Journal of the National Cancer Institute Advance Access originally published online on April 29, 2008
JNCI Journal of the National Cancer Institute 2008 100(9):682-683; doi:10.1093/jnci/djn105
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© The Author 2008. Published by Oxford University Press.
CORRESPONDENCE |
Response:Re: Adaptive Therapy for Androgen-Independent Prostate Cancer: A Randomized Selection Trial of Four Regimens
Affiliation of authors: M. D. Anderson Cancer Center, University of Texas, Houston, TX
Correspondence to: Randall E. Millikan, PhD, MD, University of Texas, M. D. Anderson Cancer Center P.O. Box 301439 Houston, TX 77030 (e-mail: rmillika{at}mdanderson.org).
Armstrong et al. correctly point out that we made no comparison to the accepted standard therapy for symptomatic castration-resistant prostate cancer (namely, docetaxel and prednisone every 3 weeks). The purpose of our study was to evaluate the feasibility of a method for efficiently ranking, by order of clinical efficacy, four regimens widely used at the time that our study was designed. Importantly, we did this by means of a clinically grounded algorithm that also serves each individual patient. We believe our experience demonstrates that the method does efficiently rank order the regimens. In particular, the observation that the carboplatin-containing regimen fared better than others was unexpected for us, given carboplatin's lack of efficacy as a single agent. This finding is in line with other recent reports and supports the validity of our approach.
The suggestion that incorporation of patients without metastatic disease or with other markedly favorable prognostic features could account for the observed survival is an important consideration for placing our results in context. Actually, there were five patients (3% of all 150 patients) in our trial with locally advanced disease (including invasion of bladder and invasion into pelvic sidewall) but without clinically detectable metastases. These five patients had survival times ranging from 10.9 to 38.5 months (median = 22.4 months), which is remarkably representative of all patients treated. By contrast, the notion that optimizing treatment strategy may contribute to improved survival is supported by the results of applying the prognostic model of Armstrong et al. (1) to the 145 patients with metastatic disease in our dataset (Figure 1). As shown, there was an apparently favorable survival of patients in our trial compared with the predicted survival that was based on the fitted model of Armstrong et al. (1) if the patients had received docetaxel every 3 weeks. As with all comparisons of results from separate trials, however, this comparison suffers from confounding of between-trial effects. We make no claim that any regimens that we used are superior to docetaxel or, indeed, that outcomes observed on our trial are superior to those in TAX327. Rather, we claim only that we treated patients with rather typical prognostic features and that our approach did not produce an obviously inferior patient outcome, despite the concerns expressed about the toxicity of the regimens we studied. We concur with Armstrong et al. (1) that the toxicity of estramustine adds urgency to the question of its contribution to any combination therapy.
|
2 vs <2), analgesic-requiring pain at baseline (present vs absent), Karnofsky performance status (dichotomized at 70%), Gleason score of primary tumor (>7 vs 
7), progression just before chemotherapy as measurable disease (present vs absent), progression just before chemotherapy on bone scan (present vs absent), baseline prostate-specific antigen (PSA) (log transformed), baseline alkaline phosphatase (log transformed), hemoglobin level (in g/dL), and PSA doubling time (categorized as 1–2, >2–3, >3–6, >6 months; all categories vs <1 month). Our previous report (Cancer treatment for an individual patient usually consists not of a single regimen but rather of sequentially applied regimens. Although this is routine in actual oncology practice, nearly all clinical trial designs, and most analyses, ignore this fact. Accounting for the sequential nature of most actual therapy is important because there may be carryover effects of first-line treatment on outcomes observed after second-line treatment. Our aim was to address these issues in a prospective trial that used an algorithm that reflects how oncologists practice. We previously published a detailed description (2) of the design's properties and operating characteristics, demonstrating that indeed this design provides substantial probabilities of correctly selecting more active regimens. Although we have offered only one approach, we believe that Armstrong, Garrett-Mayer, and Eisenberger would agree that an objective method for selecting the most promising regimens for time-consuming and expensive phase III evaluation would be a substantial improvement over the current, largely subjective approach. Our findings, which require further confirmation, suggest that not only improvement in short-term outcomes that provide a basis for selection but also improvement in patient survival can be achieved by optimizing the treatment strategy.
NOTES
We thank Drs Armstrong and Garrett-Mayer for providing the details of their model to us.
REFERENCES
1. Armstrong AJ, Garrett-Mayer ES, Yang YCO, de Wit R, Tannock AF, Eisenberger M. A contemporary prognostic nomogram for men with hormone-refractory metastatic cancer: a TAX327 study analysis. Clin Cancer Res. (2007) 13(21):6396–6403.
2. Thall PF, Millikan R, Sung H-G. Evaluating multiple treatment courses in clinical trials. Stat Med (2000) 19(8):1011–1028.[CrossRef][Web of Science][Medline]
Related Correspondence
CiteULike 
Connotea 
Del.icio.us What's this?
J Natl Cancer Inst 2008 100: 681-682.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||