Annals of Oncology Advance Access originally published online on August 19, 2005
Annals of Oncology 2006 17(2):352-353; doi:10.1093/annonc/mdj010
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© 2005 European Society for Medical Oncology
letter to the editor |
Analysis of breast cancer survival by clinical response to neoadjuvant chemoendocrine therapy
In their recent article, Cleator et al. [1
] present analyses of various long-term end points, splitting patients who received neoadjuvant chemoendocrine therapy for breast cancer into those who achieved complete response (CR) or minimal residual disease (MRD) and those who did not (partial response, no change or progressive disease). Such an analysis has the potential to investigate the prognostic effect of CR/MRD for overall survival and other end points, but has a number of pitfalls. One of the requirements, which was not done in this study, is to perform a landmark analysis as applied in Wolmark et al. [2]. In a small study, the absence of statistically significant P values for the association of patient and tumour characteristics with good versus poor clinical response is not a guarantee that an unadjusted comparison is unbiased. Surely this cannot be interpreted as proof that prognostic factors for CR/MRD do not exist. A multivariate analysis where adjustment was made for other covariates should have been presented in this paper.
Wolmark et al. [2] conclude after a similar analysis on 760 patients who received pre-operative chemotherapy that it is premature to suggest that objective tumour regression during the course of neoadjuvant therapy is a definitive surrogate marker for eventual patient outcome, but that their data suggest that it is a distinct possibility. Van der Hage et al. [3], out of the 350 patients in the pre-operative arm of their study, call response a prognostic factor, but it did not remain in the model in a multivariate analysis. It is therefore odd to conclude out of a 144 patient analysis that clinical response can be used as an early surrogate without even doing an adjusted analysis.
If a better identification of patients who will have CR/MRD becomes possible at a later time as suggested, one would still like to verify whether such a factor is predictive for the long-term effect of chemotherapy. It may be exactly the same type of factor that makes the patient's disease more aggressive and may explain long-term effects regardless of therapy.
Additional multivariate analyses as described above can show the prognostic value of clinical response. We suggest that the use of the term ‘surrogate’ for a marker is inappropriate in this setting. It is restricted to situations where there is proof that the marker is not only prognostic, but also captures the full effect of an intervention on the final clinical outcome. It is our opinion that in order to prove that clinical response is an early surrogate of treatment effect, more data and analyses are required [4, 5] and therefore the conclusions of the authors with regards to this aspect are not correct and should be disregarded.
European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium
* (E-mail: jan.bogaerts{at}eortc.be)
References
1. Cleator SJ, Makris A, Ashley SE, Lal R, Powles TJ. Good clinical response of breast cancers to neoadjuvant chemoendocrine therapy is associated with improved overall survival. Ann Oncol 2005, 16: 267–272.
2. Wolmark N, Wang J, Mamounas E et al. Preoperative chemotherapy in patients with operable breast cancer: nine-year results from National Surgical Adjuvant Breast and Bowel Project B-18. J Natl Cancer Inst Monogr 2001; 96–102.[Web of Science][Medline]
3. van der Hage JA, van de Velde CJ, Julien JP et al. Preoperative chemotherapy in primary operable breast cancer: results from the European Organization for Research and Treatment of Cancer trial 10902. J Clin Oncol 2001; 19: 4224–4237.
4. Prentice RL. Surrogate endpoints in clinical trials: definition and operational criteria. Statist Med 1989; 8: 431–440.
5. Molenberghs G, Buyse M, Geys H, Renard D, Burzykowski T, Alonso A. Statistical challenges in the evaluation of surrogate endpoints in randomized trials. Control Clin Trials 2002; 23: 607–625.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||