|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 23, No 27 (September 20), 2005: pp. 6804-6805 © 2005 American Society of Clinical Oncology. DOI: 10.1200/JCO.2005.02.5163
In Reply:The University of Texas, M.D. Anderson Cancer Center, Houston, TX We appreciate the interest expressed by Puglisi et al in our article. They raised several issues, which deserve clarification and comment. The concern about the accuracy of sentinel lymph node biopsy in patients with a clinically negative axilla is a concern of general. However, as described in our manuscript, only patients with a negative axilla as determined both by physical examination and sonography were offered this procedure. Sentinel nodes were identified using a combination of blue dye and radioisotope injection, with the use of an intraoperative probe (identifying nodes of interest), which is associated with a higher positive predictive value and a lower false-negative rate.1 A total of seven patients in the chemotherapy-alone arm and nine patients in the chemotherapy-with-trastuzumab arm had a sentinel node biopsy procedure. The median numbers of sentinel lymph nodes sampled were three (range, one to five), and four (range, one to five) in the chemotherapy-alone and chemotherapy-with-trastuzumab arms, respectively. Only one patient in each arm had a single sentinel lymph node sampled; all others had multiple sentinel nodes removed and histologically examined. The false-negative rate of 25% referred to in the review by Kuerer and Newmann2 is from two small series, one with 15 patients and the second with 40 patients. The larger series has false-negative rates ranging from 0% to 12%, and a recent meta-analysis of all published series on sentinel node biopsy after preoperative chemotherapy demonstrated a sensitivity of 90%, positive predictive value of 100%, and overall accuracy of 95.2%.3,4 This meta-analysis demonstrated that the sensitivity of sentinel node biopsy after chemotherapy was similar to that of patients undergoing sentinel node biopsy who have not received chemotherapy. Puglisi's second concern was the early termination of this randomized trial.5 The decision to stop a clinical trial is seldom easy. Considerations include (1) the safety of patients in the trial; (2) preserving the scientific integrity of the trial, including following the protocol to the extent possible; (3) the probabilities that the hypotheses under consideration will eventually be shown to hold true, given the interim results; (4) the impact of the trial results on science and clinical practice if the trial is stopped early and if it is not stopped early; and (5) the accrual rate and the estimated time the final trial results will become available. In the neoadjuvant trastuzumab trial in question, the Data Monitoring Committee (DMC) did not have concerns about the safety, but carefully reviewed all the other issues we listed in this letter. At the time that the trial was stopped by the DMC there were 34 patients with response information: four pathologic complete responses (pCRs; 25%) of 16 patients on the control arm and 12 pCRs (67%) of 18 patients on the experimental arm. The protocol called for accruing 164 patients, randomly assigned equally to the two arms, with no interim analyses. Therefore, we agree with the correspondents that the frequentist P value of .016 has no inferential meaning. However, Bayesian probabilities do have meaning and can be calculated at any time. As we indicated in our article, the DMC was impressed by the Bayesian predictive probability of 95% that the results would still be statistically significant if 164 patients were accrued. There were several mitigating factors for the DMC. One is that this predictive probability was conservative in the sense that it was based on assuming that the two arms were exchangeable a priori. Empirical evidence from metastatic trastuzumab trials and the biologic rationale of this targeted therapy justify assigning a substantial prior probability with trastuzumab being effective in the neoadjuvant setting. Incorporating such information would give a probability of eventual statistical significance even greater than 95%. Put another way, the data from this trial would be far less compelling if the two arms were arbitrarily selected therapies used for an arbitrary disease population. Another mitigating factor was the trial's accrual rate, which had slowed to fewer than two patients per month. This meant that it would take many years to accrue the originally targeted 164 patients. The question addressed in the trial would probably be clinically irrelevant at that time. The DMC felt that preserving the letter of the protocol was untenable and that the patients who had volunteered to contribute to science by participating in this trial would be ill served by keeping a lid on the results. The DMC understood that the trial was small and that some clinicians and policymakers might discount its conclusions for that reason. However, the study produced legitimate comparative information, and readers would clearly see that it was small. Having the trial's results in the public domain would facilitate other investigators in designing and running future trials. An additional benefit was that the availability of these results might make it easier for other DMCs to stop other related trials. The correspondents indicated that one should not assume improvements in pCR rates translate into overall survival benefit. We agree, and the DMC understood this as well. However, the trial was never designed to address overall survival benefit. Even with 164 patients, the possibility of showing an overall survival benefit was remote. Therefore, continuing the trial with the hope of showing an overall survival benefit was not considered. The DMC understood that no trial stands alone. As we have already indicated, our neoadjuvant results were consistent with those of the trastuzumab trials evaluating breast cancer patients with metastatic disease. Based on the results of the two large cooperative group adjuvant trials that were recently stopped early with a positive conclusion regarding the benefits of trastuzumab, we now know that our results are consistent with those in the adjuvant setting as well.6 Of course, our trial pales in sample size comparisons to both of these other settings. But the consistency of the various trials means that despite its size, our trial demonstrates that the benefits of trastuzumab extend to a therapeutic strategy that would otherwise be unexplored. Moreover, given the survival benefit of trastuzumab shown in the adjuvant studies, it would now be impossible to conduct a study such as ours, and there would be a permanent and nagging uncertainty regarding the benefits of trastuzumab on pCR rate in the neoadjuvant setting. Although we agree with the correspondents that "it is reasonable to disagree with the decision to stop this trial," we continue to feel that the DMC made the proper recommendation. Authors' Disclosures of Potential Conflicts of Interest Although all authors completed the disclosure declaration, the following authors or their immediate family members indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.
Dollar Amount Codes (A) < $10,000 (B) $10,000-99,999 (C)
REFERENCES
1. McMasters KM, Tuttle TM, Carlson DJ, et al: Sentinel lymph node biopsy for breast cancer: A suitable alternative to routine axillary dissection in multi-institutional practice when optimal technique is used. J Clin Oncol 18:2560-2566, 2000
2. Kuerer HM, Newmann LA: Lymphatic mapping and sentinel lymph node biopsy for breast cancer: Developments and resolving controversies. J Clin Oncol 23:1698-1705, 2005 3. Xing Y, Ding M, Cox D, et al: Meta-analysis of sentinel lymph node biopsy following preoperative chemotherapy in patients with operable breast cancer. Proc Am Soc Clin Oncol 22:17s, 2004 (suppl; abstr 561)
4. Mamounas EP, Brown A, Anderson S, et al: Sentinel node biopsy after neoadjuvant chemotherapy in breast cancer: Results from National Surgical Adjuvant Breast And Bowel Project Protocol B-27. J Clin Oncol 23:2694-2702, 2005
5. Buzdar A, Ibrahim NK, Francis D, et al: Significantly higher pathologic complete remission rate after neoadjuvant therapy with trastuzumab, paclitaxel, and epirubicin chemotherapy: Results of a randomized trial in human epidermal growth factor receptor 2–positive operable breast cancer. J Clin Oncol 23:3676-3685, 2005 6. National Cancer Institute: Herceptin combined with chemotherapy improves disease-free survival for patients with early-stage breast cancer. http://www.cancer.gov/newscenter/pressreleases/HerceptinCombination 2005. (Accessed August 11, 2005)
Related Correspondence
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2005 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
$100,000 (N/R) Not Required
