Journal of the National Cancer Institute Advance Access originally published online on July 24, 2007
JNCI Journal of the National Cancer Institute 2007 99(15):1139-1141; doi:10.1093/jnci/djm080
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© The Author 2007. Published by Oxford University Press.
EDITORIALS |
Breast Cancer Trends: A Marriage Between Clinical Trial Evidence and Epidemiology
Affiliations of authors: Department of Biostatistics, The University of Texas M. D. Anderson Cancer Center, Houston, TX (DAB, PMR)
Correspondence to: Donald A. Berry, PhD, Department of Biostatistics, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd, Unit 447, Houston, TX 77030-4095 (e-mail: dberry{at}mdanderson.org).
In this issue of the Journal, Glass et al. (1) report on the changes in screening mammography, menopausal hormone therapy, and breast cancer incidence between 1980 and 2006 in the Kaiser Permanente Northwest (KPNW) health plan. In particular, they report a drop in breast cancer incidence between 2000 and 2004. Similar drops have been reported in other population-based studies (2,3). Considering the various interventions and practices employed by plan participants, the authors conclude that the patterns of screening mammography and menopausal hormone therapy "parallel" breast cancer incidence. They stop short of saying that either screening mammography or decreased menopausal hormone therapy use caused the drop in incidence between 2000 and 2004. But the paralleling implicates the latter.
In observational studies such as those of Glass et al. (1), the question of causation can be inferred but not directly proven. On the other hand, randomized trials are specifically designed to address causation. The relationship between and the importance of both randomization and epidemiology are well illustrated by the studies of menopausal hormone therapy. Investigators of early epidemiologic studies drew a variety of conclusions about the impact of using menopausal hormone therapy on women's health. Fortunately, the limitations of these studies were recognized, and, despite some reservations, randomized studies (4,5) that could definitively address causation went forward and, with respect to some issues, proved the epidemiology studies to be wrong. But these randomized studies had the limitation that the inclusion criteria and highly specified treatment plans left open the question as to whether the results can be broadly applied to the population as a whole (6–8). Such questions can be addressed only by epidemiologic studies. So the two approaches nicely complement each other.
Many randomized trials are based on logical hypotheses about disease processes and not epidemiology. Randomized trials of screening mammography were undertaken because of the intuitive and scientific appeal of early detection. These trials showed that screening causes an increase in breast cancer incidence, presumably because of capturing prevalent cases, early detection of cases (lead-time effects), and detection of cases that may have never become clinically relevant (overdiagnosis) (9,10). The increased incidence of breast cancer in the 1980s in the KPNW population as mammography was introduced provides evidence that the same phenomenon happens in clinical practice.
Whether the increased breast cancer incidence in the 1990s and the decreased incidence between 2000 and 2004 are due to changes in use of hormone therapy is less clear. One reason is that the Women's Health Initiative (WHI) trials of menopausal hormone therapy drew different conclusions for unopposed estrogens and estrogens plus progestins. The former modestly decreased incidence (11) and the latter modestly increased incidence (12). Because information about type and duration of menopausal hormone therapy use are not available for individual patients in the KPNW data, it is not clear whether the WHI trial conclusions apply to the KPNW population. Thus, the question of cause as regards menopausal hormone therapy use is complicated. The safest approach is to say that it may or may not be causal. Glass et al. (1) take this approach, and we have taken the same tack in response to questions regarding our own report (13) based on Surveillance, Epidemiology, and End Results (SEER) trends in breast cancer incidence and menopausal hormone therapy use. However, both studies examined other possible explanations (particularly for the sharp decrease in incidence in 2003) and found that these explanations seem unlikely to explain the effect. Of course, as in all such epidemiologic analyses, it is possible that we are overlooking an important parameter or interaction between parameters.
Addressing cause in a randomized trial means calculating various statistical inferential quantities (including P values and confidence intervals). Such measures are not very helpful in population studies. For one thing, the sample sizes are generally so large that almost every observation is statistically significant! More important, known and unknown biases trump any statistical calculation. When attempting to ascertain causation in a population-based study, the researcher becomes in effect a forensic epidemiologist—identifying the suspects and weighing the evidence regarding each. A problem is that one can never be sure that one has a complete list of suspects. In the case of breast cancer incidence, women who use menopausal hormone therapy may, unbeknownst to anyone, use interventions or change their lifestyle to minimize its side effects, and it may be these interventions that are the true culprits.
Apart from the drop in menopausal hormone therapy use, the most likely explanation for the drop in breast cancer incidence between 2000 and 2004 is decreased use of screening mammography. Indeed, there is evidence that some women who stop taking menopausal hormone therapy are also less likely to have screening mammography (14). However, screening habits of women in the KPNW did not change enough over this period to explain more than a fraction of the drop. Another possibility is tamoxifen/raloxifene, but their use was nearly constant before and during this period. Lifestyle, including diet and exercise, may affect breast cancer incidence. But the effects tend to be minor. More important, lifestyles do not change abruptly. For example, obesity has been implicated as a risk factor for breast cancer, and obesity has increased in the United States in the last 20 years (15). But the increase has been gradual. Any increase in breast cancer due to increased obesity would have to be gradual. The only known factor that would seem to explain the precipitous drop in incidence is the sharp decrease (16,17) in use of menopausal hormone therapy.
Does menopausal hormone therapy cause breast cancer? Probably not. If breast cancer were caused by menopausal hormone therapy and millions of women were to stop taking menopausal hormone therapy, then there would be a drop in breast cancer incidence ... eventually. But breast cancer has a long preclinical period that varies in duration from one tumor to another. So a drop in breast cancer incidence would be gradual and would not be discernible for several years if menopausal hormone therapy was affecting breast cancer incidence. If menopausal hormone therapy does not cause breast cancer, then how can it be involved? A possibility is that use of menopausal hormone therapy makes cancers more visible on mammograms and cessation of hormone therapy makes them less visible. However, if anything, the reverse is true (18). The most plausible explanation is that stopping menopausal hormone therapy removes the fuel that is promoting the growth of some tumors. The growth of these tumors—preferentially ER-positive tumors—may slow or stop entirely, or the tumors may even regress.
Suppose stopping usage of menopausal hormone therapy slows or stops the growth of some tumors. Would a decreased incidence in breast cancer be noticeable in the population? Yes. Consider a woman with a tumor that was growing and was destined to be detected in 2003. She had a negative mammogram in early 2002. When the WHI results were announced in July 2002, she stopped taking menopausal hormone therapy. Deprived of a major source of fuel, her tumor stopped growing or its growth slowed, allowing it to come under the radar in 2003. The overall population impact on breast cancer incidence rates would be immediate and noticeable.
Is it possible to infer from the available evidence whether stopping menopausal hormone therapy slows or stops tumor growth or does both? If the effect of stopping menopausal hormone therapy use were to slow tumor growth, then incidence would eventually level off and begin to increase (at rates that depend on the rate of tumor growth slowing), perhaps returning to previous high levels. The KPNW data suggest a leveling off in incidence from 2004 to 2006, but not yet an increase. Similarly, incidence seems to have leveled off in SEER between 2003 and 2004 (13). This evidence is not sufficient to distinguish between slowing and stopping tumor growth as an explanation. Future incidence rates will help discriminate between the two.
An important question is the effect the decrease in incidence will have on mortality, regardless of whether tumor growth is slowed or stopped by discontinuing menopausal hormone therapy. The safe answer, again, is that we do not know. But the effect will probably be minimal to none. One reason is that estrogen receptor (ER)–positive tumors are sensitive to hormonal therapies such as selective ER modulators (notably tamoxifen) and aromatase inhibitors. The prognosis of patients with ER-positive tumors has been improving over the last 20 or so years, partly because of early detection (including overdiagnosis), partly because of hormonal therapies, and partly because of improvements in chemotherapies. Indeed, the very tumors whose clinical incidence decreases because of stopping menopausal hormone therapy use may be the ones that are successfully treated and therefore least likely to cause breast cancer–related mortality.
The observations of Glass et al. (1) are consistent with those regarding other databases and other populations, some published and some still in press (2,3,19). An anomalous finding in the article of Glass et al. (1) as compared with other databases known to us is the decrease in the incidence of ER-negative tumors (about 50%) over the period 2002–2006 (see their figure 3). They have no good explanation for it, and neither do we. In a related vein, they found a decrease in breast cancer incidence for women younger than 45 years between 2002 and 2006 (see their figure 1). If a decline in menopausal hormone therapy use is the underlying cause for the changes in incidence, then there should be no change for young women or for ER-negative tumors. So these observations argue against menopausal hormone therapy use as an explanation. The authors do not indicate the number of cases, but for women younger than 45 years the number of cases is not sufficient to make the drop statistically significant (in the sense of joinpoint analysis). For ER-negative tumors, the drop was statistically significant, but the number of cases was apparently less than 50 per year. So part of the explanation for this "anomaly" may be statistical fluctuation. We note that between 1985 and 1990 the drop in ER-negative tumors was nearly as great, and this was a period when screening rates were increasing, and so ER-negative cancer rates should have been increasing.
Glass et al. (1) present information from KPNW regarding menopausal hormone therapy use, screening mammograms, and whether and when the women were diagnosed with cancer. Therefore, this study allows an examination and correlation of apparent changes in these parameters. One limitation of the study that might be corrected in the future is that it was not done at the level of individual patient data. Merging this information at the individual patient level would enable a very powerful analysis of the menopausal hormone therapy question. For example, the authors could track each woman in their database and see whether the women who stopped using menopausal hormone therapy were less likely to get screened and whether they were less likely to be diagnosed with cancer. Additionally, they could examine in a more detailed manner the correlation of the type and duration of menopausal hormone therapy use and the changes in breast cancer incidence on stopping such therapy. These additional analyses will be hopefully forthcoming from registries and data resources like those of the KPNW group and might allow new insights into how to use menopausal hormone therapy with the lowest risk of negative effects and to the greatest advantage.
REFERENCES
(1) Glass AG, Lacey JV Jr, Carreon JD, Hoover RN. Breast cancer incidence, 1980–2006: combined roles of menopausal hormone therapy, screening mammography, and estrogen receptor status. J Natl Cancer Inst (2007) 99:1152–61.
(2) Katalinic A, Rawal R. Decline in breast cancer incidence after decrease in utilization of hormone replacement therapy. (April 24, 2007). Breast Cancer Res Treat (2007) (DOI: 10.1007/s10549-007-9566-z).
(3) Jemal A, Ward E, Thun MJ. Recent trends in breast cancer incidence rates by age and tumor characteristics among U.S. women. (May 3, 2007). Breast Cancer Res (2007) 9:R28. (doi: 10.1186/bcr1672).[CrossRef][Medline]
(4) Rossouw JE, Finnegan LP, Harlan WR, Pinn VW, Clifford C, McGowan JA. The evolution of the Women's Health Initiative: perspectives from the NIH. J Am Med Womens Assoc (1995) 50:50–5.[Medline]
(5) Vickers MR, Meade TW, Wilkes HC. Hormone replacement therapy and cardiovascular disease: the case for a randomized controlled trial. Ciba Found Symp (1995) 191:150–64.[Medline]
(6) Black N. Why we need observational studies to evaluate the effectiveness of health care. BMJ (1996) 312:1215–8.
(7) Benson K, Hartz AJ. A comparison of observational and randomized controlled trials. N Engl J Med (2000) 342:1878–86.
(8) Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observational studies and the hierarchy of research designs. N Engl J Med (2000) 342:1887–92.
(9) Jonsson H, Johansson R, Lenner P. Increased incidence of invasive breast cancer after the introduction of service screening with mammography in Sweden. Int J Cancer (2005) 117:842–7.[CrossRef][Web of Science][Medline]
(10) Zackrisson S, Andersson I, Janzon L, Manjer J, Garne JP. Rate of over-diagnosis of breast cancer 15 years after end of Malmo mammographic screening trial: follow-up study. BMJ (2006) 332:689–92.
(11) Anderson GL, Limacher M, Assaf AR, Bassford T, Beresford SA, Black H, et al, Women's Health Initiative Steering Committee. Effects of conjugated equine estrogen in postmenopausal women with hysterectomy: the Women's Health Initiative randomized controlled trial. JAMA (2004) 291:1701–12.
(12) Rossouw JE, Anderson GL, Prentice RL, LaCroix AZ, Kooperberg C, Stefanick ML, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women's Health Initiative randomized controlled trial. JAMA (2002) 288:321–33.
(13) Ravdin PM, Cronin KA, Howlader N, Berg CD, Chlebowski RT, et al. The decrease in breast-cancer incidence in 2003 in the United States. N Engl J Med (2007) 356:1670–4.
(14) Breen N, Cronin K, Meissner HI, Taplin SH, Tangka FK, Tiro JA. Reported drop in mammography: is this cause for concern? Cancer (2007) 109:2405–9.[CrossRef][Medline]
(15) Li CI, Malone KE, Daling JR. Interactions between body mass index and hormone therapy and postmenopausal breast cancer risk (United States). Cancer Causes Control (2006) 17:695–703.[CrossRef][Web of Science][Medline]
(16) Buist DSM, Newton KM, Miglioretti DL, Beverly K, Connelly MT, Andrade S, et al. Hormone therapy prescribing patterns in the United States. Obstet Gynecol (2004) 104:1042–50.[CrossRef][Web of Science][Medline]
(17) Hersh AL, Stefanick ML, Stafford RF. National use of postmenopausal hormone therapy: annual trends and response to recent evidence. JAMA (2004) 291:47–53.
(18) Chlebowski RT, Hendrix SL, Langer RD, Stefanick ML, Gass M, Lane D, Rodabough RJ, Gilligan MA, Cyr MG, Thomson CA, Khandekar J, Petrovitch H, McTiernan A. WHI Investigators. Influence of estrogen plus progestin on breast cancer and mammography in healthy postmenopausal women: the Women's Health Initiative randomized trial. JAMA (2003) 289:3243–53.
(19) Verkooijen HM, Koot VC, Fioretta G, van der Heiden M, Schipper ME, Rapiti E, et al. Hormone replacement therapy, mammography screening and changing age-specific incidence rates of breast cancer: an ecological study comparing two European populations. (March 13, 2007). Breast Cancer Res Treat (2007) (Doi: 10.1007/s10549-007-9554-3).
Related Articles in JNCI
CiteULike 
Connotea 
Del.icio.us What's this?
J Natl Cancer Inst 2007 99: 1152-1161.
J Natl Cancer Inst 2007 99: 1137.
This article has been cited by other articles:
|
|
 |
|
  W. F. Anderson Cancer Surveillance Research Cancer Epidemiol. Biomarkers Prev., June 1, 2009; 18(6): 1669 - 1671. [Full Text] [PDF]
|
|
|
|
|
|
S. Harper, J. Lynch, S. C. Meersman, N. Breen, W. W. Davis, and M. C. Reichman Trends in Area-Socioeconomic and Race-Ethnic Disparities in Breast Cancer Incidence, Stage at Diagnosis, Screening, Mortality, and Survival among Women Ages 50 Years and Over (1987-2005) Cancer Epidemiol. Biomarkers Prev., January 1, 2009; 18(1): 121 - 131. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. R. Garwood, A. S. Kumar, and V. Shim Menopausal Hormone Therapy and Breast Cancer Phenotype: Does Dose Matter? Ann. Surg. Oncol., September 1, 2008; 15(9): 2526 - 2532. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||