|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 26, No 12 (April 20), 2008: pp. 2062-a-2063 © 2008 American Society of Clinical Oncology. DOI: 10.1200/JCO.2008.16.2354
Reporting of Subset Analysis: Where Do We Draw the Line?Department of Oncology, Montefiore Medical Center, Bronx, NY To the Editor:
It is highly disconcerting to see an article on an unplanned subset analysis of the Eastern Cooperative Oncology Group trial 4599 published in the Journal of Clinical Oncology (JCO).1 The effect of the addition of bevacizumab to the standard combination of paclitaxel and carboplatin (PCB) was compared with paclitaxel and carboplatin (PC) alone. The introduction states that platinum-based doublets seem to offer the same benefit in elderly as in younger patients. The authors performed this subset analysis with the rationale that there are little data on newer treatment options and there is need to identify the therapeutic index of these newer agents in the elderly. The original trial enrolled 878 patients, of which 850 patients were included in the primary analysis.2 Of these, 224 patients were There are a number of issues that I find inappropriate in the article. First, the authors have evaluated the effect of the addition of bevacizumab in elderly in two statistically different ways. On the one hand, they have directly compared the two arms in terms of overall survival, progression-free survival, response rate, and survival at 1 and 2 years, and have failed to find a statistically significant difference, suggesting a lack of benefit in efficacy when bevacizumab is added to PC chemotherapy in the elderly. However, if one looks at the number of patients (224 in all), the power to detect a significant difference (hazard ratio [HR] of 0.8) is only 35% to 40%, depending on the different scenarios that one may feed into the power analysis. This is a major drawback of the subgroup analysis—lack of adequate patients and power. On a separate analysis, they have performed an age-by-treatment interaction to test if the magnitude of the benefit of addition of bevacizumab was different for elderly and younger patients, and the result was insignificant (P = .34), suggesting that the benefit of bevacizumab when added to PC is not dependent on age (and therefore is as beneficial in the elderly as in other age groups), in stark contrast to the previous set of analyses. They fail to discuss these two different statistical conclusions. Second, the 2-year survival in the elderly of 14% in the PC arm and 24% in the PCB arm, as shown in Ramalingam et al (Table 5),1 closely approximates that in the original trial for all patients (15% in PC and 23% in PCB).1,2 The authors however, do not perform a statistical analysis for this difference. Using the data provided, the P value for the difference is .0267 (one-sided) and .0535 (two-sided). This clearly suggests a trend favoring the addition of bevacizumab in the elderly at 2 years. It is unclear as to why this P value was omitted in the results.
Third, a multivariate analysis using various factors was performed in the initial publication, including age 65 years as a cutoff.2 The HR for the
Fourth, an editorial5 had been published in JCO criticizing the initial decision by the US Food and Drug Administration to analyze the effect of infusional fluorouracil, leucovorin, and oxaliplatin chemotherapy as compared with fluorouracil/leucovorin in stage II and III colon cancer in a subset analysis, and then grant approval for oxaliplatin only for stage III patients as the treatment benefit was not substantive in stage II (the Multicenter International Study of Oxaliplatin/5-Fluorouracil/Leucovorin in the Adjuvant Treatment of Colon Cancer trial6). It appeared then that the contributing authors and JCO did not favor subset analysis. So the question is why this practice is acceptable now. Although subset analysis can be informative in raising hypotheses, it is difficult to fathom that this could be important enough to be published in a high-impact journal such as JCO. AUTHOR'S DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST The author(s) indicated no potential conflicts of interest.
REFERENCES
1. Ramalingam SS, Dahlberg SE, Langer CJ, et al: Outcomes for elderly, advanced-stage non–small-cell lung cancer patients treated with bevacizumab in combination with carboplatin and paclitaxel: Analysis of Eastern Cooperative Oncology Group trial 4599. J Clin Oncol 26:60-65, 2008 2. Sandler A, Gray R, Perry MC, et al: Paclitaxel-carboplatin alone or with bevacizumab for non–small-cell lung cancer. N Engl J Med 355:2542-2550, 2006 3. Ross H, Bonomi P, Langer C, et al: Effect of gender on outcome in two randomized phase III trials of paclitaxel poliglumex (PPX) in chemonaïve pts with advanced NSCLC and poor performance status (PS2). J Clin Oncol 25:373s, 2006 (suppl; abstr 7039) 4. Albain KS, Belani CP, Bonomi P, et al: PIONEER: A phase III randomized trial of paclitaxel poliglumex versus paclitaxel in chemotherapy-naive women with advanced-stage non-small-cell lung cancer and performance status of 2. Clin Lung Cancer 7:417-419, 2006[Medline] 5. Grothey A, Sargent DJ: FOLFOX for stage II colon cancer? A commentary on the recent FDA approval of oxaliplatin for adjuvant therapy of stage III colon cancer J Clin Oncol 23:3311-3313, 2005 6. André T, Boni C, Mounedji-Boudiaf L, et al: Multicenter International Study of Oxaliplatin/5-Fluorouracil/Leucovorin in the Adjuvant Treatment of Colon Cancer (MOSAIC) Investigators: Oxaliplatin, fluorouracil, and leucovorin as adjuvant treatment for colon cancer. N Engl J Med 350:2343-2351, 2004 Related Reply
Related Article
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2008 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
70 years of age, and the current analysis evaluated the effect of the addition of bevacizumab in these patients. 
