|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Originally published as JCO Early Release 10.1200/JCO.2009.23.3601 on September 28 2009 © 2009 American Society of Clinical Oncology.
Selection Bias Is Not a Good Reason for Advising More Than 5 Years of Adjuvant Hormonal Therapy for All Patients With Locally Advanced Prostate Cancer Treated With RadiotherapyStatistics Department, European Organisation for Research and Treatment of Cancer Headquarters, Brussels, Belgium
Department of Urology, University Hospital Bern, Bern, Switzerland To the Editor:
In a secondary analysis of the randomized trial RTOG (Radiation Therapy Oncology Group) 85-31, Souhami et al1 analyzed outcome in patients treated with radiation therapy (65 to 70 Gy) combined with lifelong adjuvant monthly luteinizing-hormone– releasing hormone treatment for locally advanced prostate cancer. The authors concluded that patients who received androgen-deprivation treatment for more than 5 years had improved overall, disease-specific, and progression-free survival compared with patients who underwent shorter hormonal therapy duration (HTD). On the basis of these findings, the authors concluded that "decreasing HTD to The interpretation of these results is severely misleading, because patients who received HT for more than 5 years (ie, group 3) first had to live for 5 years to fulfill these secondary selection criteria. If a patient died fewer than 5 years after radiation therapy, then obviously he had to be included in group 1 (HTD < 1 year) or group 2 (HTD between 1 and 5 years). This is a well-established example of secondary selection bias, and therefore, it cannot be overinterpreted in favor of long-term androgen-deprivation therapy for patients who undergo radiation therapy for locally advanced prostate cancer. The presence of bias can be seen in Figure 1A in Souhami et al.1 The curves indicate no events in group 2 for the first year of the study and no events (ie, no drops in the survival curves) in group 3 until year 5. Similarly, median follow-up for group 3 (11.0 years), as listed in Table 1, is 4 years longer than that reported for the other two groups, again reflecting selection of group 3 on the basis of a longer observed survival time. The fact that the definition of the groups is affected by the survival of the patients invalidates any comparison of the three groups in terms of long-term end points such as disease-free or overall survival. The bias present in Souhami et al1 is known as selection bias in statistical literature and is well documented. Anderson et al2 described it in Journal of Clinical Oncology 25 years ago. Anderson et al also proposed a simple method to correct for this bias, known as the landmark method, which involves defining an observation time (eg, 5 years) that is used to classify patients into groups; only observations past that landmark are used to compare the groups.
To illustrate our statements, we simulated data similar to survival data reported by Souhami et al1 by randomly generating exponentially distributed survival data (parameter
We then applied a landmark of 5 years. Figure 2 shows survival from year 5 for patients in the three groups who survived to that time, with no difference in survival. Readers may argue that this analysis is not informative, because many patients in groups 1 and 2 would have died before year 5, and perhaps the benefit of longer duration of androgen-deprivation therapy would have been observed before year 5. We agree that the significance of this landmark analysis may be limited, but this only reflects the limits inherent in the data collected in a clinical trial1 that did not randomize between varying durations of treatment. We are convinced this cannot be overcome by statistical analyses.
There is ample evidence that patients who have locally advanced prostate cancer may benefit from combined short-term androgen- deprivation therapy if the local dose is less than 76 Gy (as in the RTOG 85-31 trial).3 The benefit of androgen-deprivation treatment together with relatively low-dose radiation therapy is attributed to a synergistic effect.4,5 The additional benefit of long-term HT (24 months) versus short-term HT (6 months) is only marginal and limited to patients with high-risk prostate cancer (Gleason score  8), as suggested by the subgroup analysis of RTOG Protocol 92-02.6 An earlier meta-analysis suggested no significant benefit with 3-year adjunctive androgen-deprivation therapy,7 and the recently published EORTC (European Organisation for Research and Treatment of Cancer) 22961 trial showed that 36 months of androgen-deprivation therapy compared with 6 months of androgen-deprivation therapy improved 5-year overall survival by only 4%.8
In summary, we appreciate that the question of optimal duration of androgen-deprivation therapy in this patient group is extremely relevant, but we warn readers against the conclusion in the analysis of Souhami et al,1 which in our view erroneously attributes the effect of selection bias to a benefit from longer androgen deprivation. The conclusion of the authors is misleading because patients who received HT for more than 5 years had first to survive for at least this period of time. Short-term androgen-deprivation therapy improves local control of disease if the radiation dose is—according to the actual standards—suboptimal (ie, AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST The author(s) indicated no potential conflicts of interest. REFERENCES
1. Souhami L, Bae K, Pilepich M, et al: Impact of the duration of adjuvant hormonal therapy in patients with locally advanced prostate cancer treated with radiotherapy: A secondary analysis of RTOG 85-31. J Clin Oncol 27:2137–2143, 2009. 2. Anderson JR, Cain KC, Gelber RD: Analysis of survival by tumor response. J Clin Oncol 1:710–719, 1983.[Abstract] 3. D'Amico AV, Manola J, Loffredo M, et al: 6-month androgen suppression plus radiation therapy vs radiation therapy alone for patients with clinically localized prostate cancer: A randomized controlled trial. JAMA 292:821–827, 2004. 4. Joon DL, Hasegawa M, Sikes C, et al: Supraadditive apoptic response of R3327-G rat prostate tumors to androgen ablation and radiation. Int J Radiat Oncol Biol Phys 38:1071–1077, 1997.[CrossRef][Medline] 5. Pinthus JH, Bryskin I, Trachtenberg J, et al: Androgen induces adaptation to oxidative stress in prostate cancer: Implications for treatment with radiation therapy. Neoplasia 9:68–80, 2007.[CrossRef][Medline] 6. Horwitz EM, Bae K, Hanks GE, et al: Ten-year follow-up of Radiation Therapy Oncology Group Protocol 92-02: A phase III trial of the duration of elective androgen deprivation in locally advanced prostate cancer. J Clin Oncol 26:2497–2504, 2008. 7. D'Amico AV, Denham JW, Bolla M, et al: Short- vs long-term androgen suppression plus external beam radiation therapy and survival in men of advanced age with node-negative high-risk adenocarcinoma of the prostate. Cancer 109:2004–2010, 2007.[CrossRef][Medline] 8. Bolla M, De Reijke THM, Van Tienhoven G, et al: Six-month concomitant and adjuvant hormonal treatment with external beam irradiation is inferior to 3-years hormonal treatment for locally advanced prostate cancer: Results of the EORTC randomized phase III trial 22961. Eur Urol 7:117; 2008 (suppl) abstr 186. 9. Studer UE, Collette L, Whelan P, et al: Using PSA to guide timing of androgen deprivation in patients with T0-4 N0-2 M0 prostate cancer not suitable for local curative treatment (EORTC 30891). Eur Urol 53:941–949, 2008.[CrossRef][Medline]
Related Article
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2009 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
5 years may have a detrimental effect on patients with locally advanced prostate cancer," because overall survival was significantly better in patients who received androgen-deprivation therapy for more than 5 years, as shown in 
= 0.0861, corresponding to 70% survival at 5 years), which we then manipulated to reproduce the selection bias observed in the article (by adding 1 and 5 years to all survival times in groups 2 and 3, respectively). The simulated survival data indicate a largely significant difference (
