled is not significantly different than in a more conventional two-arm randomized trial of placebo versus drug. In fact, had the trial continued as originally planned, with a constant 17% randomization rate, a total of 588 patients would have had to be enrolled to randomize 100. A more formal incorporation of Bayesian stopping rules as applied here, utilizing information from the nonrandomized portion of the trial, should be integrated into future RDT designs. Including all nonrandomized patients as "failures" rather than just those who progressed as was done in this trial also has the potential for increasing the design's efficiency. Even under these conditions, the total trial size will still be sizable, but the total number of patients exposed to placebo is significantly less than a conventional two-arm trial, making this design very attractive to patients. In addition, since this is a phase II "explanatory trial" rather than a phase III "pragmatic" trial,24 followup can be much shorter and requirements for data collection in the prerandomization phase are much less. Thus, the cost of the trial is much less than a phase III trial with a survival, or even time to progression, end point.

    In conclusion, accrual to the RDT design with an oral putative cytostatic agent was brisk in patients with metastatic renal cell cancer. Because of the trial size and the randomization, the conclusions reached are robust. Specifically, CAI should not be studied further in patients with metastatic renal cell cancer, and under the hypothesis that it targets the tumor vasculature, further study in other solid tumors is unlikely to be fruitful either. Due to study discontinuation as a result of toxicity and other reasons, stable disease rate at an arbitrary time point in this disease is not predictable and not useful as an end point in a single-arm uncontrolled clinical trial. We thus suggest that the randomized discontinuation design be explored further as a phase II explanatory trial for evaluating cytostatic agents. We suggest further that such trials could include multiple tumor types, with study discontinuation in those for which an interim futility analysis suggests activity is unlikely to be observed.

    Appendix

    The following institutions participated in this study: CALGB Statistical Center, Durham, NC每Stephen George, PhD, supported by CA33601; Christiana Care Health Services, Inc. CCOP, Wilmington, DE每Stephen Grubbs, MD., supported by CA45418; Dartmouth Medical School, Lebanon, NH每Marc S. Ernstoff, MD, supported by CA04326; Georgetown University Medical Center, Washington, DC, Edward Gelmann, MD, supported by CA77597; Green Mountain Oncology Group CCOP, Bennington, VT每L. Herbert Maurer, MD, supported by CA35091; Duke University Medical Center, Durham, NC每Jeffrey Crawford, MD, supported by CA47577; Illinois Oncology Research Assoc, Peoria, IL, John W. Kugler, MD, supported by CA35113; Missouri Baptist Medical Center, St. Louis, MO每Alan P Lyss, MD; Mount Sinai Medical Center, Miami, FL每Rogerio Lilenbaum, MD, supported by CA45564; Rhode Island Hospital, Providence, RI每William Sikov, MD, supported by CA08025; Roswell Park Cancer Institute, Buffalo, NY每Ellis Levine, MD, supported by CA02599; Southeast Cancer Control Consortium Inc. CCOP, Goldsboro, NC每James N. Atkins, MD, supported by CA45808; Southern Nevada Cancer Research Foundation CCOP, Las Vegas, NV每John Ellerton, MD, supported by CA35421; SUNY Upstate Medical University, Syracuse, NY每Stephen L. Graziano, MD, supported by CA21060; Syracuse Hematology-Oncology Assoc. CCOP, Syracuse, NY每Jeffrey Kirshner, MD, supported by CA45389; The Ohio State University Medical Center, Columbus, OH每Clara D Bloomfield, MD, supported by CA77658; University of California at San Diego, San Diego, CA每Stephen L Seagren, MD, supported by CA11789; University of California at San Francisco, San Francisco, CA每Alan P. Venook, MD, supported by CA60138; University of Chicago Medical Center, Chicago, IL -Gini Fleming, MD, supported by CA41287; University of Iowa, Iowa City, IA每Gerald Clamon, MD, supported by CA47642; University of Missouri/Ellis Fischel Cancer Center, Columbia, MO每Michael C Perry, MD, supported by CA12046; University of Nebraska Medical Center, Omaha, NE每Anne Kessinger, MD, supported by CA77298; University of North Carolina at Chapel Hill, Chapel Hill, NC每Thomas C. Shea, MD, supported by CA47559; Vermont Cancer Center, Burlington, VT每Hyman B. Muss, MD, supported by CA77406; Wake Forest University School of Medicine, Winston-Salem, NC每David D Hurd, MD, supported by CA03927; Washington University School of Medicine, St. Louis, MO每Nancy Bartlett, MD, supported by CA77440; Western Pennsylvania Cancer Institute, Pittsburgh, PA每Richard K. Shadduck, MD, N/A

    Authors' Disclosures of Potential Conflicts of Interest

    The authors indicated no potential conflicts of interest.

    Acknowledgment

    The investigators wish to thank the data managers, nurses, and clinical research associates for providing excellent and essential support in completing this research. We also thank our patients without whom this research would not have been possible.

    NOTES

    Supported by CA41287, supported by CA33601, supported by CA60138, supported by CA47642, supported by CA45808. The research for CALGB 69901 was supported, in part, by grants from the National Cancer Institute (CA31946) to the Cancer and Leukemia Group B (Richard L. Schilsky, MD, Chairman). The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute.

    Presented in part at the 39th Annual Meeting of the American Society of Clinical Oncology, Chicago, IL, May 31-June 3, 2003.

    Authors' disclosures of potential conflicts of interest are found at the end of this article.

    REFERENCES

    Vogelzang NJ, Stadler WM: Kidney cancer [see comments]. Lancet 352:1691-1696, 1998

    George CM, Stadler WM: The role of systemic chemotherapy in the treatment of kidney cancer, in Figlin R (ed): Kidney Cancer. Norwell, MA, Kluwer Academic Publishers, 2003

    Gnarra JR, Tory K, Weng Y, et al: Mutations of the VHL tumour suppressor gene in renal carcinoma. Nat Genet 7:85-90, 1994

    Herman JG, Latif F, Weng Y, et al: Silencing of the VHL tumor-suppressor gene by DNA methylation in renal carcinoma. Proc Natl Acad Sci U S A 91:9700-9704, 1994

    Turner KJ, Moore JW, Jones A, et al: Expression of hypoxia-inducible factors in human renal cancer: Relationship to angiogenesis and to the von Hippel-Lindau gene mutation. Cancer Res 62:2957-2961, 2002

    Iliopoulos O, Levy AP, Jiang C, et al: Negative regulation of hypoxia-inducible genes by the von Hippel-Lindau protein. Proc Natl Acad Sci U S A 93:10595-10599, 1996

    Wiesener MS, Munchenhagen PM, Berger I, et al: Constitutive activation of hypoxia-inducible genes related to overexpression of hypoxia-inducible factor-1alpha in clear cell renal carcinomas. Cancer Res 61:5215-5222, 2001

    Therasse P, Arbuck SG, Eisenhauer EA, et al: New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst 92:205-216, 2000

    Gleave ME, Elhilali M, Fradet Y, et al: Interferon gamma-1b compared with placebo in metastatic renal-cell carcinoma: Canadian Urologic Oncology Group. N Engl J Med 338:1265-1271, 1998

    Motzer RJ, Mazumdar M, Bacik J, et al: Survival and prognostic stratification of 670 patients with advanced renal cell carcinoma. J Clin Oncol 17:2530-2540, 1999

    Stadler WM, Ratain MJ: Development of target-based antineoplastic agents. Invest New Drugs 18:7-16, 2000

    Rosner GL, Stadler W, Ratain MJ: Randomized discontinuation design: Application to cytostatic antineoplastic agents. J Clin Oncol 20:4478-4484, 2002

    Felder CC, Ma AL, Liotta LA, et al: The antiproliferative and antimetastatic compound L651582 inhibits muscarinic acetylcholine receptor-stimulated calcium influx and arachidonic acid release. J Pharmacol Exp Ther 257:967-971, 1991

    Kohn EC, Felder CC, Jacobs W, et al: Structure-function analysis of signal and growth inhibition by carboxyamido-triazole, CAI. Cancer Res 54:935-942, 1994

    Kohn EC, Alessandro R, Spoonster J, et al: Angiogenesis: Role of calcium-mediated signal transduction. Proc Natl Acad Sci U S A 92:1307-1311, 1995

    Teicher BA, Holden SA, Chen YN, et al: CAI: Effects on cytotoxic therapies in vitro and in vivo. Cancer Chemother Pharmacol 34:515-521, 1994

    Kohn EC, Sandeen MA, Liotta LA: In vivo efficacy of a novel inhibitor of selected signal transduction pathways including calcium, arachidonate, and inositol phosphates. Cancer Res 52:3208-3212, 1992

    Kohn EC, Figg WD, Sarosy GA, et al: Phase I trial of micronized formulation carboxyamidotriazole in patients with refractory solid tumors: Pharmacokinetics, clinical outcome, and comparison of formulations. J Clin Oncol 15:1985-1993, 1997

    Kohn EC, Reed E, Sarosy G, et al: Clinical investigation of a cytostatic calcium influx inhibitor in patients with refractory cancers. Cancer Res 56:569-573, 1996

    Berlin J, Tutsch KD, Hutson P, et al: Phase I clinical and pharmacokinetic study of oral carboxyamidotriazole, a signal transduction inhibitor. J Clin Oncol 15:781-789, 1997

    Heitjan DF: Bayesian interim analysis of phase II cancer clinical trials. Stat Med 16:1791-1802, 1997

    Spiegelhalter DJ, Freedman LS, Blackburn PR: Monitoring clinical trials: Conditional or predictive power Control Clin Trials 7:8-17, 1986

    Friedland D, Leon L, Manola J, et al: Carboxyamidotriazole (CAI) in patients with advanced renal cell cancer (RCC) refractory to immunotherapy: An ECOG phase II trial. Proc Amer Soc Clin Oncol 21:152b, 2002 (abstr 2424)

    Schwartz D, Lellouch J: Explanatory and pragmatic attitudes in therapeutical trials. J Chronic Dis 20:637-648, 1967(Walter M. Stadler, Gary R)