|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 25, No 19 (July 1), 2007: pp. 2862-2863 © 2007 American Society of Clinical Oncology. DOI: 10.1200/JCO.2006.10.4596
In ReplyService de Chirurgie Digestive, Hôpital Tenon; Université Pierre et Marie Curie; Assistance Publique; INSERM U707; Faculté de Médecine Saint-Antoine; Université Pierre et Marie Curie, Paris, France; Division of Biostatistics, School of Public Health, University of California, Berkeley, CA
INSERM U707; Faculté de Médecine Saint-Antoine; Université Pierre et Marie Curie, Paris, France
Service de Chirurgie Digestive, Hôpital Tenon; Université Pierre et Marie Curie; Assistance Publique, Paris, France
Division of Biostatistics, School of Public Health; University of California, Berkeley, CA
INSERM U602; Hôpital Paul Brousse; Université Paris XI, Villejuif, France In their correspondence, Cascinu et al raise two points of concern about our recently published article.1 The first one is the lack of information about clinical and pathologic features known to be of prognostic significance. None of our patients received emergency surgery, meaning that no patient had occlusion or perforation as clinical presentation. Additional information about the number of examined lymph nodes should have been given in our report. In fact, we have only included real stage II patients, meaning that a minimum of 15 lymph nodes were pathologically assessed for all 50 patients. As outlined by Cascinu et al, in contrast to the tumor differentiation, the parietal invasion depth and the vascular invasion status were not part of our analysis. Following their correspondence, both these pathologic features, as well as the perineural invasion status, were studied and did not reveal any difference between the two groups of patients (recurrence- and disease-free groups). The second point of concern raised by Cascinu et al is the variability of prognosis signatures in terms of gene composition, and its potential consequence on the use of gene expression profiling in clinical practice. In our article, the Monte-Carlo cross-validation clearly showed that there was no unique prognosis predictor. Indeed, prognosis predictors included a small core of stable genes and a large number of variable genes. Even if perturbing, this instability of gene selection is the current price to pay for the tremendous amount of data provided by microarray technology and the comparatively small number of patients. A recent study suggested that building a stable prognosis predictor could require several thousands of patients.2 The absence of gene overlap between the 23-gene prognosis predictor proposed by Wang et al,3 based on 38 patients, and our 30-gene prognosis predictor,1 based on 50 patients, is therefore quite expected. A pragmatic view to prognosis prediction by gene expression profiling is to determine if it can be of use in clinical practice, rather than focusing on the expected instability of gene selection from study to study. Therefore, we must emphasize that the rule developed by Wang et al3 was predictive of survival in our data, although it did not fare better than our proposed 30-gene rule. We could not validate our rule on the data of Wang et al because, to our knowledge, it is not publicly available. As of today, gene expression profiling has demonstrated a potential to predict the prognosis of nonmetastatic colon cancer. This is supported by the article of Wang et al,3 our publications,1,4 and those from other teams.5,6 However, gene expression profiling is not ready yet for a clinical use. As outlined by Cascinu et al, all available studies are retrospective and, before being used in clinical practice, gene expression profiling needs to be assessed in a prospective manner. Above all, studies including larger numbers of patients are needed, and this will mainly rely on the willingness of the different research teams to work together because including several hundreds patients is hardly possible for a single team. There are two ways to proceed. The first is that several groups agree on a common project (direct collaboration). Another way is allowing mutual access to data collected by other teams; this naturally implies that each research group posts all their data on the Internet (indirect collaboration). As an example, data of our two most recent studies1,4 are downloadable, 7 allowing other researchers to compare and/or pool them with their own data, to test our signature on their patients, or, reciprocically, test their signatures on our patients. Another aspect of collaboration is the absolute need to validate existing prognosis signatures on totally different sets of patients, as we did with the predictor proposed by Wang et al.3 Additional studies on stage II colon cancer, even including more patients, would be of little practical value if they were only designed to propose another prognosis predictor. Such a predictor would probably be totally different, in terms of gene composition, from the predictor proposed by Wang et al3 and ours,1 which may reinforce the impression of nonreproducibility, and thus inapplicability, of gene expression profiling. In contrast, new studies would greatly gain in interest if they were also designed to assess performance of previously proposed predictors. Similarly, new studies on stage III colon cancer should take into account the genes identified by Arango et al.6 Pragmatically, if a given signature was successfully validated on a totally independent set of patients, it could then be assessed in a prospective way, first step to a possible use in the clinical setting. AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST The author(s) indicated no potential conflicts of interest. ACKNOWLEDGMENTS Supported by the J. David Gladstone Institutes and General Clinical Research Center at San Francisco General Hospital, and by a grant from the California Institute for Quantitative Biomedical Research, University of California Berkeley (A.B.). REFERENCES
1. Barrier A, Boelle PY, Roser F, et al: Stage II colon cancer prognosis prediction by tumor gene expression profiling. J Clin Oncol 24:4685-4691, 2006 2. Ein-Dor L, Zuk O, Domany E: Thousands of samples are needed to generate a robust gene list for predicting outcome in cancer. Proc Natl Acad Sci U S A 103:5923-5928, 2006 3. Wang Y, Jatkoe T, Zhang Y, et al: Gene expression profiling and molecular markers to predict recurrence of Dukes’ B colon cancer. J Clin Oncol 22:1564-1571, 2004 4. Barrier A, Roser F, Boelle PY, et al: Prognosis of stage II colon cancer by non-neoplastic mucosa gene expression profiling. Oncogene 26:2642-2648, 2007[CrossRef][Medline] 5. Eschrich S, Yang I, Bloom G, et al: Molecular staging for survival prediction of colorectal patients. J Clin Oncol 23:3526-3535, 2005 6. Arango D, Laiho P, Kokko A, et al: Gene-expression profiling predicts recurrence in Dukes'C colorectal cancer. Gastroenterology 129:874-884, 2005[CrossRef][Medline] 7. Alain Barrier's Gene Chips Repository: http://www.u707.jussieu.fr/boelle/genechips/index.html
Related Correspondence
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2007 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
