The Surveillance, Epidemiology, and End Results program of the National Cancer Institute provides information on annual numbers of cancer cases registered in cancer registries covering about 26% of the US population. These data were used by Jemal et al.1 to show that between 1970 and 1999 age-standardized annual breast cancer incidence rates rose steadily from around 130 to around 135 per 100,000, but that between 2000 and 2003 they declined sharply to around 125 per 100,000. Irregularities from year to year are compatible with random variation and do not disturb either of the 2 longer-term trends. Several other authors using the same data have focused on the decline since 2000,2–5 and similar trends have been reported in the smaller areas of Northern California6 and Germany.7
Some authors have observed that the decreased use of exogenous hormones that followed reports of their association with increased risk of breast cancer may have been responsible for the decline in breast cancer incidence.6–9 This possibility cannot be ruled out, although the time relationship seems rather too short. Tarone10 focused on incidence trends in younger US women and a cohort analysis of mortality rates from breast cancer, and suggested that there were “beneficial changes in underlying risk or protective factors for breast cancer beginning at a relatively early age for US women born after 1945.” Comparing temporal changes in incidence rates and mortality is difficult for a disease in which improved therapy has undoubtedly prolonged survival. More persuasively, Weiss11 noted that “it is indeed likely that a substantial proportion of women with in situ breast cancer (identified by screening) would develop invasive disease within a few years if not treated.”
The increase in breast cancer incidence rates before 2000 probably does reflect the increasing use of more effective methods of breast cancer screening and earlier diagnosis. It is clear that cases detected by screening during a fixed time period will increase the number of cases diagnosed and hence the incidence rate of the diagnosed disease over time. Less obvious is the fact that screening decreases the number of preclinical cases to be found by screening in subsequent years. Only Weiss seems to have considered this in regard to the decline since 2000.11 Diagnosed incidence will increase as long as the effect of efficacy or availability of screening continues to improve, and exceeds the decline in the pool of prevalent unscreened cases. When the reverse becomes the case, the incidence will decrease. Ultimately, when a steady state is reached incidence will stabilize at the level present before the introduction of screening. This situation has not yet come about. Furthermore, it is possible that rates could stabilize at an even lower level than that which prevailed before the introduction of screening. This could occur if some of the lesions identified at screening would not have progressed to the stage of clinical diagnosis during the life of the individual. However, since cases will be diagnosed on average at an earlier stage of disease than if they had come to attention clinically, the objective of the screening program still is served.
The efficacy of programs for screening (and consequent treatment) will come ultimately from observation of trends in mortality from a disease, not its incidence. In this respect, the picture is moderately encouraging. After remarkable stability at slightly above 30 deaths per 100,000 per year between 1930 and 1990, age-standardized death rates from breast cancer in the United States declined steadily after 1990 to around 25 in 2002–2003—some 10 years before the decline in incidence rates became evident.1,8 The extent to which such a decline can be attributed to early diagnosis or to improved therapy in general (or both) is not discernible from the study of time trends, unless one or the other shows much more secular variation than has been evident so far.
ABOUT THE AUTHORS
BRIAN MACMAHON was Professor emeritus and past chairman of the Department of Epidemiology at the Harvard School of Public Health, Boston. He died on 5 December 2007 at the age of 84. (See page 357.) PHILIP COLE is Professor emeritus and past chairman of the Department of Epidemiology, School of Public Health, University of Alabama at Birmingham.
1. Jemal A, Siegel R, Ward E, et al. Cancer statististics. 2007. CA Cancer J Clin
2. Berry DA, Cronin KA, Plevritis SK, et al. Effect of screening and adjuvant therapy on mortality from breast cancer. N Engl J Med
3. Stewart SL, Sabatio SA, Foster SL, et al. Decline in breast cancer incidence—United States, 1999–2003. Morb Mortal Wkly Rep
4. Ravdin PM, Cronin KA, Howlader N, et al. The decrease in breast-cancer incidence in 2003 in the United States. N Engl J Med
5. Tarone RE. Breast cancer trends among young women in the United States. (Brief report). Epidemiology
6. Clarke CA, Glaser SL, Uratsu CS, et al. Recent decline in hormone therapy utilization and breast cancer incidence: clinical and population-based evidence. J Clin Oncol
7. Katalinic A, Rawal R. Decline in breast cancer incidence after decrease in utilization of hormone replacement therapy. Breast Cancer Res Treat
. 2007; E-pub ahead of print.
8. McNeil J. Breast cancer decline mirrors fall in hormone use, spurs both debate and research. J Natl Cancer Inst
9. Jatoi I, Chen BE, Anderson WF, et al. Breast cancer mortality trends in the United States according to estrogen status and age at diagnosis. J Clin Oncol
10. Tarone RE. Breast cancer trends. (Letter. The author replies). Epidemiology
11. Weiss NS. Breast cancer trends (Letter); Epidemiology