|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 25, No 35 (December 10), 2007: pp. 5673-5674 © 2007 American Society of Clinical Oncology. DOI: 10.1200/JCO.2007.14.5268
Are Prostate-Specific Antigen Changes Valid Surrogates for Survival in Hormone-Refractory Prostate Cancer? A Meta-Analysis Is Needed!Statistics Department, European Organisation for Research and Treatment of Cancer Data Center, Brussels, Belgium
International Drug Development Institute, Louvain-la-Neuve, Belgium
Center for Statistics, Hasselt University, Diepenbeek, Belgium To the Editor:
Armstrong et al,1 in the September 1 issue of the Journal of Clinical Oncology, report on their analysis of surrogacy of post-therapy prostate-specific antigen (PSA) changes as replacement for overall survival in patients with hormone-refractory prostate cancer (HRPC) undergoing chemotherapy in the randomized phase III TAX 327 trial. They used the Prentice criteria2 and the proportion of treatment effect (PTE) explained by the surrogate3 to validate measures of PSA decline during the first 3 months of treatment as surrogates for overall survival. Their analysis attempted to confirm the findings of Petrylak at al,4 who suggested in 2006 that such measures fulfilled the Prentice conditions for surrogacy in the randomized phase III Southwest Oncology Group 99-16 trial that compared the combination of docetaxel and estramustine versus the combination of mitoxantrone and prednisone in a similar patient population. Petrylak et al tested several cutoffs for PSA declines during the first 3 months, and reported a PTE equal to 1 (95% CI, 0.73 to 1.0) for a
In TAX 327, Armstrong et al1 found a PTE of 0.66 (95% CI, 0.23 to 1.0) for a These two reports bring more confusion, rather than clarity, to the important issue of whether PSA declines should be used at all as end points in clinical trials in HRPC. Clinicians reading these reports may wonder why similar analyses generated such contrasting conclusions. We wish to draw attention to the fact that the PTE is neither an adequate nor a complete measure of surrogacy. The consistency of the PTE, or lack thereof, across different series of patients is largely due to random variation,5 and does not provide the type of information required to make claims for, or against, surrogacy. Although there is currently no consensus about the conditions that must be fulfilled for an intermediate end point or a biomarker to be an acceptable surrogate for a clinical end point, it has been demonstrated that the PTE is not a valid measure of surrogacy.6,7 The key problem with the PTE is that it amalgamates the correlation between the surrogate end point and the clinical end point (here, PSA changes and survival) and the relative effect of treatment on the surrogate end point and the clinical end point. In the statistical literature, the former has been called an "individual-level" quantity and the latter a "trial-level" quantity.8 Interestingly, measures of explained variation, which were estimated both by Armstrong et al1 and by Petrylak et al,4 provide a valid approach to individual-level surrogacy, but were not interpreted as such by these authors. More seriously, neither article addressed the key issue of trial-level surrogacy (ie, whether the effect of treatment on PSA declines during the first 3 months can reliably predict the effect of treatment on survival). These effects could be estimated within each of these two trials from the comparison of the randomized groups in terms of the surrogate end point and the final end point. Each trial would thus contribute a pair of treatment effects (the effect on PSA declines and the effect on survival), and a complete validation effort would investigate whether these effects are correlated. We have used this approach to investigate the potential surrogacy of PSA declines for overall survival in patients with hormone-sensitive prostate cancer.9,10 It could also be usefully applied to a meta-analysis of individual patient data from these two trials and other randomized trials of cytotoxic agents against HRPC. Such a meta-analytic validation approach would yield a model to predict the magnitude of the effect of specific treatments or classes of treatment on survival, based on the effect of these treatments on PSA declines. The so-called surrogate threshold effect (ie, the minimum effect on the surrogate end point required to predict a significant impact on the true end point) could be estimated from this model and would constitute a statistically valid and clinically useful measure of the validity of the surrogate.11 Although the articles by Armstrong et al1 and Petrylak et al4 show convincingly that PSA declines are useful to predict survival in individual patients, only a meta-analysis will unveil evidence, if any, that treatment effects on PSA can also be used to reliably predict treatment effects on survival. AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST The author(s) indicated no potential conflicts of interest. ACKNOWLEDGMENTS We thank the European Organisation for Research and Treatment of Cancer Charitable Trust for supporting the contribution to this publication (L.C.). REFERENCES
1. Armstrong AJ, Garrett-Mayer E, Ou Yang YC, et al: Prostate-specific antigen and pain surrogacy analysis in metastatic hormone-refractory prostate cancer. J Clin Oncol 25:3965-3970, 2007 2. Prentice RL: Surrogate endpoints in clinical trials: Definitions and operational criteria. Stat Med 8:431-440, 1989[Medline] 3. Freedman LS, Graubard BI, Schatzkin A: Statistical validation of intermediate endpoints in chronic disease. Stat Med 11:167-178, 1992[Medline] 4. Petrylak DP, Ankerst DP, Jiang CS, et al: Evaluation of prostate-specific antigen declines for surrogacy in patients treated on SWOG 99-16. J Natl Cancer Inst 98:516-521, 2006 5. Freedman LS: Confidence intervals and statistical power of the ‘validation’ ratio for surrogate or intermediate endpoints. J Stat Plan Inference 96:143-153, 2001[CrossRef] 6. Buyse M, Molenberghs G: Criteria for the validation of surrogate endpoints in randomized experiments. Biometrics 54:1014-1029, 1998[CrossRef][Medline] 7. Molenberghs G, Buyse M, Geys H, et al: Statistical challenges in the evaluation of surrogate endpoints in randomized trials. Control Clin Trials 23:607-625, 2002[CrossRef][Medline] 8. Buyse M, Molenberghs G, Burzykowski T, et al: The validation of surrogate endpoints in meta-analyses of randomized experiments. Biostatistics 1:49-67, 2000[Abstract] 9. Buyse M, Vangeneugden T, Bijnens L, et al: Validation of biomarkers as surrogates for clinical endpoints, in Bloom JC, Dean RA (eds): Biomarkers in Clinical Drug Development, Marcel Dekker, New York, NY, 2003, pp 149-168 10. Collette L, Burzykowski T, Carroll K, et al: Is prostate-specific antigen a valid surrogate endpoint for survival in hormonally treated patients with metastatic prostate cancer? J Clin Oncol 23:6139-6148, 2005 11. Burzykowski T, Buyse M: Surrogate threshold effect: An alternative measure for meta-analytic surrogate endpoint validation. Pharm Stat 5:173-186, 2006[CrossRef][Medline]
Related Reply
Related Article
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2007 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
30% PSA decline, as well as a PTE equal to 1.0 (95% CI, 0.67 to 1.0) for the 3-month PSA velocity. These authors concluded that these measures of PSA declines can be considered surrogates for overall survival. 
