Critical Reviews in Oncology/Hematology 93 (2015) 50–59
Progression-free survival as primary endpoint in randomized clinical trials of targeted agents for advanced renal cell carcinoma. Correlation with overall survival, benchmarking and power analysis Emilio Bria a,∗ , Francesco Massari a , Francesca Maines a , Sara Pilotto a , Maria Bonomi a , Camillo Porta b , Sergio Bracarda c , Daniel Heng d , Daniele Santini e , Isabella Sperduti f , Diana Giannarelli f , Francesco Cognetti g , Giampaolo Tortora a , Michele Milella g a
Medical Oncology, Azienda Ospedaliera Universitaria Integrata, University of Verona, Verona, Italy b Medical Oncology, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy c Medical Oncology Ospedale San Donato, Istituto Toscano Tumori, Arezzo, Italy d Medical Oncology, University of Calgary, Tom Baker Cancer Center, Calgary, Alberta, Canada e Medical Oncology, University Campus Bio-medico, Roma, Italy f Biostatistics, Regina Elena National Cancer Institute, Roma, Italy g Medical Oncology, Regina Elena National Cancer Institute, Roma, Italy Accepted 8 August 2014
Contents 1. 2.
3.
4. 5.
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Outcome definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Data extraction and synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Search results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Correlation analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Benchmarking analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Power analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conflicts of interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reviewers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51 51 51 51 52 52 52 53 55 56 57 57 58 58 58 59
Abstract Purpose: A correlation, power and benchmarking analysis between progression-free and overall survival (PFS, OS) of randomized trials with targeted agents or immunotherapy for advanced renal cell carcinoma (RCC) was performed to provide a practical tool for clinical trial design. ∗ Corresponding author at: Medical Oncology, Azienda Ospedaliera Universitaria Integrata, University of Verona, P.zza L.A. Scuro 10, 37124 Verona, Italy. Tel.: +39 0458128502; fax: +39 0458128140. E-mail addresses: [email protected] (E. Bria), [email protected] (F. Massari), [email protected] (F. Maines), [email protected] (S. Pilotto), [email protected] (M. Bonomi), [email protected] (C. Porta), [email protected] (S. Bracarda), [email protected] (D. Heng), [email protected] (D. Santini), [email protected] (I. Sperduti), [email protected] (D. Giannarelli), [email protected] (F. Cognetti), [email protected] (G. Tortora).
http://dx.doi.org/10.1016/j.critrevonc.2014.08.001 1040-8428/© 2014 Elsevier Ireland Ltd. All rights reserved.
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
51
Results: For 1st-line of treatment, a significant correlation was observed between 6-month PFS and 12-month OS, between 3-month PFS and 9-month OS and between the distributions of the cumulative PFS and OS estimates. According to the regression equation derived for 1st-line targeted agents, 7859, 2873, 712, and 190 patients would be required to determine a 3%, 5%, 10% and 20% PFS advantage at 6 months, corresponding to an absolute increase in 12-month OS rates of 2%, 3%, 6% and 11%, respectively. Conclusions: These data support PFS as a reliable endpoint for advanced RCC receiving up-front therapies. Benchmarking and power analyses, on the basis of the updated survival expectations, may represent practical tools for future trial’ design. © 2014 Elsevier Ireland Ltd. All rights reserved.
Keywords: Correlation; Overall survival; Progression-free survival; Renal cell carcinoma
1. Introduction
2. Materials and methods
Since 2006, many active targeted agents have been approved for the treatment of advanced renal cell carcinoma (RCC), leading to an important improvement in RCC management, as compared with the past [1]. With the exception of temsirolimus [2], targeted agents have been registered on the basis of an advantage in progression free survival (PFS), without significant differences in overall survival (OS). The lack of OS benefit in individual studies may be, in part, explained by cross-over issues and by the availability of multiple subsequent lines of active therapy, which dilute out potential OS advantages [3]. Nevertheless, OS remains the gold standard for the evaluation of clinical efficacy of an experimental drug, because it is the only end-point able to directly demonstrate a relevant clinical benefit and is completely independent of investigator biases, including timing of disease assessment. Although objective, OS requires large patient populations and long follow-up, with higher study costs and longer duration, potentially delaying patients’ access to new active treatments. Moreover, Broglio and Berry demonstrated that, in many clinical trials for metastatic diseases, a long median postprogression survival may confound OS comparisons [4]. In such cases, a lack of statistical significance in OS differences does not imply an actual lack of improvement, since OS endpoints may not accurately reflect the benefit of the agent under investigation. Many studies have been conducted in solid tumors to assess whether PFS is indeed a reliable surrogate end-point, regardless of disease and setting [5–7], particularly in modern age, in which small benefits at high costs will no longer be affordable [8]. This concept especially applies to the development of new drugs with a mechanism of action similar to that of already approved agents, as I happens with agents targeting the VEGF/VEGFR axis or the mTOR pathway in metastatic RCC, a disease setting in which a plethora of active agents is already available across multiple lines of treatment. In order to provide a practical tool for clinical trial design in such context, we performed a pooled correlation analysis between PFS and OS according to treatment strategy using a meta-analytic approach and a power and benchmarking analysis of randomized clinical trials conducted in the setting of advanced RCC.
The analysis was conducted according to 4 pre-specified steps: (1) definition of the question the analysis was designed to answer; (2) definition of the trial selection criteria; (3) definition of the search strategy; and (4) detailed description of the statistical methods. 2.1. Outcome definition The analysis was conducted to assess if a correlation between PFS and OS may be demonstrated in clinical trials evaluating specific treatments (targeted therapies, immunotherapy and placebo) for advanced RCC. 2.2. Data extraction and synthesis Full reports and trials’ updates were gathered through Medline (PubMed), American Society of Clinical Oncology (ASCO), European Society for Medical Oncology (ESMO), European Association of Urology (EAU) and American Urological Association (AUA) website searches. Keywords used for searching were: renal cell carcinoma, RCC, immunotherapy, cytokines, phase 3, and randomized, tyrosine kinase inhibitors. Furthermore, meetings lectures having RCC as the topic were checked. All phase 3 randomized trials published in peer-reviewed journals or presented at the ASCO, ASCO-Genitourinary Symposium, ECCO, and ESMO meetings up to December 31st 2013, accruing either treatment-naïve or pretreated advanced RCC patients to receive targeted agents, immunotherapy or placebo were considered eligible. Trials examining first-line targeted agent administration after only one previous systemic treatment consisting of cytokines (as in the TARGET [9], pazopanib registration [10], and AXIS [11] trials) were included among first-line trials. As a mandatory entry criterion, both PFS and OS data (as a direct report or extracted by curves) with at least a 12-month follow-up must have been reported. Given the purpose of the current analysis (correlation between efficacy outcomes), randomized phase II trials were excluded. Trial arms examining chemotherapy, radiotherapy, experimental combinations or drugs without Food and Drug Administration (FDA) and/or European Medical Agency (EMA) regulatory approval were
52
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
considered not eligible. The last available update of each trial was considered as the original source. Monthly PFS and OS rates from month 1 to month 12 were extracted from publications or Kaplan–Meier curves, by 5 different investigators (E.B., F.M., F.Ma., S.P., and M.M.). Treatment arms were merged according to 3 groups: (1) targeted agents, (2) immunotherapy, and (3) placebo. Moreover, cumulative monthly PFS and OS rates were determined using a weighted-average approach for each of the 3 groups. Correlation analysis between 3-month PFS and 9month OS or 6-month PFS and 12-month OS, respectively, was accomplished according to parametric (Pearson’s r, with 95% confidence intervals, CI) and non-parametric (Spearman’s Rho and Kendall’s Tau) coefficients [12]. A regression equation was calculated according to the regression analysis (parametric R2 ). In order to assess if the eventual correlation between PFS and OS should be considered disease-related or setting-related (first-line, second-line, etc.) the same correlation analysis was accomplished taking into account those trials (and/or arms) whereas patients previously treated with targeted agents were subsequently randomized to receive further targeted therapies. In addition, an exploratory correlation analysis by extracting trials’ hazard ratio was performed, given this approach is generally considered as the preferred statistics because it summarizes the treatment effect under weak assumptions. In order to provide a benchmark for the design of new 1stline clinical trials with updated and more reliable hypotheses and in order to estimate the overall impact of the introduction of targeted agents on PFS/OS endpoints in metastatic RCC, cumulative PFS and OS rates at 1–12 months’ time-points were determined, according to a patients’ weighted-average fashion, for all the 3 treatment groups. A power-analysis-model in order to estimate the patients’ sample necessary to determine 3%, 5%, 10% and 20% survival gains (alpha-error 0.05, power 0.80) was developed according to the regression equation for the targeted agents group. Data were separately analyzed by five investigators (E.B., F.Ma., S.P., I.S. and D.G.), by using a licensed MedCalc software (Version 11.0).
3. Results 3.1. Search results The detailed search flow diagram relative to first-line clinical trials is shown in Fig. 1. Of the 107 potential trials identified, 27 (including 13,106 patients) were considered for the analysis of the correlation between PFS and OS; of these, 19 trials (Table 1) were evaluable for both PFS and OS (10,065 patients) and employed either a targeted agent (10 trials, 7236 patients) [2,9–18] or immunotherapy (9 trials, 2829 patients) [19–27]. Three arms (accounting for a total of 802 patients) were excluded because of
non-approved drugs (i.e. retinoic acid and 5-fluorouracil). Overall, 8791 patients were evaluable for the analysis; of these, 4900, 3195 and 696 received targeted agents, immunotherapy or placebo, respectively, as first-line treatment (Fig. 1). Trials examining first-line targeted agent administration after only one previous systemic treatment consisting of cytokines (as in the TARGET [9], pazopanib registration [10], and AXIS [11] trials) were included among first-line trials. Regard to AXIS trial [11], only patients who had previously received cytokines before being treated with targeted agents (251 pts) were included in the first-line analysis. Three phase III clinical trials including 1651 patients were considered for the exploratory analysis in the secondline setting [11,28,29] (Supplementary Table S1). For the AXIS trial [11], only patients previously treated with sunitinib (389 pts) were included in the analysis, because PFS and OS Kaplan–Meier curves were lacking for those previously receiving bevacizumab or temsirolimus. Supplementary Table S1 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j. critrevonc.2014.08.001.
3.2. Correlation analysis In the first-line setting, data were analyzed according to individual treatment arms (targeted agents, immunotherapy, or placebo). For targeted agents, a statistically significant correlation was found between 3-month PFS and 9-month OS, with a Pearson coefficient of 0.82 (95% CI 0.53–0.93, p = 0.0002), a R2 of 0.67 and significant non-parametric coefficients (Rho 0.59, 95% CI 0.11–0.84, p = 0.001; Tau 0.47, 95% CI −0.02 to 0.81, p = 0.015) (Fig. 2A). A significant correlation was also observed between 6-month PFS and 12-month OS, with a Pearson coefficient of 0.85 (95% CI 0.61–0.95, p < 0.0001), a R2 of 0.73, and significant non-parametric coefficients (Rho 0.69, 95% CI 0.28–0.89, p = 0.003; Tau 0.55, 95% CI 0.13–0.87, p = 0.0049; Fig. 2B). For immunotherapy, a strong correlation was found between 3-month PFS and 9-month OS, with a Pearson coefficient of 0.83 (95% CI 0.60–0.93, p < 0.0001), a R2 of 0.63, and significant non-parametric coefficients (Rho 0.80, 95% CI 0.55–0.92, p < 0.0001; Tau 0.61, 95% CI 0.38–0.80, p = 0.0003; Fig. 2C); an equally significant correlation was found between 6-month PFS and 12-month OS, with a Pearson coefficient of 0.84 (95% CI 0.63–0.93, p < 0.0001), a R2 of 0.71, and significant non-parametric coefficients (Rho 0.85, 95% CI 0.64–0.94, p < 0.0001; Tau 0.69, 95% CI 0.49–0.88, p < 0.0001; Fig. 2D). Despite the low number of treatment arms identified (only 3), a significant correlation was observed between 3-month PFS and 9-month OS (Pearson 0.99, p = 0.03) and between 6-month PFS and 12-month OS (Pearson 0.99, p = 0.018) in placebo groups (Supplementary Figure S1).
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
53
Fig. 1. Flow diagram for first-line trials’ selection.
Supplementary Figure S1 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j. critrevonc.2014.08.001. With regard to the exploratory analysis of second-line treatments, no significant correlations according to parametric and non- parametric coefficients were found between 3-month PFS and 9-month OS (Pearson 0.50, p = 0.667; R2 0.25; Rho 0.50, p = 0.79; Tau 0.50, p = 1.0) and between 6month PFS and 12-month OS (Pearson 0.42, p = 0.719; R2 0.182; Rho 0.00, p = 1.0; Tau 0.00, p = 1.0) (data not shown). With regard to the correlation analysis on the basis of the extracted HRs, 10 and 4 trials report both PFS and OS estimates for the targeted agents and the immunotherapy
populations. Although the significant data attrition may certainly limit these results, no significant correlations were found between PFS and OS for the targeted agents (Pearson 0.45, p = 0.19; R2 0.20; Rho 0.78, p = 0.16; Tau 0.34, p = 0.19) and the immunotherapy (Pearson 0.63, p = 0.36; R2 0.66; Rho 0.80, p = 0.20; Tau 0.66, p = 0.30) trials’ sample. 3.3. Benchmarking analysis Cumulative survival rates for each treatment group at every month from treatment start up to 12 months were extracted and pooled according to a weighted-average calculation, for the 1st-line setting; a punctual averaged estimate
54
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
Fig. 2. Regression line according to correlation time-points (3 months PFS and 9 months OS: Panels A and C; 6 months PFS and 12 months OS: Panels B and D) and treatment group (targeted agents: Panels A and B; immunotherapy: Panels C and D).
Fig. 3. Cumulative survival rates (PFS: Panel A; OS: Panel B).
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
55
Table 1 Characteristics of the eligible first-line randomized clinical trials included in the analysis.
Targeted agents
Author
Arms
No. pts (per arm)
PFS: HR (95% CI)
OS: HR (95% CI)
Hudes et al. [2]
Temsirolimus Temsirolimus + IFN-␣ IFN-␣
209 210 207
0.66b (0.53–0.81) p < 0.001
Motzer et al. [13]
Sunitinib IFN-␣ Bevacizumab + IFN-␣ IFN-␣ Bevacizumab + IFN-␣ IFN-␣ Sorafenib Placebo Pazopanib Placebo Axitinib Sorafenib Pazopanib Sunitinib Tivozanib Sorafenib Temsirolimus + bevacizumab IFN + bevacizumab
375 375 327 322 369 363 451 452 290 145 126a 125a 557 553 260 257 400 391
0.53 (0.45–0.64) p < 0.001 0.63 (0.52–0.75) p < 0.001 0.71 (0.61–0.83) p < 0.001 0.44 (0.35–0.55) p < 0.001 0.46 (0.34–0.62) p < 0.001 0.46a (0.31–0.67) p < 0.001 1.05 (0.90–1.22) 0.79 (0.63–0.99) p = 0.042 1.07 (0.89–1.28) p = 0.759
0.73b (0.58–0.92) p = 0.009 0.96c (0.76–1.20) p = 0.70 0.82 (0.67–1.00) p = 0.51 0.91 (0.76–1.10) p = 0.336 0.86 (0.73–1.01) p = 0.069 0.88 (0.74–1.04) p = 0.146 0.91 (0.71–1.16) p = 0.224 0.81a (0.55–1.19) p = 0.144 0.91 (0.76–1.08) p = 0.28 1.24 (0.95–1.62) p = 0.105 1.04 (0.85–1.26) p = 0.638
IFN-␣-2a IFN-␣-2a + 13-CRA r-IFN-␣2c r-IFN-␣2c + r-IFN-␥ IL-2 + IFN HD-IL-2 IFN-␣-2a IFN-␣-2a + 13-CRA IFN-␣ IFN-␣ + IL-2 + 5-FU Autologous CIK IL-2 + IFN␣-2a IL-2 i.v. + IFN-␣ IL-2 s.c. + IFN-␣ IL-2 IFN-␣ IL-2 + IFN-␣ IFN-␥-1b Placebo
161 159 54 48 96 96 145 139 502 504 74 74 80 75 138 147 140 98 99
0.67 (0.53–0.85) p = 0.007 p = 0.98
0.78 (0.61–1.0) p = 0.048 p = 0.73
p = 0.082
0.81 p = 0.211
p = 0.13
p = 0.26
1.02 (0.89–1.16) p = 0.81 0.88 (0.84–0.93) p < 0.001 1.16 (0.83–1.63) p = 0.188 p = 0.01
1.05 (0.90–1.21) p = 0.55 0.58 (0.48–0.69) p < 0.001 1.20 (0.78–1.83) p = 0.202 p = 0.55
p = 0.49
p = 0.52
Escudier et al. [15] Rini et al. [14] Escudier et al. [9] Sternberg et al. [10] Rini et al. [11] Motzer et al. [16] Motzer et al. [17] Rini et al. [18] Immunotherapy
Aass et al. [19] De Mulder et al. [20] Mc Dermott et al. [21] Motzer et al. [22] Gore et al. [23] Liu et al. [24] Negrier et al. [25] Negrier et al. [26]
Gleave et al. [27]
CI, confidential interval; HR, hazard ratio; pts, patients; PFS, progression-free survival; OS, overall survival; 13-CRA, 13-cis retinoic acid; r, recombinant; IFN, interferon; HD, high dose; 5-FU, 5-fluorouracil; CIK, cytokine-induced killer; i.v., intravenous; s.c., subcutaneous. a In patients previously treated with cytokines. b Comparing temsirolimus alone with interferon alone. c Comparing the combination of temsirolimus plus interferon with interferon alone.
for each group was then calculated for both PFS and OS at every month. Outcome probability according to treatment group is shown in Fig. 3A and B for PFS and OS, respectively, with punctual estimates at 3, 6, 9 and 12 months reported in the embedded table. A highly significant correlation between the distributions of the calculated PFS and OS estimates according to treatment groups was found for targeted agents and immunotherapy arms, with Pearson coefficients of 0.98 (95% CI 0.94–0.99, p < 0.0001), 0.95 (95% CI 0.85–0.98, p < 0.0001), respectively (Fig. 4A and B), and to a lesser extent for the
placebo group (Pearson 0.87, 95% CI 0.60–0.96, p = 0.0002) (Fig. 4C). 3.4. Power analysis According to the regression equation derived for targeted agents (y = 33.2949 + 0.6382x), a power-analysis model was developed in order to estimate the patient sample size necessary to determine a 3%, 5%, 10% and 20% PFS advantage at 6 months in the context of 1st-line setting. With a power of 80%, and an alpha error of 0.05, 7859, 2873, 712, and 190
56
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
Fig. 5. Power analysis according to the regression equation for targeted agents (y = 13.2598 + 0.9484x); power 80%, alpha-error 0.05. The determined patients’ sample, corresponds to a 12-month OS difference of 19%, 9%, 4% and 2%, respectively.
Fig. 4. Regression line between PFS and OS cumulative survival rates, according to treatment group (targeted agents: Panel A; immunotherapy: Panel B; placebo: Panel C).
patients would be required; these figures, in turn, would correspond to an absolute increase in 12-month OS rates of 2%, 3%, 6% and 11%, respectively (Fig. 5).
4. Discussion Here we formally analyzed the correlation between PFS and OS in the context of randomized trials of either targeted
agents or immunotherapy conducted in advanced RCC, using a meta-analytic approach. OS is traditionally considered the ‘gold standard’ of efficacy endpoints, above all because it can be objectively established and, therefore, not subject to biases introduced by investigator assessment. However, OS is a difficult endpoint to pursue, because it needs expensive clinical trials with huge numbers of patients and lengthy follow-up; moreover, the use of further lines of therapy and crossover strategies, though necessary for ethical reasons and to guarantee patients’ access to active treatments as they become available, may dilute and obscure eventual survival benefits, particularly in first-line trials. For all these reasons, reliable surrogate endpoints are clearly required [30], not only to identify new active agents, but also to make them rapidly available, particularly at times when the whole community, including health care stakeholders, regulatory agencies, and patients alike, strongly calls for the rapid development of novel treatment opportunities. This necessity is tempered by the need to restrict wide availability only to those drugs that really impact on disease course and from which patients could derive a true, clinically significant, benefit. Three lines of evidence support the use of PFS as a surrogate endpoint for OS in advanced RCC: first, a recent retrospective, non-randomized study performed mainly on patients who received targeted agents as first-line therapy (68.6% of all enrolled patients) demonstrated that PFS at 3 and 6 months significantly correlates with OS (p < 0.0001) and suggested that it may replace OS as a reliable intermediate endpoint for measuring the efficacy of new targeted therapies [12]; second, in a recent systematic review [31], encompassing trials of either cytokines or targeted agents, a significant correlation was found between the log transformation of PFS/TTP and OS (correlation coefficient 0.80; p = 0.0001); third, a recent analysis of 1381 patients treated within two large phase 3 trials, evaluating the efficacy of
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
interferon-alpha with or without bevacizumab in the treatment of advanced RCC, demonstrated a strong correlation between PFS (at 3 and 6 months) and OS (p < 0.00001) with an adjusted HR for death of 2.6 for patients who had disease progression at 3 months [32]. Our current work, significantly adds to such body of knowledge and strongly confirms that PFS may be considered an acceptable intermediate endpoint for ultimate survival benefit, particularly in the context of first-line treatment with targeted agents. PFS may thus represent a good choice, especially when multiple therapy options exist, such as in advanced RCC, as it evaluates only the activity of the study drug and it is not affected by subsequent therapies and crossover strategies. However, the lack of correlation between PFS and OS in first-line has been recently observed in TIVO-1 trial where tivozanib demonstrated to improve PFS but not OS [17]. A possible explanation is that OS in this study was confounded by a differential use of subsequent targeted cancer therapies; in fact, not only a one-way cross-over to tivozanib was allowed for the patients progressing with sorafenib but also a greater proportion of patients in the sorafenib arm had received next-line targeted treatments (63% vs 13% in the tivozanib arm). However, other factors such as reduction of relationship between these two endpoints when an active, targeted therapy comparator is used cannot be excluded, as also observed in INTORSECT and AXIS trials [11,29]. The correlation between PFS and OS is far less stringent in the secondand subsequent-line setting. Indeed, in contrast with firstline results, our exploratory analysis performed on patients previously treated with targeted agents (second-line setting), failed to demonstrate a significant correlation between PFS and OS. Although such discrepancy is currently difficult to explain, our findings suggest that this phenomenon may be setting-, rather than disease-related. The breakdown of relationship between PFS and OS in the second-line treatment has been recently observed in INTORSECT trial comparing temsirolimus with sorafenib in sunitinib-refractory metastatic RCC patients [29]. Results showed that there was no significant difference in PFS between these two agents but paradoxically, a longer OS has been observed in patients treated with sorafenib. Benchmarking and power analysis provide meaningful tools for clinicians who aim at designing future trials with updated and more reliable hypotheses. Such “statistical” evidence could contribute to changing the methodological approach, enhancing the role of PFS as a useful endpoint for efficacy and limiting the use of OS in the context of first-line RCC trials. Benchmarking analysis conducted on cumulated survival rates extracted from the analyzed trials also provides the first unequivocal demonstration of the impact of targeted agents on advanced RCC survival: indeed, the confidence intervals of the landmark analysis of patients treated with targeted agents do not overlap with those of patients receiving either immunotherapy or placebo, particularly from the 9-month timepoint onward. Although this is not a log-rank analysis of Kaplan–Meier curves, it does effectively represent the changing landscape
57
in the natural history of RCC, highlighting how the introduction of targeted agents has impacted on OS expectations, whereas cytokines have failed to improve survival in the long run. In addition, power analysis helps estimate how many patients could be spared if trials were designed on the basis of PFS (Fig. 5). These calculations may be useful to design future trials more effectively, avoiding what has been recently observed with the phase III AGILE trial [33], which, because of an overestimation of differences in treatment effects, was underpowered to demonstrate a potentially relevant difference in efficacy between axitinib and sorafenib in the first-line setting.
5. Conclusions Overall, despite the potential trials selection, the reporting biases and the missing data our data add significantly to current literature, supporting the use of PFS as a reliable endpoint for OS in first-line trials of targeted therapy for advanced RCC and provide useful tools for the design of future clinical trials in this setting, based on more realistic and updated survival expectations. However, the demonstration of a correlation between PFS and OS using aggregate data and a meta-analytic approach is not sufficient to considering PFS as a surrogate endpoint for OS, since it is not able to satisfy all four the Prentice’s criteria [34]. Particularly, the lack of confirmatory individual patient data, does not allow to establish the conditional independence of the impact of the treatment on the true endpoint, given the surrogate endpoint. Moreover, heterogeneity of analyzed clinical trials, including drugs with different mechanisms of action and consequently, with a potentially different effect on median PFS and OS and immature overall survival data from some studies require caution in interpretation of the final results, suggesting the need of further confirming analyses.
Conflicts of interest E. Bria had advisory role for Eli-Lilly, Pfizer, Amgen and speaker’s fee from Eli-Lilly; C. Porta has consultancies for Pfizer, GSK, Roche, Bayer-Schering, Novartis, Astellas, Aveo, Boehringer-Ingelheim, speaker’s fee from Pfizer, GSK, Roche, Bayer-Schering, Novartis, Astellas and research funding from Bayer-Schering, Novartis. Clinical trial support: Pfizer; S. Bracarda is advisory board member for Pfizer, Novartis, Bayer, Boehringer-Ingelheim, GSK, Aveo/Astellas, and received honoraria from: Novartis, GSK, Pfizer; D. Heng has advisory role at Aveo, Pfizer, Novartis, and Bayer; G. Tortora has consultancy role for Novartis, Pfizer and GSK; M. Milella had advisory role for Pfizer, Novartis, Eli-Lilly and received speaker’s fee from Pfizer, GSK, Novartis, Bayer, Astra-Zeneca; F. Massari, F. Maines, S. Pilotto, M. Bonomi, I. Sperduti, D. Giannarelli and F. Cognetti declare no competing interests.
58
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
Reviewers Dr. Valder Torri, Oncology Department, Mario Negri Institute, Via G. La Masa 19, I-20156 Milan, Italy. Prof. Walter M. Stadler, The University of Chicago, Department of Medicine, Section of Hematology/Oncology, AMB I214 (MC 2115), 5841 South Maryland Avenue, Chicago, IL 60637, United States. Dr. Roberto Iacovelli, PhD Student, Sapienza University of Rome, Department of Radiology, Oncology and Human Pathology, Oncology Unit B, Via Regina Elena 324, I-00161 Rome, Italy. Acknowledgments This work was supported by grants of the Italian Association for Cancer Research (AIRC) (MFAG 14282, Investigator Grant 11930, 5X1000 12182 and 12214) and of Italian Ministry of Education, Universities and Research (MIUR) – Interest National Research Project (PRIN) (2009X23L78 005) without any implication in study design, data collection, data analysis, data interpretation or writing of the paper. References [1] Escudier B, Albiges L, Sonpavde G. Optimal management of metastatic renal cell carcinoma: current status. Drugs 2013;73(5):427–38. [2] Hudes G, Carducci M, Tomczak P, et al. Temsirolimus, interferon alfa, or both for advanced renal-cell carcinoma. N Engl J Med 2007;356(22):2271–81. [3] Zietemann VD, Schuster T, Duell TH. Post-study therapy as a source of confounding in survival analysis of first-line studies in patients with advanced non-small-cell lung cancer. J Thorac Dis 2011;3(2):88–98. [4] Broglio KR, Berry DA. Detecting an overall survival benefit that is derived from progression-free survival. J Natl Cancer Inst 2009;101(23):1642–9. [5] Burzykowski T, Buyse M, Piccart-Gebhart MJ, et al. Evaluation of tumor response, disease control, progression-free survival, and time to progression as potential surrogate end points in metastatic breast cancer. J Clin Oncol 2008;26(12):1987–92. [6] Gill S, Berry S, Biagi J, et al. Progression-free survival as a primary endpoint in clinical trials of metastatic colorectal cancer. Curr Oncol 2011;18(Suppl. 2):S5–10. [7] Halabi S, Vogelzang NJ, Ou SS, Owzar K, Archer L, Small EJ. Progression-free survival as a predictor of overall survival in men with castrate-resistant prostate cancer. J Clin Oncol 2009;27(17):2766–71. [8] Sullivan R, Peppercorn J, Sikora K, et al. Delivering affordable cancer care in high-income countries. Lancet Oncol 2011;12(10):933–80. [9] Escudier B, Eisen T, Stadler WM, et al. Sorafenib in advanced clear-cell renal-cell carcinoma. N Engl J Med 2007;356(2):125–34. [10] Sternberg CN, Davis ID, Mardiak J, et al. Pazopanib in locally advanced or metastatic renal cell carcinoma: results of a randomized phase III trial. J Clin Oncol 2010;28(6):1061–8. [11] Rini BI, Escudier B, Tomczak, et al. Comparative effectiveness of axitinib versus sorafenib in advanced renal cell carcinoma (AXIS): a randomised phase 3 trial. Lancet 2011;378:1931–9. [12] Heng DY, Xie W, Bjarnason GA, et al. Progression-free survival as a predictor of overall survival in metastatic renal cell carcinoma treated with contemporary targeted therapy. Cancer 2011;117(12):2637–42.
[13] Motzer RJ, Hutson TE, Tomczak P, et al. Sunitinib versus interferon alfa in metastatic renal-cell carcinoma. N Engl J Med 2007;356(2): 115–24. [14] Rini BI, Halabi S, Rosenberg JE, et al. Bevacizumab plus interferon alfa compared with interferon alfa monotherapy in patients with metastatic renal cell carcinoma: CALGB 90206. J Clin Oncol 2008;26(33):5422–8. [15] Escudier B, Pluzanska A, Koralewski P, et al. Bevacizumab plus interferon alfa-2a for treatment of metastatic renal cell carcinoma: a randomised, double-blind phase III trial. Lancet 2007;370(9605): 2103–11. [16] Motzer RJ, Hutson TE, Cella D, et al. Pazopanib versus sunitinib in metastatic renal cell carcinoma. N Engl J Med 2013;369:722–31. [17] Motzer RJ, Nosov D, Eisen T, et al. Tivozanib versus sorafenib as initial targeted therapy for patients with metastatic renal cell carcinoma: results from a phase III trial. J Clin Oncol 2013;31(30):3791–9. [18] Rini BI, Bellmunt J, Clancy J, et al. Randomized phase III trial of temsirolimus and bevacizumab versus interferon alfa and bevacizumab in metastatic renal cell carcinoma: INTORACT trial. J Clin Oncol 2014;32(8):752–9. [19] Aass N, De Mulder PHM, Mickisch GHJ, et al. Randomized phase II/III trial of interferon alfa-2a with and without 13-cis-retinoic acid in patients with progressive metastatic renal cell carcinoma: the European Organisation for Research and Treatment of Cancer Genito-Urinary Tract Cancer Group (EORTC 30951). J Clin Oncol 2005;23(18):4172–8. [20] De Mulder PHM, Oosterhof GON, Bouffioux C, et al. EORTC (30885) randomised phase III study with recombinant interferon alpha and recombinant interferon alpha and gamma in patients with advanced renal cell carcinoma. Br J Cancer 1995;71:371–5. [21] Mc Dermott DF, Regan MM, Clark JI, et al. Randomised phase III trial of high-dose interleukin-2 versus subcutaneous interleukin-2 and interferon in patients with metastatic renal cell carcinoma. J Clin Oncol 2005;23(1):133–41. [22] Motzer RJ, Murphy BA, Bacik J, et al. Phase III trial of interferon alfa2a with or without 13-cis-retinoic acid for patients with advanced renal cell carcinoma. J Clin Oncol 2000;18(16):2972–80. [23] Gore ME, Griffin CL, Hancock B, et al. Interferon alfa-2a versus combination therapy with interferon alfa-2a, interleukin-2, and fluorouracil in patients with untreated metastatic renal cell carcinoma (MRC RE04/EORTC GU 30012): an open-label randomised trial. Lancet 2010;375(9715):641–8. [24] Liu L, Zhang W, Qi X, et al. Randomised study of autologous cytokineinduced killer cell immunotherapy in metastatic renal carcinoma. Clin Cancer Res 2012;18(6):1751–9. [25] Negrier S, Perol D, Ravaud A, et al. Randomised study of intravenous versus subcutaneous interleukin-2 and IFN␣ in patients with good prognosis metastatic renal cancer. Clin Cancer Res 2008;14:5907–12. [26] Negrier S, Escudier B, Lasset C, et al. Recombinant human interleukin2, recombinant human interferon alfa-2a, or both in metastatic renal-cell carcinoma. N Engl J Med 1998;338:1272–8. [27] Gleave ME, Elhilali M, Fradet Y, et al. Inteferon gamma-1b compared with placebo in metastatic renal-cell carcinoma. N Engl J Med 1998;18:1265–71. [28] Motzer RJ, Escudier B, Oudard S, et al. Phase 3 trial of everolimus for metastatic renal cell carcinoma: final results and analysis of prognostic factors. Cancer 2010;116(18):4256–65. [29] Hutson TE, Escudier B, Esteban E, et al. Randomized phase III trial of temsirolimus versus sorafenib as second-line therapy after sunitinib in patients with metastatic renal cell carcinoma. J Clin Oncol 2014;32(8):760–77. [30] Sargent DJ, Hayes DF. Assessing the measure of a new drug: is survival the only thing that matters? J Clin Oncol 2008;26(12):1922–3. [31] Delea TE, Khuu A, Heng DY, Haas T, Soulieres D. Association between treatment effects on disease progression end points and overall survival in clinical studies of patients with metastatic renal cell carcinoma. Br J Cancer 2012;107(7):1059–68.
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59 [32] Halabi S, Rini B, Escudier B, et al. Progression-free survival as a surrogate endpoint of overall survival in patients with metastatic renal cell carcinoma. Cancer 2014;120(1):52–60. [33] Hutson TE, Lesovoy V, Al-Shukri S, et al. Axitinib versus sorafenib as first-line therapy in patients with metastatic renal-cell carcinoma: a randomised open-label phase 3 trial. Lancet Oncol 2013;14(13):1287–94. [34] Prentice RL. Surrogate endpoints in clinical trials: definition and operational criteria. Stat Med 1989;8(4):431–40.
Biography Emilio Bria graduated in medicine and surgery, and got the Medical Oncology Fellowship with full marks and honors at ‘La Sapienza’ University in Rome (Italy). He trained at the Memorial Sloan Kettering Cancer Center, New York (USA), Department of Medicine, Head and Neck/Genitourinary Service, at the Herbert Hirving Comprehensive Cancer Center, Columbia Presbyterian Medical Center, College of Physicians & Surgeons of Columbia University, New York (USA), and at the Monter Cancer Center, Division of Hematology and Oncology, North Shore-Long Island Jewish Health System, New York (USA). He spent 8 years at the Regina Elena National Cancer Institute, Department of Medical Oncology with a grant in Methodology of Research in Clinical
59
Oncology, with a particular focus upon lung and breast cancer. His main areas of research interest are meta-analyses in clinical oncology, surrogate end-points for clinical trials and definition of treatment guidelines, mathematical models in clinical oncology and new trials’ design for targeted agents in lung and breast cancer. He is currently involved in a series of Phase I–III trials, and in full-time involved in many translational and clinical research projects in lung cancer. He won national and international awards and grants and he is authors of several papers published in peer-reviewed journal and authors of book-chapters. He is also member of the editorial boards of international peer-reviewed journals. He also presented original researches or educational lectures at national and international meetings. He is currently member of the Expert panel of the Antiemetic Study Group, on behalf of the MASCC (Multinational Association for Supportive Care in Cancer) and the ESMO (European Society for Medical Oncology), for the development and updated of the Guidelines for Antiemetic treatment. He is also currently member of the Publication Committee of the International Association for the Study of Lung Cancer (IASLC), and member of the Expert Board of the National Oncological Committee for the national regulatory agency (Italian Drug Agency, AIFA).
Progression-free survival as primary endpoint in randomized clinical trials of targeted agents for advanced renal cell carcinoma. Correlation with overall survival, benchmarking and power analysis Emilio Bria a,∗ , Francesco Massari a , Francesca Maines a , Sara Pilotto a , Maria Bonomi a , Camillo Porta b , Sergio Bracarda c , Daniel Heng d , Daniele Santini e , Isabella Sperduti f , Diana Giannarelli f , Francesco Cognetti g , Giampaolo Tortora a , Michele Milella g a
Medical Oncology, Azienda Ospedaliera Universitaria Integrata, University of Verona, Verona, Italy b Medical Oncology, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy c Medical Oncology Ospedale San Donato, Istituto Toscano Tumori, Arezzo, Italy d Medical Oncology, University of Calgary, Tom Baker Cancer Center, Calgary, Alberta, Canada e Medical Oncology, University Campus Bio-medico, Roma, Italy f Biostatistics, Regina Elena National Cancer Institute, Roma, Italy g Medical Oncology, Regina Elena National Cancer Institute, Roma, Italy Accepted 8 August 2014
Contents 1. 2.
3.
4. 5.
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Outcome definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Data extraction and synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Search results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Correlation analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Benchmarking analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Power analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conflicts of interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reviewers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51 51 51 51 52 52 52 53 55 56 57 57 58 58 58 59
Abstract Purpose: A correlation, power and benchmarking analysis between progression-free and overall survival (PFS, OS) of randomized trials with targeted agents or immunotherapy for advanced renal cell carcinoma (RCC) was performed to provide a practical tool for clinical trial design. ∗ Corresponding author at: Medical Oncology, Azienda Ospedaliera Universitaria Integrata, University of Verona, P.zza L.A. Scuro 10, 37124 Verona, Italy. Tel.: +39 0458128502; fax: +39 0458128140. E-mail addresses: [email protected] (E. Bria), [email protected] (F. Massari), [email protected] (F. Maines), [email protected] (S. Pilotto), [email protected] (M. Bonomi), [email protected] (C. Porta), [email protected] (S. Bracarda), [email protected] (D. Heng), [email protected] (D. Santini), [email protected] (I. Sperduti), [email protected] (D. Giannarelli), [email protected] (F. Cognetti), [email protected] (G. Tortora).
http://dx.doi.org/10.1016/j.critrevonc.2014.08.001 1040-8428/© 2014 Elsevier Ireland Ltd. All rights reserved.
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
51
Results: For 1st-line of treatment, a significant correlation was observed between 6-month PFS and 12-month OS, between 3-month PFS and 9-month OS and between the distributions of the cumulative PFS and OS estimates. According to the regression equation derived for 1st-line targeted agents, 7859, 2873, 712, and 190 patients would be required to determine a 3%, 5%, 10% and 20% PFS advantage at 6 months, corresponding to an absolute increase in 12-month OS rates of 2%, 3%, 6% and 11%, respectively. Conclusions: These data support PFS as a reliable endpoint for advanced RCC receiving up-front therapies. Benchmarking and power analyses, on the basis of the updated survival expectations, may represent practical tools for future trial’ design. © 2014 Elsevier Ireland Ltd. All rights reserved.
Keywords: Correlation; Overall survival; Progression-free survival; Renal cell carcinoma
1. Introduction
2. Materials and methods
Since 2006, many active targeted agents have been approved for the treatment of advanced renal cell carcinoma (RCC), leading to an important improvement in RCC management, as compared with the past [1]. With the exception of temsirolimus [2], targeted agents have been registered on the basis of an advantage in progression free survival (PFS), without significant differences in overall survival (OS). The lack of OS benefit in individual studies may be, in part, explained by cross-over issues and by the availability of multiple subsequent lines of active therapy, which dilute out potential OS advantages [3]. Nevertheless, OS remains the gold standard for the evaluation of clinical efficacy of an experimental drug, because it is the only end-point able to directly demonstrate a relevant clinical benefit and is completely independent of investigator biases, including timing of disease assessment. Although objective, OS requires large patient populations and long follow-up, with higher study costs and longer duration, potentially delaying patients’ access to new active treatments. Moreover, Broglio and Berry demonstrated that, in many clinical trials for metastatic diseases, a long median postprogression survival may confound OS comparisons [4]. In such cases, a lack of statistical significance in OS differences does not imply an actual lack of improvement, since OS endpoints may not accurately reflect the benefit of the agent under investigation. Many studies have been conducted in solid tumors to assess whether PFS is indeed a reliable surrogate end-point, regardless of disease and setting [5–7], particularly in modern age, in which small benefits at high costs will no longer be affordable [8]. This concept especially applies to the development of new drugs with a mechanism of action similar to that of already approved agents, as I happens with agents targeting the VEGF/VEGFR axis or the mTOR pathway in metastatic RCC, a disease setting in which a plethora of active agents is already available across multiple lines of treatment. In order to provide a practical tool for clinical trial design in such context, we performed a pooled correlation analysis between PFS and OS according to treatment strategy using a meta-analytic approach and a power and benchmarking analysis of randomized clinical trials conducted in the setting of advanced RCC.
The analysis was conducted according to 4 pre-specified steps: (1) definition of the question the analysis was designed to answer; (2) definition of the trial selection criteria; (3) definition of the search strategy; and (4) detailed description of the statistical methods. 2.1. Outcome definition The analysis was conducted to assess if a correlation between PFS and OS may be demonstrated in clinical trials evaluating specific treatments (targeted therapies, immunotherapy and placebo) for advanced RCC. 2.2. Data extraction and synthesis Full reports and trials’ updates were gathered through Medline (PubMed), American Society of Clinical Oncology (ASCO), European Society for Medical Oncology (ESMO), European Association of Urology (EAU) and American Urological Association (AUA) website searches. Keywords used for searching were: renal cell carcinoma, RCC, immunotherapy, cytokines, phase 3, and randomized, tyrosine kinase inhibitors. Furthermore, meetings lectures having RCC as the topic were checked. All phase 3 randomized trials published in peer-reviewed journals or presented at the ASCO, ASCO-Genitourinary Symposium, ECCO, and ESMO meetings up to December 31st 2013, accruing either treatment-naïve or pretreated advanced RCC patients to receive targeted agents, immunotherapy or placebo were considered eligible. Trials examining first-line targeted agent administration after only one previous systemic treatment consisting of cytokines (as in the TARGET [9], pazopanib registration [10], and AXIS [11] trials) were included among first-line trials. As a mandatory entry criterion, both PFS and OS data (as a direct report or extracted by curves) with at least a 12-month follow-up must have been reported. Given the purpose of the current analysis (correlation between efficacy outcomes), randomized phase II trials were excluded. Trial arms examining chemotherapy, radiotherapy, experimental combinations or drugs without Food and Drug Administration (FDA) and/or European Medical Agency (EMA) regulatory approval were
52
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
considered not eligible. The last available update of each trial was considered as the original source. Monthly PFS and OS rates from month 1 to month 12 were extracted from publications or Kaplan–Meier curves, by 5 different investigators (E.B., F.M., F.Ma., S.P., and M.M.). Treatment arms were merged according to 3 groups: (1) targeted agents, (2) immunotherapy, and (3) placebo. Moreover, cumulative monthly PFS and OS rates were determined using a weighted-average approach for each of the 3 groups. Correlation analysis between 3-month PFS and 9month OS or 6-month PFS and 12-month OS, respectively, was accomplished according to parametric (Pearson’s r, with 95% confidence intervals, CI) and non-parametric (Spearman’s Rho and Kendall’s Tau) coefficients [12]. A regression equation was calculated according to the regression analysis (parametric R2 ). In order to assess if the eventual correlation between PFS and OS should be considered disease-related or setting-related (first-line, second-line, etc.) the same correlation analysis was accomplished taking into account those trials (and/or arms) whereas patients previously treated with targeted agents were subsequently randomized to receive further targeted therapies. In addition, an exploratory correlation analysis by extracting trials’ hazard ratio was performed, given this approach is generally considered as the preferred statistics because it summarizes the treatment effect under weak assumptions. In order to provide a benchmark for the design of new 1stline clinical trials with updated and more reliable hypotheses and in order to estimate the overall impact of the introduction of targeted agents on PFS/OS endpoints in metastatic RCC, cumulative PFS and OS rates at 1–12 months’ time-points were determined, according to a patients’ weighted-average fashion, for all the 3 treatment groups. A power-analysis-model in order to estimate the patients’ sample necessary to determine 3%, 5%, 10% and 20% survival gains (alpha-error 0.05, power 0.80) was developed according to the regression equation for the targeted agents group. Data were separately analyzed by five investigators (E.B., F.Ma., S.P., I.S. and D.G.), by using a licensed MedCalc software (Version 11.0).
3. Results 3.1. Search results The detailed search flow diagram relative to first-line clinical trials is shown in Fig. 1. Of the 107 potential trials identified, 27 (including 13,106 patients) were considered for the analysis of the correlation between PFS and OS; of these, 19 trials (Table 1) were evaluable for both PFS and OS (10,065 patients) and employed either a targeted agent (10 trials, 7236 patients) [2,9–18] or immunotherapy (9 trials, 2829 patients) [19–27]. Three arms (accounting for a total of 802 patients) were excluded because of
non-approved drugs (i.e. retinoic acid and 5-fluorouracil). Overall, 8791 patients were evaluable for the analysis; of these, 4900, 3195 and 696 received targeted agents, immunotherapy or placebo, respectively, as first-line treatment (Fig. 1). Trials examining first-line targeted agent administration after only one previous systemic treatment consisting of cytokines (as in the TARGET [9], pazopanib registration [10], and AXIS [11] trials) were included among first-line trials. Regard to AXIS trial [11], only patients who had previously received cytokines before being treated with targeted agents (251 pts) were included in the first-line analysis. Three phase III clinical trials including 1651 patients were considered for the exploratory analysis in the secondline setting [11,28,29] (Supplementary Table S1). For the AXIS trial [11], only patients previously treated with sunitinib (389 pts) were included in the analysis, because PFS and OS Kaplan–Meier curves were lacking for those previously receiving bevacizumab or temsirolimus. Supplementary Table S1 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j. critrevonc.2014.08.001.
3.2. Correlation analysis In the first-line setting, data were analyzed according to individual treatment arms (targeted agents, immunotherapy, or placebo). For targeted agents, a statistically significant correlation was found between 3-month PFS and 9-month OS, with a Pearson coefficient of 0.82 (95% CI 0.53–0.93, p = 0.0002), a R2 of 0.67 and significant non-parametric coefficients (Rho 0.59, 95% CI 0.11–0.84, p = 0.001; Tau 0.47, 95% CI −0.02 to 0.81, p = 0.015) (Fig. 2A). A significant correlation was also observed between 6-month PFS and 12-month OS, with a Pearson coefficient of 0.85 (95% CI 0.61–0.95, p < 0.0001), a R2 of 0.73, and significant non-parametric coefficients (Rho 0.69, 95% CI 0.28–0.89, p = 0.003; Tau 0.55, 95% CI 0.13–0.87, p = 0.0049; Fig. 2B). For immunotherapy, a strong correlation was found between 3-month PFS and 9-month OS, with a Pearson coefficient of 0.83 (95% CI 0.60–0.93, p < 0.0001), a R2 of 0.63, and significant non-parametric coefficients (Rho 0.80, 95% CI 0.55–0.92, p < 0.0001; Tau 0.61, 95% CI 0.38–0.80, p = 0.0003; Fig. 2C); an equally significant correlation was found between 6-month PFS and 12-month OS, with a Pearson coefficient of 0.84 (95% CI 0.63–0.93, p < 0.0001), a R2 of 0.71, and significant non-parametric coefficients (Rho 0.85, 95% CI 0.64–0.94, p < 0.0001; Tau 0.69, 95% CI 0.49–0.88, p < 0.0001; Fig. 2D). Despite the low number of treatment arms identified (only 3), a significant correlation was observed between 3-month PFS and 9-month OS (Pearson 0.99, p = 0.03) and between 6-month PFS and 12-month OS (Pearson 0.99, p = 0.018) in placebo groups (Supplementary Figure S1).
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
53
Fig. 1. Flow diagram for first-line trials’ selection.
Supplementary Figure S1 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j. critrevonc.2014.08.001. With regard to the exploratory analysis of second-line treatments, no significant correlations according to parametric and non- parametric coefficients were found between 3-month PFS and 9-month OS (Pearson 0.50, p = 0.667; R2 0.25; Rho 0.50, p = 0.79; Tau 0.50, p = 1.0) and between 6month PFS and 12-month OS (Pearson 0.42, p = 0.719; R2 0.182; Rho 0.00, p = 1.0; Tau 0.00, p = 1.0) (data not shown). With regard to the correlation analysis on the basis of the extracted HRs, 10 and 4 trials report both PFS and OS estimates for the targeted agents and the immunotherapy
populations. Although the significant data attrition may certainly limit these results, no significant correlations were found between PFS and OS for the targeted agents (Pearson 0.45, p = 0.19; R2 0.20; Rho 0.78, p = 0.16; Tau 0.34, p = 0.19) and the immunotherapy (Pearson 0.63, p = 0.36; R2 0.66; Rho 0.80, p = 0.20; Tau 0.66, p = 0.30) trials’ sample. 3.3. Benchmarking analysis Cumulative survival rates for each treatment group at every month from treatment start up to 12 months were extracted and pooled according to a weighted-average calculation, for the 1st-line setting; a punctual averaged estimate
54
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
Fig. 2. Regression line according to correlation time-points (3 months PFS and 9 months OS: Panels A and C; 6 months PFS and 12 months OS: Panels B and D) and treatment group (targeted agents: Panels A and B; immunotherapy: Panels C and D).
Fig. 3. Cumulative survival rates (PFS: Panel A; OS: Panel B).
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
55
Table 1 Characteristics of the eligible first-line randomized clinical trials included in the analysis.
Targeted agents
Author
Arms
No. pts (per arm)
PFS: HR (95% CI)
OS: HR (95% CI)
Hudes et al. [2]
Temsirolimus Temsirolimus + IFN-␣ IFN-␣
209 210 207
0.66b (0.53–0.81) p < 0.001
Motzer et al. [13]
Sunitinib IFN-␣ Bevacizumab + IFN-␣ IFN-␣ Bevacizumab + IFN-␣ IFN-␣ Sorafenib Placebo Pazopanib Placebo Axitinib Sorafenib Pazopanib Sunitinib Tivozanib Sorafenib Temsirolimus + bevacizumab IFN + bevacizumab
375 375 327 322 369 363 451 452 290 145 126a 125a 557 553 260 257 400 391
0.53 (0.45–0.64) p < 0.001 0.63 (0.52–0.75) p < 0.001 0.71 (0.61–0.83) p < 0.001 0.44 (0.35–0.55) p < 0.001 0.46 (0.34–0.62) p < 0.001 0.46a (0.31–0.67) p < 0.001 1.05 (0.90–1.22) 0.79 (0.63–0.99) p = 0.042 1.07 (0.89–1.28) p = 0.759
0.73b (0.58–0.92) p = 0.009 0.96c (0.76–1.20) p = 0.70 0.82 (0.67–1.00) p = 0.51 0.91 (0.76–1.10) p = 0.336 0.86 (0.73–1.01) p = 0.069 0.88 (0.74–1.04) p = 0.146 0.91 (0.71–1.16) p = 0.224 0.81a (0.55–1.19) p = 0.144 0.91 (0.76–1.08) p = 0.28 1.24 (0.95–1.62) p = 0.105 1.04 (0.85–1.26) p = 0.638
IFN-␣-2a IFN-␣-2a + 13-CRA r-IFN-␣2c r-IFN-␣2c + r-IFN-␥ IL-2 + IFN HD-IL-2 IFN-␣-2a IFN-␣-2a + 13-CRA IFN-␣ IFN-␣ + IL-2 + 5-FU Autologous CIK IL-2 + IFN␣-2a IL-2 i.v. + IFN-␣ IL-2 s.c. + IFN-␣ IL-2 IFN-␣ IL-2 + IFN-␣ IFN-␥-1b Placebo
161 159 54 48 96 96 145 139 502 504 74 74 80 75 138 147 140 98 99
0.67 (0.53–0.85) p = 0.007 p = 0.98
0.78 (0.61–1.0) p = 0.048 p = 0.73
p = 0.082
0.81 p = 0.211
p = 0.13
p = 0.26
1.02 (0.89–1.16) p = 0.81 0.88 (0.84–0.93) p < 0.001 1.16 (0.83–1.63) p = 0.188 p = 0.01
1.05 (0.90–1.21) p = 0.55 0.58 (0.48–0.69) p < 0.001 1.20 (0.78–1.83) p = 0.202 p = 0.55
p = 0.49
p = 0.52
Escudier et al. [15] Rini et al. [14] Escudier et al. [9] Sternberg et al. [10] Rini et al. [11] Motzer et al. [16] Motzer et al. [17] Rini et al. [18] Immunotherapy
Aass et al. [19] De Mulder et al. [20] Mc Dermott et al. [21] Motzer et al. [22] Gore et al. [23] Liu et al. [24] Negrier et al. [25] Negrier et al. [26]
Gleave et al. [27]
CI, confidential interval; HR, hazard ratio; pts, patients; PFS, progression-free survival; OS, overall survival; 13-CRA, 13-cis retinoic acid; r, recombinant; IFN, interferon; HD, high dose; 5-FU, 5-fluorouracil; CIK, cytokine-induced killer; i.v., intravenous; s.c., subcutaneous. a In patients previously treated with cytokines. b Comparing temsirolimus alone with interferon alone. c Comparing the combination of temsirolimus plus interferon with interferon alone.
for each group was then calculated for both PFS and OS at every month. Outcome probability according to treatment group is shown in Fig. 3A and B for PFS and OS, respectively, with punctual estimates at 3, 6, 9 and 12 months reported in the embedded table. A highly significant correlation between the distributions of the calculated PFS and OS estimates according to treatment groups was found for targeted agents and immunotherapy arms, with Pearson coefficients of 0.98 (95% CI 0.94–0.99, p < 0.0001), 0.95 (95% CI 0.85–0.98, p < 0.0001), respectively (Fig. 4A and B), and to a lesser extent for the
placebo group (Pearson 0.87, 95% CI 0.60–0.96, p = 0.0002) (Fig. 4C). 3.4. Power analysis According to the regression equation derived for targeted agents (y = 33.2949 + 0.6382x), a power-analysis model was developed in order to estimate the patient sample size necessary to determine a 3%, 5%, 10% and 20% PFS advantage at 6 months in the context of 1st-line setting. With a power of 80%, and an alpha error of 0.05, 7859, 2873, 712, and 190
56
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
Fig. 5. Power analysis according to the regression equation for targeted agents (y = 13.2598 + 0.9484x); power 80%, alpha-error 0.05. The determined patients’ sample, corresponds to a 12-month OS difference of 19%, 9%, 4% and 2%, respectively.
Fig. 4. Regression line between PFS and OS cumulative survival rates, according to treatment group (targeted agents: Panel A; immunotherapy: Panel B; placebo: Panel C).
patients would be required; these figures, in turn, would correspond to an absolute increase in 12-month OS rates of 2%, 3%, 6% and 11%, respectively (Fig. 5).
4. Discussion Here we formally analyzed the correlation between PFS and OS in the context of randomized trials of either targeted
agents or immunotherapy conducted in advanced RCC, using a meta-analytic approach. OS is traditionally considered the ‘gold standard’ of efficacy endpoints, above all because it can be objectively established and, therefore, not subject to biases introduced by investigator assessment. However, OS is a difficult endpoint to pursue, because it needs expensive clinical trials with huge numbers of patients and lengthy follow-up; moreover, the use of further lines of therapy and crossover strategies, though necessary for ethical reasons and to guarantee patients’ access to active treatments as they become available, may dilute and obscure eventual survival benefits, particularly in first-line trials. For all these reasons, reliable surrogate endpoints are clearly required [30], not only to identify new active agents, but also to make them rapidly available, particularly at times when the whole community, including health care stakeholders, regulatory agencies, and patients alike, strongly calls for the rapid development of novel treatment opportunities. This necessity is tempered by the need to restrict wide availability only to those drugs that really impact on disease course and from which patients could derive a true, clinically significant, benefit. Three lines of evidence support the use of PFS as a surrogate endpoint for OS in advanced RCC: first, a recent retrospective, non-randomized study performed mainly on patients who received targeted agents as first-line therapy (68.6% of all enrolled patients) demonstrated that PFS at 3 and 6 months significantly correlates with OS (p < 0.0001) and suggested that it may replace OS as a reliable intermediate endpoint for measuring the efficacy of new targeted therapies [12]; second, in a recent systematic review [31], encompassing trials of either cytokines or targeted agents, a significant correlation was found between the log transformation of PFS/TTP and OS (correlation coefficient 0.80; p = 0.0001); third, a recent analysis of 1381 patients treated within two large phase 3 trials, evaluating the efficacy of
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
interferon-alpha with or without bevacizumab in the treatment of advanced RCC, demonstrated a strong correlation between PFS (at 3 and 6 months) and OS (p < 0.00001) with an adjusted HR for death of 2.6 for patients who had disease progression at 3 months [32]. Our current work, significantly adds to such body of knowledge and strongly confirms that PFS may be considered an acceptable intermediate endpoint for ultimate survival benefit, particularly in the context of first-line treatment with targeted agents. PFS may thus represent a good choice, especially when multiple therapy options exist, such as in advanced RCC, as it evaluates only the activity of the study drug and it is not affected by subsequent therapies and crossover strategies. However, the lack of correlation between PFS and OS in first-line has been recently observed in TIVO-1 trial where tivozanib demonstrated to improve PFS but not OS [17]. A possible explanation is that OS in this study was confounded by a differential use of subsequent targeted cancer therapies; in fact, not only a one-way cross-over to tivozanib was allowed for the patients progressing with sorafenib but also a greater proportion of patients in the sorafenib arm had received next-line targeted treatments (63% vs 13% in the tivozanib arm). However, other factors such as reduction of relationship between these two endpoints when an active, targeted therapy comparator is used cannot be excluded, as also observed in INTORSECT and AXIS trials [11,29]. The correlation between PFS and OS is far less stringent in the secondand subsequent-line setting. Indeed, in contrast with firstline results, our exploratory analysis performed on patients previously treated with targeted agents (second-line setting), failed to demonstrate a significant correlation between PFS and OS. Although such discrepancy is currently difficult to explain, our findings suggest that this phenomenon may be setting-, rather than disease-related. The breakdown of relationship between PFS and OS in the second-line treatment has been recently observed in INTORSECT trial comparing temsirolimus with sorafenib in sunitinib-refractory metastatic RCC patients [29]. Results showed that there was no significant difference in PFS between these two agents but paradoxically, a longer OS has been observed in patients treated with sorafenib. Benchmarking and power analysis provide meaningful tools for clinicians who aim at designing future trials with updated and more reliable hypotheses. Such “statistical” evidence could contribute to changing the methodological approach, enhancing the role of PFS as a useful endpoint for efficacy and limiting the use of OS in the context of first-line RCC trials. Benchmarking analysis conducted on cumulated survival rates extracted from the analyzed trials also provides the first unequivocal demonstration of the impact of targeted agents on advanced RCC survival: indeed, the confidence intervals of the landmark analysis of patients treated with targeted agents do not overlap with those of patients receiving either immunotherapy or placebo, particularly from the 9-month timepoint onward. Although this is not a log-rank analysis of Kaplan–Meier curves, it does effectively represent the changing landscape
57
in the natural history of RCC, highlighting how the introduction of targeted agents has impacted on OS expectations, whereas cytokines have failed to improve survival in the long run. In addition, power analysis helps estimate how many patients could be spared if trials were designed on the basis of PFS (Fig. 5). These calculations may be useful to design future trials more effectively, avoiding what has been recently observed with the phase III AGILE trial [33], which, because of an overestimation of differences in treatment effects, was underpowered to demonstrate a potentially relevant difference in efficacy between axitinib and sorafenib in the first-line setting.
5. Conclusions Overall, despite the potential trials selection, the reporting biases and the missing data our data add significantly to current literature, supporting the use of PFS as a reliable endpoint for OS in first-line trials of targeted therapy for advanced RCC and provide useful tools for the design of future clinical trials in this setting, based on more realistic and updated survival expectations. However, the demonstration of a correlation between PFS and OS using aggregate data and a meta-analytic approach is not sufficient to considering PFS as a surrogate endpoint for OS, since it is not able to satisfy all four the Prentice’s criteria [34]. Particularly, the lack of confirmatory individual patient data, does not allow to establish the conditional independence of the impact of the treatment on the true endpoint, given the surrogate endpoint. Moreover, heterogeneity of analyzed clinical trials, including drugs with different mechanisms of action and consequently, with a potentially different effect on median PFS and OS and immature overall survival data from some studies require caution in interpretation of the final results, suggesting the need of further confirming analyses.
Conflicts of interest E. Bria had advisory role for Eli-Lilly, Pfizer, Amgen and speaker’s fee from Eli-Lilly; C. Porta has consultancies for Pfizer, GSK, Roche, Bayer-Schering, Novartis, Astellas, Aveo, Boehringer-Ingelheim, speaker’s fee from Pfizer, GSK, Roche, Bayer-Schering, Novartis, Astellas and research funding from Bayer-Schering, Novartis. Clinical trial support: Pfizer; S. Bracarda is advisory board member for Pfizer, Novartis, Bayer, Boehringer-Ingelheim, GSK, Aveo/Astellas, and received honoraria from: Novartis, GSK, Pfizer; D. Heng has advisory role at Aveo, Pfizer, Novartis, and Bayer; G. Tortora has consultancy role for Novartis, Pfizer and GSK; M. Milella had advisory role for Pfizer, Novartis, Eli-Lilly and received speaker’s fee from Pfizer, GSK, Novartis, Bayer, Astra-Zeneca; F. Massari, F. Maines, S. Pilotto, M. Bonomi, I. Sperduti, D. Giannarelli and F. Cognetti declare no competing interests.
58
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59
Reviewers Dr. Valder Torri, Oncology Department, Mario Negri Institute, Via G. La Masa 19, I-20156 Milan, Italy. Prof. Walter M. Stadler, The University of Chicago, Department of Medicine, Section of Hematology/Oncology, AMB I214 (MC 2115), 5841 South Maryland Avenue, Chicago, IL 60637, United States. Dr. Roberto Iacovelli, PhD Student, Sapienza University of Rome, Department of Radiology, Oncology and Human Pathology, Oncology Unit B, Via Regina Elena 324, I-00161 Rome, Italy. Acknowledgments This work was supported by grants of the Italian Association for Cancer Research (AIRC) (MFAG 14282, Investigator Grant 11930, 5X1000 12182 and 12214) and of Italian Ministry of Education, Universities and Research (MIUR) – Interest National Research Project (PRIN) (2009X23L78 005) without any implication in study design, data collection, data analysis, data interpretation or writing of the paper. References [1] Escudier B, Albiges L, Sonpavde G. Optimal management of metastatic renal cell carcinoma: current status. Drugs 2013;73(5):427–38. [2] Hudes G, Carducci M, Tomczak P, et al. Temsirolimus, interferon alfa, or both for advanced renal-cell carcinoma. N Engl J Med 2007;356(22):2271–81. [3] Zietemann VD, Schuster T, Duell TH. Post-study therapy as a source of confounding in survival analysis of first-line studies in patients with advanced non-small-cell lung cancer. J Thorac Dis 2011;3(2):88–98. [4] Broglio KR, Berry DA. Detecting an overall survival benefit that is derived from progression-free survival. J Natl Cancer Inst 2009;101(23):1642–9. [5] Burzykowski T, Buyse M, Piccart-Gebhart MJ, et al. Evaluation of tumor response, disease control, progression-free survival, and time to progression as potential surrogate end points in metastatic breast cancer. J Clin Oncol 2008;26(12):1987–92. [6] Gill S, Berry S, Biagi J, et al. Progression-free survival as a primary endpoint in clinical trials of metastatic colorectal cancer. Curr Oncol 2011;18(Suppl. 2):S5–10. [7] Halabi S, Vogelzang NJ, Ou SS, Owzar K, Archer L, Small EJ. Progression-free survival as a predictor of overall survival in men with castrate-resistant prostate cancer. J Clin Oncol 2009;27(17):2766–71. [8] Sullivan R, Peppercorn J, Sikora K, et al. Delivering affordable cancer care in high-income countries. Lancet Oncol 2011;12(10):933–80. [9] Escudier B, Eisen T, Stadler WM, et al. Sorafenib in advanced clear-cell renal-cell carcinoma. N Engl J Med 2007;356(2):125–34. [10] Sternberg CN, Davis ID, Mardiak J, et al. Pazopanib in locally advanced or metastatic renal cell carcinoma: results of a randomized phase III trial. J Clin Oncol 2010;28(6):1061–8. [11] Rini BI, Escudier B, Tomczak, et al. Comparative effectiveness of axitinib versus sorafenib in advanced renal cell carcinoma (AXIS): a randomised phase 3 trial. Lancet 2011;378:1931–9. [12] Heng DY, Xie W, Bjarnason GA, et al. Progression-free survival as a predictor of overall survival in metastatic renal cell carcinoma treated with contemporary targeted therapy. Cancer 2011;117(12):2637–42.
[13] Motzer RJ, Hutson TE, Tomczak P, et al. Sunitinib versus interferon alfa in metastatic renal-cell carcinoma. N Engl J Med 2007;356(2): 115–24. [14] Rini BI, Halabi S, Rosenberg JE, et al. Bevacizumab plus interferon alfa compared with interferon alfa monotherapy in patients with metastatic renal cell carcinoma: CALGB 90206. J Clin Oncol 2008;26(33):5422–8. [15] Escudier B, Pluzanska A, Koralewski P, et al. Bevacizumab plus interferon alfa-2a for treatment of metastatic renal cell carcinoma: a randomised, double-blind phase III trial. Lancet 2007;370(9605): 2103–11. [16] Motzer RJ, Hutson TE, Cella D, et al. Pazopanib versus sunitinib in metastatic renal cell carcinoma. N Engl J Med 2013;369:722–31. [17] Motzer RJ, Nosov D, Eisen T, et al. Tivozanib versus sorafenib as initial targeted therapy for patients with metastatic renal cell carcinoma: results from a phase III trial. J Clin Oncol 2013;31(30):3791–9. [18] Rini BI, Bellmunt J, Clancy J, et al. Randomized phase III trial of temsirolimus and bevacizumab versus interferon alfa and bevacizumab in metastatic renal cell carcinoma: INTORACT trial. J Clin Oncol 2014;32(8):752–9. [19] Aass N, De Mulder PHM, Mickisch GHJ, et al. Randomized phase II/III trial of interferon alfa-2a with and without 13-cis-retinoic acid in patients with progressive metastatic renal cell carcinoma: the European Organisation for Research and Treatment of Cancer Genito-Urinary Tract Cancer Group (EORTC 30951). J Clin Oncol 2005;23(18):4172–8. [20] De Mulder PHM, Oosterhof GON, Bouffioux C, et al. EORTC (30885) randomised phase III study with recombinant interferon alpha and recombinant interferon alpha and gamma in patients with advanced renal cell carcinoma. Br J Cancer 1995;71:371–5. [21] Mc Dermott DF, Regan MM, Clark JI, et al. Randomised phase III trial of high-dose interleukin-2 versus subcutaneous interleukin-2 and interferon in patients with metastatic renal cell carcinoma. J Clin Oncol 2005;23(1):133–41. [22] Motzer RJ, Murphy BA, Bacik J, et al. Phase III trial of interferon alfa2a with or without 13-cis-retinoic acid for patients with advanced renal cell carcinoma. J Clin Oncol 2000;18(16):2972–80. [23] Gore ME, Griffin CL, Hancock B, et al. Interferon alfa-2a versus combination therapy with interferon alfa-2a, interleukin-2, and fluorouracil in patients with untreated metastatic renal cell carcinoma (MRC RE04/EORTC GU 30012): an open-label randomised trial. Lancet 2010;375(9715):641–8. [24] Liu L, Zhang W, Qi X, et al. Randomised study of autologous cytokineinduced killer cell immunotherapy in metastatic renal carcinoma. Clin Cancer Res 2012;18(6):1751–9. [25] Negrier S, Perol D, Ravaud A, et al. Randomised study of intravenous versus subcutaneous interleukin-2 and IFN␣ in patients with good prognosis metastatic renal cancer. Clin Cancer Res 2008;14:5907–12. [26] Negrier S, Escudier B, Lasset C, et al. Recombinant human interleukin2, recombinant human interferon alfa-2a, or both in metastatic renal-cell carcinoma. N Engl J Med 1998;338:1272–8. [27] Gleave ME, Elhilali M, Fradet Y, et al. Inteferon gamma-1b compared with placebo in metastatic renal-cell carcinoma. N Engl J Med 1998;18:1265–71. [28] Motzer RJ, Escudier B, Oudard S, et al. Phase 3 trial of everolimus for metastatic renal cell carcinoma: final results and analysis of prognostic factors. Cancer 2010;116(18):4256–65. [29] Hutson TE, Escudier B, Esteban E, et al. Randomized phase III trial of temsirolimus versus sorafenib as second-line therapy after sunitinib in patients with metastatic renal cell carcinoma. J Clin Oncol 2014;32(8):760–77. [30] Sargent DJ, Hayes DF. Assessing the measure of a new drug: is survival the only thing that matters? J Clin Oncol 2008;26(12):1922–3. [31] Delea TE, Khuu A, Heng DY, Haas T, Soulieres D. Association between treatment effects on disease progression end points and overall survival in clinical studies of patients with metastatic renal cell carcinoma. Br J Cancer 2012;107(7):1059–68.
E. Bria et al. / Critical Reviews in Oncology/Hematology 93 (2015) 50–59 [32] Halabi S, Rini B, Escudier B, et al. Progression-free survival as a surrogate endpoint of overall survival in patients with metastatic renal cell carcinoma. Cancer 2014;120(1):52–60. [33] Hutson TE, Lesovoy V, Al-Shukri S, et al. Axitinib versus sorafenib as first-line therapy in patients with metastatic renal-cell carcinoma: a randomised open-label phase 3 trial. Lancet Oncol 2013;14(13):1287–94. [34] Prentice RL. Surrogate endpoints in clinical trials: definition and operational criteria. Stat Med 1989;8(4):431–40.
Biography Emilio Bria graduated in medicine and surgery, and got the Medical Oncology Fellowship with full marks and honors at ‘La Sapienza’ University in Rome (Italy). He trained at the Memorial Sloan Kettering Cancer Center, New York (USA), Department of Medicine, Head and Neck/Genitourinary Service, at the Herbert Hirving Comprehensive Cancer Center, Columbia Presbyterian Medical Center, College of Physicians & Surgeons of Columbia University, New York (USA), and at the Monter Cancer Center, Division of Hematology and Oncology, North Shore-Long Island Jewish Health System, New York (USA). He spent 8 years at the Regina Elena National Cancer Institute, Department of Medical Oncology with a grant in Methodology of Research in Clinical
59
Oncology, with a particular focus upon lung and breast cancer. His main areas of research interest are meta-analyses in clinical oncology, surrogate end-points for clinical trials and definition of treatment guidelines, mathematical models in clinical oncology and new trials’ design for targeted agents in lung and breast cancer. He is currently involved in a series of Phase I–III trials, and in full-time involved in many translational and clinical research projects in lung cancer. He won national and international awards and grants and he is authors of several papers published in peer-reviewed journal and authors of book-chapters. He is also member of the editorial boards of international peer-reviewed journals. He also presented original researches or educational lectures at national and international meetings. He is currently member of the Expert panel of the Antiemetic Study Group, on behalf of the MASCC (Multinational Association for Supportive Care in Cancer) and the ESMO (European Society for Medical Oncology), for the development and updated of the Guidelines for Antiemetic treatment. He is also currently member of the Publication Committee of the International Association for the Study of Lung Cancer (IASLC), and member of the Expert Board of the National Oncological Committee for the national regulatory agency (Italian Drug Agency, AIFA).
