Review. A. Di Leo 1 *, M. Buyse 2 & H. Bleiberg 1. Introduction. Design and main results of the trials

Transcription

1 Review Annals of Oncology 15: , 2004 DOI: /annonc/mdh127 Is overall survival a realistic primary end point in advanced colorectal cancer studies? A critical assessment based on four clinical trials comparing fluorouracil plus leucovorin with the same treatment combined either with oxaliplatin or with CPT-11 A. Di Leo 1 *, M. Buyse 2 & H. Bleiberg 1 1 Jules Bordet Institute, Brussels, Belgium; 2 International Drug Development Institute, Cambridge, MA, USA Received 6 May 2003; revised 19 August 2003; accepted 14 September 2003 Introduction Background: The adequacy of overall survival (OS) as study end point in phase III trials for advanced solid tumors is questionable. The present review highlights the limits of OS as study end point to evaluate the efficacy of new drugs. Methods: Four phase III clinical trials comparing a fluorouracil-based regimen with the same regimen plus either CPT-11 or oxaliplatin in advanced colorectal cancer patients were reviewed. The primary aim of the critical assessment was to explain the lack of OS advantage observed in two of the four trials, despite the presence of increased response rate (RR) and time to progression (TTP). Four possible reasons for the lack of OS benefit (i.e. statistical power, cross-over, magnitude of the effect on RR and TTP, non-tumor-related deaths) were systematically reviewed in the trials, and the detectable 1-year OS difference, assuming a statistical power of 80%, was calculated for each. Results: None of these reasons for the lack of OS advantage in presence of RR and TTP benefits convincingly explained the results of the evaluated trials. Three of the four trials had roughly the same statistical power to detect 1-year OS differences, while the fourth trial was underpowered to detect realistic OS differences. The lack of OS advantage observed in the two oxaliplatin trials is therefore likely fortuitous, and due to lack of statistical power. Conclusions: Although increase in OS remains the ultimate goal of many clinical trials, the choice of OS benefit as a mandatory requirement to register new compounds can lead to a serious underestimation of a drug s real efficacy. Key words: colorectal cancer, overall survival, phase III trials, study end point It is still a matter of debate whether overall survival (OS) represents the most practical end point for evaluating the superiority of an experimental therapy over standard treatment in patients with metastatic tumors. The undeniable advantage of OS as study end point is that it represents the ultimate clinical benefit for the patient, provided that quality of life is not compromised. Moreover, survival time is an objective end point, whereas the evaluation of tumor response and/or tumor progression may be quite dependent upon the evaluation methods and schedules used. Hence, response rate (RR) and time to progression (TTP) are less reliable end points than OS. However, potential differences in OS between the experimental and the control arms of a phase III trial could be masked by the use of active drugs delivered as salvage therapy after tumor progression to the evaluated treatments. In particular, the administration of experimental drugs to patients progressing after having received standard therapy could be a major explanation for the lack of OS advantage in clinical trials of new drugs. In this paper, we discuss the results of four clinical trials dealing with the evaluation of CPT-11 and oxaliplatin as first-line therapies for advanced colorectal cancer patients [1 4]. These trials offer an interesting opportunity to evaluate the effectiveness of OS as study end point in phase III trials. Design and main results of the trials Figure 1 summarizes the study design and the main results of the four phase III trials analyzed [1 4]. Each of the trials had a control arm consisting of fluorouracil plus leucovorin and an experimental arm consisting of the same treatment combined with either oxaliplatin or CPT-11. Fluorouracil was delivered as a continuous infu- *Correspondence to: Dr A. Di Leo, Sandro Pitigliani Medical Oncology Unit, Department of Oncology, Hospital of Prato, Piazza dell Ospedale, Prato, Italy. Tel/Fax: ; adileo@us14.toscana.it 2004 European Society for Medical Oncology

2 546 Figure 1. Trial design and main results. Chrono, chronotherapy; c.i., continuous infusion. Table 1. Detectable 1-year survival difference in the four trials (with statistical power 80%) Study (ref.) No. of patients Observed 1-year OS in the control arm (%) OS, overall survival. sion in the trials by Douillard et al. [1] and de Gramont et al. [3], as a chronotherapy in the trial by Giacchetti et al. [4], and as a bolus injection in the trial by Saltz et al. [2]. The study by Saltz et al. was the only one to include three treatment arms; patients were randomly treated either with fluorouracil plus leucovorin or with the same regimen combined with CPT-11, or with CPT-11 alone [2]. While the two CPT-11 trials showed a statistically significant benefit in favor of the experimental arm in terms of RR, TTP and OS [1, 2], in the oxaliplatin trials the advantage associated with the experimental treatment was apparently confined to RR and TTP. OS as an end point in the trials OS was not the primary study end point in any of the trials [1 4]. The main study end points were RR in the trials by Douillard et al. [1] and Giacchetti et al. [4], and TTP in the other two studies [2, 3]. The OS analysis was therefore a secondary objective in all cases, and in none of the trials was the sample size calculated to detect plausible OS differences. In addition, none of the published papers quantify the power of the respective trials to detect a survival advantage. Observed 1-year OS in the experimental arm (%) Giacchetti et al. [4] de Gramont et al. [3] Saltz et al. [2] Douillard et al. [1] Detectable 1-year OS difference (%) Table 1 shows the detectable 1-year OS difference for each study, which is defined as the smallest difference in the 1-year OS rate between the control and the experimental arms that could have been detected with a statistical power of 80%. To perform this calculation, survival was assumed to be exponentially distributed, as is approximately the case in advanced colorectal cancer. Thus, the 1-year OS differences could easily be transformed into overall hazard ratios. The study sample size, as well as the 1-year OS rate of the control arm, were taken into account in the calculations as well. The detectable 1-year OS difference was 18% in the trial by Giacchetti et al. [4]. This difference seems too large to be achievable with the investigational treatment, indicating that the trial was clearly underpowered to detect realistic differences in OS between the two treatment arms. The remaining three trials all had a power of 80% to detect a difference of 13% in the 1-year OS rate. The observed 1-year OS benefit was very similar in the three studies (8% in both the de Gramont and Saltz trials, 10% in the Douillard trial), even though this difference reached statistical significance in only two of them [1, 2]. At least three reasons could explain the lack of a survival benefit in the presence of an improved RR or TTP: (i) an imbalance between the control and the experimental arms in the proportion of patients

3 547 Table 2. Second-line therapies (% of patients treated with each regimen) Second-line treatment CPT-11 trials Oxali trials Douillard [1] Saltz [2] Giacchetti [4] de Gramont [3], not available. (n = 187) CPT-11 (n = 198) (n = 226) CPT-11 (n = 231) (n = 100) Oxali (n = 100) (n = 210) Oxaliplatin <5 < CPT Second surgery Oxali (n = 210) Table 3. Increase in RR and TTP CPT-11 trials Oxali trials Douillard [1] Saltz [2] Giacchetti [4] de Gramont [3] CPT-11 CPT-11 Oxali Oxali % OR (ITT) % CR <1 1.4 Median response duration (months) ?? Median TTP (months) RR, response rate; TTP, time to progression; OR, objective response; CR, complete response; ITT, intention-to-treat analysis. receiving active second-line therapies (including cross-overs to the experimental therapy for patients progressing to control therapy); (ii) the fact that the benefit in terms of RR and TTP was modest and not large enough to lead to OS benefit; and (iii) a higher risk of non-tumor-related deaths in one of the two study arms, which would counterbalance a benefit in terms of tumor-related survival. We will examine each of these possible explanations in turn. (i) Imbalance in use of active second-line therapies Table 2 shows, for each arm of the four trials, the proportion of patients receiving CPT-11, oxaliplatin or second surgery (mainly liver metastasectomy after tumor response to first-line chemotherapy) as second-line therapies. In all of the trials, a significant proportion of patients in the control arm received the experimental drug after progression. The percentage of cross-over was 28% and 57% in the oxaliplatin trials [3, 4] and 31% and 56% in the CPT-11 trials [1, 2]. The proportion of patients receiving CPT-11 as second-line therapy in the oxaliplatin trials, or the proportion of patients treated with second-line oxaliplatin in the CPT-11 trials, was well balanced between the two study arms. In the trial by Giacchetti et al. [4], no data regarding the use of second-line CPT-11 were reported. All in all, similar proportions of patients in the control arms were crossed-over in the oxaliplatin trials and in the CPT-11 trials, and therefore the impact of second-line therapies on OS should be similar. A difference between the study arms in the rate of second surgery might be another potential explanation for the lack of statistically significant OS benefit in the two oxaliplatin trials. The information on second surgery is lacking in the CPT-11 trials, while in the two other studies, a slightly higher percentage of patients in the oxaliplatin arms underwent liver metastases resection compared with patients in the control arms. This difference may be due to the increase in RR observed with oxaliplatin, which led more frequently to a radical liver metastasectomy. Nevertheless, the increased rate of liver metastectomy observed in the experimental arms of the two oxaliplatin trials did not lead to a statistically significant increase in OS. (ii) Benefit in RR and TTP not large enough to lead to OS benefit Table 3 shows, for each trial, the percentage of RR according to an intention-to-treat analysis, the percentage of complete responses, as well as the median response duration and the median TTP (in months). The absolute increase in RR associated with the use of oxaliplatin was 28% and 37% in the Giacchetti and de Gramont trials, respectively, while the increase in median TTP was 2.6 and 2.8 months [3, 4]. The absolute increase in RR associated with the use of CPT-11 was 13% and 22%, and the prolongation in median TTP was of 2.3 and 2.7 months in the Douillard and Saltz trials, respectively [1, 2]. The percentage of complete responses was <5% in all arms, and the response duration was similar across the treatment arms of all trials. Thus, the benefits in terms of RR and TTP associated with the use of either oxaliplatin or CPT-11 were substantial and similar for both investigational agents, and there-

4 548 fore their impact on OS, save for chance, should be about the same. (iii) Increased risk of non-tumor-related deaths Non-tumor-related deaths may be due to treatment toxicity or to accidental causes. The percentage of accidental deaths is unknown in the four trials, but there is no reason to assume that it was substantially different between the study arms of each trial, and across the four trials. The percentage of toxic deaths was <1% in each arm. It therefore seems unlikely that the lack of OS benefit in the oxaliplatin trials was related to an increased risk of nontumor-related deaths. Conclusion: are we seeing chance at play? None of the above reasons seems sufficient to explain a difference in survival outcome between the CPT-11 and the oxaliplatin studies. It may well be fortuitous that the advantage reached statistical significance in only two of the four trials; the apparently different conclusions about the impact of CPT-11 and oxaliplatin on OS could largely be due to chance as well. Notably, the Food and Drug Administration (FDA) recently approved CPT-11 as first-line therapy for advanced colorectal cancer, but not oxaliplatin. Based on the same evidence, the European Agency for the Evaluation of Medicinal Products (EMEA) has approved both of them in this indication. There are therefore differences in the criteria used for drug approval in the USA and in Europe. Differences in guidelines for approval of a new anticancer treatment in the USA and Europe The FDA and the EMEA accept clinical trials, wherever they are conducted, as critical data for marketing authorization, assuming the trials characterize drug activity and safety in a clinically relevant and statistically valid manner. This means that studies must be conducted ethically, at least to the level delineated in the Declaration of Helsinki; adequate assessment of the drug profile and protection of the subjects involved in the study must be ensured by having trials conducted by competent investigators who follow internationally accepted Good Clinical Practice standards; the data need to be validated; and both the FDA and the EMEA insist upon the right to verify the data at the investigational site at the source document level. The FDA requirements for the approval of new drugs for the treatment of colon and rectal cancer were discussed at the Oncologic Drugs Advisory Committee (ODAC) meeting of 19 April As far as advanced metastatic disease is concerned, the FDA and ODAC stated that a favorable effect on survival and quality of life in well-controlled studies was an appropriate basis for approval of a new cancer drug. Improvement in performance status and weight gain without edema were easily assessable measurements of quality of life, and would be convincing end points in randomized controlled studies. A trial with a highly significant improvement in quality of life and immature survival data could be considered as a basis for approval. Objective RRs were not deemed to be satisfactory surrogates for quality of life or survival, nor were response duration or TTP [5]. The EMEA guidelines for the evaluation of anticancer products in human beings, which came into operation in March 1997, and were modified in November 2001, stated that to study the effect of a new agent, the appropriate end points of assessment were, in order of importance: progression-free survival, OS, RR, symptom control; and quality of life. In addition, the results must be shown to be unequivocally due to the new agent, and the selected primary end point used for assessment should be justified based on solid preclinical and clinical phase I/II data, with the study designed accordingly [6]. Should we reassess our efficacy end points for trials in advanced disease? The trials discussed above illustrate some of the difficulties of using OS as an end point in advanced colorectal cancer, although the limit of the present work is that our analysis refers exclusively to four well-defined clinical trials run in the field of first-line treatment for advanced colorectal cancer. Nevertheless, difficulties that exist in considering OS as a main study end point have also been highlighted in the field of breast cancer [7]. In this setting, seven randomized trials have recently attempted to demonstrate OS benefits owing to the use of taxanes in the first-line treatment of advanced breast cancer. However, the reported trials were all lacking appropriate statistical power to demonstrate realistic OS differences between the study arms. Moreover, the frequent use of taxanes as a salvage treatment in the control arm contributed to further decrease the sensitivity of these trials to detect plausible OS benefits associated with the use of taxanes [7]. Of note, only one of the seven trials showed a statistically significant OS benefit in favor of the taxane arm, although the low rate of patients from the control arm who received taxanes as salvage therapy might explain, at least in part, the results of this trial [8]. As shown previously, improvements in OS thus remain the cornerstone for approval of new anticancer drugs in the USA, even though benefits to other end points are often chosen as the primary objectives of trials in advanced disease. Table 4 summarizes the pros and cons of the main end points that can be chosen in advanced disease (RR, TTP and OS). The ultimate goal of anticancer drugs is to prolong survival, but this does not automatically imply that OS should be the primary end point of randomized clinical trials. Indeed, OS may be affected not only by the treatment received as first-line therapy, but also by subsequent lines of treatment. As an increasing number of active new drugs are approved, it may become unrealistic to use OS as a meaningful end point. Indeed, the benefit of new drugs on survival will be smaller when these drugs are compared with alternative treatments than when they are compared with best supportive care alone. The alternative would be not to allow cross-over to the other treatment arm; nevertheless, this option seems to be unrealistic, because new and potentially effective drugs should not be denied to patients with advanced tumors. This statement becomes even more important for some forms of advanced cancer, for

5 549 Table 4. Pros and cons of different end points in metastatic disease End point Pros Cons Tumor response Measured early (3 6 months) Responses infrequent Measured easily Insensitive to disease stabilizations (cytostatics) Reflects biological activity Assessment prone to error and/or bias if not reviewed which no treatment alternatives are available (i.e. melanoma, kidney cancer, lung cancer, etc.). Accordingly, pragmatic trials that mimic clinical practice may become more common, i.e. those in which all patients receive the experimental agent, but in a different sequence with standard therapy. These trials would provide data on the efficacy of the evaluated drug as well as on the most appropriate treatment sequence. These considerations have already profoundly modified clinical research in other diseases, such as AIDS, and we believe that a similar evolution is desirable in the field of clinical oncology. If survival remains the end point of interest, trials will have to be much larger, and consequently more expensive, than the trials performed so far in advanced disease. Alternatively, a meta-analysis of those trials exploring the same research question should be performed to demonstrate realistic OS differences. However, the most serious drawback of using survival as the end point in advanced colorectal cancer is the prolonged period of observation required before the analysis is possible. With the number of promising new drugs that are now reaching the clinic, more sensitive end points that are observed earlier than death will have to replace survival in the evaluation process sooner or later. Acknowledgements Assessment can be reviewed blindly by expert committee The authors wish to thank Carolyn Straehle Ph D, for her help with the linguistic revision of the manuscript. Disease not always measurable Limited impact on survival Time to progression Reflects control of disease process Assessment subjective Unaffected by competing risks of death Very sensitive to differences in treatment efficacy Possible impact on survival Closely related to quality of life Assessment potentially biased to allow for change in therapy Assessment can be reviewed only after changes in therapy Overall survival Most meaningful Hard to affect, therefore large sample sizes needed Most objective Measured late Affected by second-line treatments Affected by competing risks Insensitive to short-term benefits References 1. Douillard JY, Cunningham D, Roth AD et al. Irinotecan combined with fluorouracil compared with fluorouracil alone as first-line treatment for metastatic colorectal cancer: a multicenter randomised trial. Lancet 2000; 335: Saltz LB, Cox JV, Blanke C et al. Irinotecan plus fluorouracil and leucovorin for metastatic colorectal cancer. N Engl J Med 2000; 343: de Gramont A, Figer A, Seymour M et al. Leucovorin and fluorouracil with or without oxaliplatin as first-line treatment in advanced colorectal cancer. J Clin Oncol 2000; 18: Giacchetti S, Perpoint B, Zidani R et al. Phase III multicenter randomised trial of oxaliplatin added to chronomodulated fluorouracil leucovorin as first-line treatment of metastatic colorectal cancer. J Clin Oncol 2000; 18: US Department of Health and Human Services, Food and Drug Administration. Center For Drug Evaluation an Research. Guidance for Industry. [On-line] (April 2003, date last accessed). 6. European Agency for the Evaluation of Medicinal Products (EMEA). Note for guidance on evaluation of anticancer medicinal products in man. [On-line] (April 2003, date last accessed). 7. Di Leo A, Buyse M, Bleiberg H. Overall survival is not a realistic end point for clinical trials of new drugs in advanced solid tumors: a critical assessment based on recently reported phase III trials in colorectal and breast cancer. J Clin Oncol 2003; 21: Jassem J, Pienkowsky T, Pluzanska A et al. Doxorubicin and paclitaxel versus fluorouracil, doxorubicin, and cyclophosphamide as first-line therapy for women with metastatic breast cancer: Final results of a randomised phase III multicenter trial. J Clin Oncol 2001; 19: