Correspondence
Molecular Prediction of Recurrence of Breast Cancer
N Engl J Med 2005; 352:1605-1607April 14, 2005
- Article
To the Editor:
Paik et al. (Dec. 30 issue)1 imply that their multigene assay provides more objective and reproducible information than an assigned tumor grade. Any RNA-based gene-expression profile of neoplasms is heavily dependent on tumor cellularity. Moreover, different areas of the tumor may be expected to show different gene-expression patterns. Therefore, selection of the area to be measured becomes a critical issue. Many breast-cancer specimens may be represented in two or more paraffin blocks, and it is unknown how the recurrence score in the current study would have been affected by the choice of block. The authors studied within-patient variability in only two cases. It was not stated how these two cases were chosen and what their mean recurrence scores were. The only number provided is the standard deviation of 2.2 recurrence-score units, and it is not clear whether this represents the average of the two cases. The reported standard deviation may be acceptable if the recurrence score was high but may not be acceptable if the latter was low. A much larger body of data on intratumoral variability in the recurrence score should be made available.
1 ReferencesJoseph Geradts, M.D.
Roswell Park Cancer Institute, Buffalo, NY 14263
joseph.geradts@roswellpark. org 1
Paik S ,Shak S ,Tang G , et al. A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer. N Engl J Med 2004;351:2817-2826
Full Text | Web of Science | Medline
To the Editor:
Paik et al. report that their multigene assay “provided significant information beyond tumor grade.” However, they beg the question of possible bias in how they designed the comparison of their assay results to tumor grade by taking the unusual step of stating the institutional affiliations of the pathologists who determined tumor grade, rather than documenting the pathologists' expertise. Shak, a coauthor of the article, has asserted that the choice of pathologists is irrelevant, since he and his colleagues intended to compare their assay results to tumor grade as it might be determined in the “community,”1 but this seems a disingenuous way to allow bias. Since they advocate submission of tissue to their reference laboratory at Genomic Health,2 the proper test is to compare their assay results to tumor grade determined by pathologists with special expertise in breast disease (i.e., those in pathology reference laboratories). To allow readers to evaluate for themselves whether bias existed in the comparison of the multigene assay results to tumor grade, the authors should document the three pathologists' experience in breast pathology at the inception of the study.
2 ReferencesWilliam H. Goodson, III, M.D.
California Pacific Medical Center Research Institute, San Francisco, CA 941151
Shak S. Development and validation of a multigene RT-PCR assay for predicting recurrence in node-negative, ER +, tamoxifen treated breast cancer patients. Presented at the General Tumor Board, California Pacific Medical Center, San Francisco, September 2, 2004.
2
Genomic Health press releases. (Accessed March 25, 2005, at http://www.genomichealth.com/pressroom/pressreleases.aspx.)
To the Editor:
The article by Paik et al. raises two concerns. First, the practical implications of the test results are unclear. If tamoxifen will be given regardless of the recurrence score, the additional cost of testing would be an enormous burden to health care systems. Second, the 21-gene panel has no prognostic value for patients who do not receive tamoxifen.1 Since Paik et al. show that the assay had prognostic value only in tamoxifen-treated patients, it is unclear whether tamoxifen lowered the risk of recurrence for patients in the low-risk group, or whether there was a bias against high-risk patients in the tamoxifen-treated group.
1 ReferencesChristopher G. Tang, B.A.
Albert Y. Lin, M.D.
Santa Clara Valley Medical Center, San Jose, CA 95128
albert.lin@hhs. co. santa-clara. ca. us To the Editor:
In their editorial, Bast and Hortobagyi1 suggest that the biomarker test described by Paik et al. allows precise prediction of risk and could spare patients with breast cancer the toxicity of chemotherapy. However, in my experience, the test has often been used for the opposite purpose. Many patients with breast cancer are determined to take aggressive treatment even for a minuscule reduction in the risk of recurrence.2-4 Women with a small tumor (<1 cm in diameter), who generally do not receive adjuvant chemotherapy, are now undergoing this test to find a justification for chemotherapy. Others, who are already undergoing chemotherapy, receive a high-risk score and on that basis are being treated more intensively than previously planned.
It is important to address the cost of this method, which is about $3,500 per test. According to Paik et al., 22 percent of women with breast cancer will have an intermediate score, which does not help with treatment decisions. Therefore, it appears that, in reality, this test may put additional pressure on women to get more treatment and may result in an increase in the cost of health care.
4 ReferencesTawee Tanvetyanon, M.D.
Loyola University Chicago Stritch School of Medicine, Maywood, IL 60153
ttanve@lumc.edu 1
Bast RC Jr ,Hortobagyi GN . Individualized care for patients with cancer -- a work in progress. N Engl J Med 2004;351:2865-2867
Full Text | Web of Science | Medline2
Ravdin PM, Siminoff IA, Harvey JA. Survey of breast cancer patients concerning their knowledge and expectations of adjuvant therapy. J Clin Oncol 1998 16:515-21.
3
Palda VA ,Llewellyn-Thomas HA ,Mackenzie RG ,Pritchard KI ,Naylor CD . Breast cancer patients' attitudes about rationing postlumpectomy radiation therapy: applicability of trade-off methods to policy-making. J Clin Oncol 1997;15:3192-3200
Web of Science | Medline4
McQuellon RP ,Muss HB ,Hoffman SL ,Russell G ,Craven B ,Yellen SB . Patient preferences for treatment of metastatic breast cancer: a study of women with early-stage breast cancer. J Clin Oncol 1995;13:858-868
Web of Science | Medline
Author/Editor Response
We agree with Dr. Geradts that selection of tissue for RNA analysis and variability between blocks are important considerations. Of the 668 cases in the published study, macrodissection based on prespecified criteria was performed to exclude normal tissue in 130 cases (19.5 percent). In addition to the reported variability between blocks in two patients, we have additional data in 36 patients. In 16 cases with macrodissection in the National Surgical Adjuvant Breast and Bowel Project (NSABP) trial B-20, the RNA concentration was 8.9 times as high in the enriched tumor region as in the normal region. The median difference between the whole section and the region enriched in tumor tissue in these macrodissected cases was 3.2 recurrence-score units. In 20 additional patients, we estimated the variability in recurrence scores among 60 blocks without macrodissection. Even with a varying degree of relative tumor volume, the standard deviation in the recurrence score between blocks from the same patient was 3.0 recurrence-score units. For 16 of the 20 patients, the standard deviation between blocks was less than 2.5 recurrence-score units. The variability in measurement of the recurrence score is less than the known large variability among readers in the subjective assessment of tumor grade.1 When the recurrence score is used as a continuous function, consideration of variability should have less effect on the interpretation of the result.
We disagree with Dr. Tanvetyanon's opinion that there is no value in an intermediate recurrence score. For patients with an intermediate recurrence score, the test indicates that they are not at the highest or at the lowest risk of distant recurrence. Moreover, the likelihood of recurrence for a tumor with a recurrence score of 19 is different from that for a tumor with a recurrence score of 30, although both are classified as “intermediate” risk. Like any new test, the recurrence score will most likely undergo further optimization over time and become integrated with contributions from clinicopathologic features.
With regard to the comments of Mr. Tang and Dr. Lin, we reiterate that the practical application of the recurrence score is to permit more informed decision making concerning the addition of adjuvant chemotherapy in patients with node-negative, estrogen-receptor–positive breast cancer. An individual tumor can have a low recurrence score and a low risk of distant recurrence because the tumor is less aggressive, responsive to tamoxifen, or both. It is likely that the genes evaluated in the calculation of the recurrence score capture both the prognosis and the responsiveness to tamoxifen.
Finally, we would like to point out that we failed to acknowledge grant support from the National Cancer Institute for the study: 5-U10-CA69651 was awarded to the NSABP Biostatistical Center and grant U10-CA12027-34 to the NSABP Operations Center.
1 ReferencesSoonmyung Paik, M.D.
Norman Wolmark, M.D.
National Surgical Adjuvant Breast and Bowel Project, Pittsburgh, PA 15212
spaik.nejm@nsabp. org Steven Shak, M.D.
Genomic Health, Redwood City, CA 940631
Robbins P ,Pinder S ,de Klerk N , et al. Histological grading of breast carcinomas: a study of interobserver agreement. Hum Pathol 1995;26:873-879
CrossRef | Web of Science | Medline
Author/Editor Response
Tanvetyanon raises several concerns regarding the use of the multigene biomarker test reported by Paik et al. If confirmed, the test may indeed be used to encourage patients who have a poor prognosis while receiving tamoxifen alone to undergo chemotherapy; for those who have a good prognosis, it may help suggest that chemotherapy can be avoided. In our view, both applications are worthwhile, provided that the test is a highly reliable predictor of good prognosis and that there is sufficiently great improvement in the predicted outcome to chemotherapy when all prognostic factors are considered. With regard to cost, the savings from avoiding chemotherapy in a substantial fraction of the patients could outweigh the cost of multigene biomarker tests for all patients with node-negative, receptor-positive breast cancers, but formal analysis is required. This is clearly a work in progress.
Robert C. Bast, Jr., M.D.
Gabriel N. Hortobagyi, M.D.
University of Texas M.D. Anderson Cancer Center, Houston, TX 77030- Citing Articles (1)
Citing Articles
1
Joseph Geradts, Sarah M. Bean, Rex C. Bentley, William T. Barry. (2010) The Oncotype DX Recurrence Score Is Correlated With a Composite Index Including Routinely Reported Pathobiologic Features. Cancer Investigation 28:9, 969-977
CrossRef
