|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 22, No 23 (December 1), 2004: pp. 4860-4862 © 2004 American Society of Clinical Oncology. DOI: 10.1200/JCO.2004.04.630
In Reply:
Istituto Nazionale Tumori, Milano, Italy Dr Daniel B. Kopans harshly criticizes our article.1 We thank him for giving us the opportunity to rediscuss some points of our report. We will address his questions point by point. First, we wish to point out that Dr Kopans attributes to us a statement ("the article. ...concludes that premenopausal women should not be screened for breast cancer...") that was not part of our conclusions because it does not correspond to our interpretation of the results. Even the point regarding the "hypothesis that [we] wish to prove" is not pertinent to our article. We are well aware that it is virtually impossible to prove that a given model is true, and that models can only be disproved (when they result in significant departure from observed findings). Our article simply suggests a possible biology-based explanation of clinical observations, and no statement about "proven hypotheses" is reported. Dr Kopans denies the occurrence of an early mortality excess for younger women. The question of early mortality excess has recently been addressed by C.J. Baines.2,3 We wish to note here that the denied occurrence of this phenomenon in the Malmo trial that Dr Kopans takes as an example is explicitly reported by its principal investigator I. Andersson4; one needs to read the papers in addition to looking at a few graphs. Moreover, Dr Kopans states that the analysis of Cox "represents an artifact" and that "any adjustment for multiple testing would render the observation nonsignificant"—an occurrence that should be verified, however, and not assumed a priori. We are aware of the complexity of planning, carrying out, and analyzing randomized controlled trials on mammographic screening. The resulting difficulty of achieving unquestionable results is witnessed by the well-known controversy about screening effectiveness. We do not have a fideistic attitude toward the P value from either univariate or multivariate analysis and we believe that early mortality excess of invited women, documented consistently across screening trials, countries, and time, may maintain its clinical and scientific significance even in the absence of statistical significance. Animal studies about the role of primary tumor removal may be performed either on macrometastases or on micrometastases, two very different experimental systems. Dr Kopans, when pointing out the "transient" surgery effect, refers to animal models in which experimental macroscopic "metastases" (mainly subcutaneous implants smaller than the "primary") were studied. Most of these investigations are focused on local surgery-induced changes and do not report data on animal survival. They prove the proliferation increase, which we consider to be one of the possible mechanisms of metastatic development acceleration, while they little say about its role in the natural history of the neoplastic disease. Dr Kopans' criticism completely ignores angiogenesis induction by primary tumor removal—a further important acceleration mechanism5 observed only in animal studies on micrometastases. Studies on micrometastases that are much more similar to the clinical setting prove that noncurative primary tumor removal may produce proliferation enhancement or angiogenesis triggering, thus resulting in tumor dormancy interruption and in prognosis worsening.6,7 In particular, a biphasic effect of surgery on life span was observed in a specific animal model6 in which very early surgery resulted in a few long-term survivors, while more delayed surgery displayed a detrimental effect on life span. Another example comes from studies on the fertility cycle influence on surgery effectiveness for mammary carcinoma in mice: the same surgical maneuver results in different outcomes when performed in different times within the estrous cycle.8 Therefore, Dr Kopans is wrong in underestimating the importance of "transient" phenomena that can throw a switch, with lifelong implications. Can he declare that similar phenomena do not occur in humans and that both favorable and detrimental effects cannot result from primary tumor removal? Metastatic growth enhancement, tumor dormancy, and modulation of host-tumor interactions, which are the main features of the metastasis development model we are proposing, are well known to "oncologists of mice," and we believe that these concepts, mutatis mutandis, may be relevant for humans and should be taken into account even by clinical oncologists without blinders. Another point of Dr Kopans is about what may be noticed in our survival curves. He compares the survival rates of pre- and postmenopausal patients within the categories of tumor size, instead of comparing survival changes when tumor size drops from more than to less than 2 cm in diameter within the same menopausal status. This "look at the graphs" approach does not allow him to perceive that premenopausal patients obtain less improvement by primary tumor undersizing than their postmenopausal counterparts, thus suggesting a possible reason for the diminished effect of screening for the former. For the last point, because of lead-time bias, length bias, and class bias, it is widely accepted that only comparisons between populations randomly allocated to be invited and not to be invited to screening may provide meaningful data. Obviously, if each invited woman had actually been screened and all noninvited women did not undergo mammography, we could compare "screened" versus "not screened" instead of "invited" versus "not invited." The compliance, which in Scandinavian trials was near 80%, is a useful measure of how near to identical the invited were to the screened group. Therefore, Dr Kopans, who is certainly aware of the trap of using surrogate end points, should be reassured about the comparability between arms. We cannot believe that Dr Kopans assumes that, ceteris paribus, throwing the invitation letter into the wastepaper basket is meaningfully different from not receiving the invitation. Jatoi et al suggest another possible and, in our opinion, probable effect of surgical primary tumor removal on the pattern of mortality in screening trials. According to our hypothesis, women in the control arm will show a mortality increase following invitation to undergo delayed screening, resulting in a trend toward mortality curves opening wide. It should be noted, however, that this phenomenon is temporary, as well as that involving invited women in the early years,10 and should not affect the long lasting results of the trial. These considerations add further elements to the complexity of understanding results from screening trials. Regarding the considerations about the timing of surgery in relation to the menstrual cycle, we agree with Jatoi. Since the early report,10 data supporting this point have been produced,11,12 and we hope that the often-solicited randomized prospective clinical trial on this issue will be at last carried out and will definitively answer the question. Authors' Disclosures of Potential Conflicts of Interest The authors indicated no potential conflicts of interest. REFERENCES
1. Demicheli R, Bonadonna G, Hrushesky WJ, et al: Menopausal status dependence of early mortality reduction due to diagnosis of smaller breast cancers (T1 vs T2-T3): Relevance to screening. J Clin Oncol 22:102-107, 2004
2. Baines CJ: Mammography screening: Are women really given informed consent? J Natl Cancer Inst 95:1508-1511, 2003
3. Baines CJ: Mammography screening: Are women really given informed consent? J Natl Cancer Inst 95:1512-1513, 2003 4. Andersson I: Mammographic screening under age 50: A review. Breast 9:125-129, 2000[Medline] 5. O'Reilly MS, Holmgren L, Shing Y, et al: Angiostatin: A novel angiogenesis inhibitor that mediates the suppression of metastases by a Lewis lung carcinoma. Cell 79:315-328, 1994[CrossRef][Medline] 6. Simpson-Herren L, Sanford AH, Holmquist JP: Effects of surgery on the cell kinetics of residual tumour. Cancer Treat Rep 60:1749-1760, 1976[Medline] 7. Folkman J: Angiogenesis in cancer, vascular, rheumatoid and other disease. Nat Med 1:27-31, 1995[CrossRef][Medline] 8. Bove K, Lincoln D, Wood P, et al: Fertility cycle influence on surgical breast cancer cure. Breast Cancer Res Treat 75:65-72, 2002[CrossRef][Medline] 9. Retzky M, Demicheli R, Hrushesky W: Premenopausal status accelerates relapse in node positive breast cancer: Hypothesis links angiogenesis, screening controversy. Breast Cancer Res Treat 65:217-224, 2001[CrossRef][Medline] 10. Hrushesky W, Bluming A, Gruber S, et al: Menstrual influence on surgical cure of breast cancer. Lancet 2:949-952, 1989[CrossRef][Medline] 11. Hagen AA, Hrushesky WJ: Menstrual timing of breast cancer surgery. Am J Surg 175:245-261, 1998[CrossRef][Medline] 12. Jatoi I: Timing of surgery for primary breast cancer with regard to the menstrual phase and prognosis. Breast Cancer Res Treat 52:217-225, 1998[CrossRef][Medline]
Related Correspondence
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2004 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
