|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 22, No 4 (February 15), 2004: pp. 699-705 © 2004 American Society of Clinical Oncology. DOI: 10.1200/JCO.2004.07.062 Prognosis of Premenopausal Breast Cancer and Childbirth Prior to DiagnosisFrom the Peter MacCallum Cancer Centre, The University of Melbourne, and The Cancer Council Victoria, Victoria, Australia; Prince of Wales Hospital, New South Wales, Australia; and University of Otago, Dunedin, New Zealand Address reprint requests to John L. Hopper, PhD, Centre for Genetic Epidemiology, University of Melbourne, Level 2, 723 Swanston St, Carlton, Victoria 3053, Australia; e-mail: j.hopper{at}unimelb.edu.au
PURPOSE: The time interval between last childbirth and subsequent breast cancer diagnosis is emerging as an important prognostic factor for premenopausal women. PATIENTS AND METHODS: We studied, prospectively, 750 women diagnosed with primary invasive breast cancer before age 45 years who participated in the population-based Australian Breast Cancer Family Study (ABCFS). RESULTS: Median follow-up time was 4.9 years (range, 0.8 to 10.8 years). Compared with nulliparous women, women who gave birth within 2 years prior to diagnosis were more likely to have axillary node-positive (58% v 41%; P = .01), and estrogen receptor-negative (58% v 39%; P = .005) tumors. The unadjusted hazard ratios for death were 2.3 (95% CI, 1.3 to 3.8; P = .002), 1.7 (95% CI, 1.1 to 2.6; P = .03), and 0.9 (95% CI, 0.6 to 1.5; P = .8) for patients who gave birth less than 2 years, 2 to 5 years, and 5 or more years before diagnosis, respectively. After adjusting for tumor characteristics, these hazard ratios were reduced to 1.9 (95%CI, 1.1 to 3.2; P = .02), 1.3 (95% CI, 0.8 to 2.1; P = .3), and 0.9 (95%CI, 0.5 to 1.4; P = .5). Modeling showed that, compared with nulliparous women, the mortality hazard ratio in parous women was 1.9, decreasing by 8% (95%CI, 4% to 13%; P < .001) for each year between last birth and breast cancer diagnosis. CONCLUSION: Proximity of last childbirth to subsequent breast cancer diagnosis is a predictor of mortality independent of histopathological tumor characteristics. Clinicians should be aware that women diagnosed with breast cancer within a few years following childbirth may have a worse outcome than that suggested solely by the standard histopathological prognostic factors of their cancer.
Childbirth is protective against the development of breast cancer in the long-term; it is well established that early age at first live birth and high parity confer a lower cumulative lifetime risk of breast cancer [1]. However, the protective effect on risk may be biphasic, with a transient increase in the risk of breast cancer shortly after pregnancy, followed by a greater long-term reduction in risk [2-6]. This might be explained by the short-term stimulation of any existing malignant clones under the influence of the hormonal milieu of pregnancy, but longer-term inhibition of breast carcinogenesis because of induction of differentiation of normal mammary stem cells in the later stages of pregnancy that otherwise have the potential for neoplastic change [7]. Having a recent childbirth before being diagnosed with breast cancer may also increase a woman's risk of dying from the disease [8-12]. Using prospective data from a population-based cohort of young women diagnosed with nonmetastatic, primary breast cancer, we assessed the impact of recent childbirth before a breast cancer diagnosis, on subsequent mortality.
Subjects Subjects were participants in the Australian Breast Cancer Family Study (ABCFS), a population-based case-control-family study of the genetic, environmental, and lifestyle factors associated with breast cancer [13]. The study commenced in 1992 and recruited women with incident primary breast cancer who were living in Sydney, Australia, or Melbourne, Australia. Recruitment was via the respective state cancer registries (reporting of cancer to these state registries is a legislative requirement). Family members of the patients with breast cancer, as well as population-based controls, were also recruited. Approval for the study was obtained from the ethics committees of the University of Melbourne and the Cancer Councils of Victoria and New South Wales. All subjects provided written informed consent for participation in the study. The overall participation rate for patients with breast cancer in the ABCFS was 69%. Nonparticipation was due to attrition by death (2%), refusal by the attending doctor (8%), refusal by the patient (16%), nonresponse by the attending doctor (1%), nonresponse by the patient (1%), or inability to locate the patient (2%). Details of the recruitment strategy and baseline data collection methods have been previously described [13]. To be eligible for the current study, women had to consent to ongoing follow-up, be diagnosed with nonmetastatic breast cancer before age 45 years, and have no previous history of invasive cancer (apart from nonmelanoma skin cancer). Overall, 1,027 women aged 45 years or younger at the time of their breast cancer diagnosis were enrolled onto the ABCFS. The first 218 patients enrolled via the New South Wales cancer registry were excluded from this study because the consent they gave at the time of entry specifically prohibited any further approaches from the researchers. An additional 37 patients were excluded because of a lack of any follow-up information; eight were excluded because they had metastatic disease at the time of diagnosis and 14 were excluded because they had a history of prior malignancy. Thus, 750 women were included in the current analysis.
Data Collection
Statistical Analysis Time to failure was considered from the date of diagnosis, with subjects left-truncated at the date of interview. For subjects who died, date of death (obtained from medical record, reported to the ABCFS study coordinator by a family member, or determined by linking to the Victorian cancer registry) was used as the time of failure. Women who were not known to have died were censored at the date last known to be alive (ie, date of last contact with ABCFS study staff, or date of last medical follow-up as obtained from the medical record). The proportional hazards assumption was assessed using Schoenfeld residuals.
Equality of proportions was tested using the
The median age at diagnosis of the 750 patients was 37 years, and 548 (73%) women were parous. The patient characteristics by time between last birth and diagnosis are detailed in Table 1. The majority of women were aged between 35 and 39 years at the time of diagnosis. Overall, 90% of patients had a history of oral contraceptive use, but there was no significant difference in this regard between the nulliparous and parous women. Compared with women who had given birth within 2 years before their diagnosis, nulliparous women were less likely to have received adjuvant chemotherapy and were more likely to have received adjuvant radiation.
The tumor characteristics of the patients, by time between last birth and diagnosis, are presented in Table 2. Those who had given birth within 2 years before their breast cancer diagnosis were more likely to have tumors that involved axillary lymph nodes (58% v 41%, respectively; P = .01), and that were estrogen receptor- (ER-) negative (58% v 39%, respectively; P = .005). There was no significant difference in the distribution of grade, size, or PR status of tumors between the two groups.
The median follow-up time was 4.9 years (mean, 5.0; standard deviation [SD], 2.1; range, 0.8 to 10.8 years), during which there were 147 deaths (mean time to death, 4.0 years; SD, 2.0; range, 0.8 to 10.0). There were 32 deaths among the 202 nulliparous women, 25 among the 80 women who gave birth within 2 years prior, 46 among the 168 who gave birth between 2 and 5 years prior, and 44 among the 300 who gave birth more than 5 years before their breast cancer diagnosis. Kaplan-Meier estimates of mortality, by time to diagnosis since last birth, are shown in Figure 1. At 5 years since diagnosis, these estimates were 16% (95% CI, 11% to 23%) for nulliparous women, 28% (95% CI, 18% to 41%) in women who gave birth within 2 years before diagnosis, 24% (95% CI, 17% to 32%) in women who gave birth between 2 and 5 years prior, and 14% (95% CI, 10% to 20%) in women who gave birth more than 5 years before their breast cancer diagnosis.
Table 3 shows the HR estimates for mortality for time between last birth and subsequent breast cancer diagnosis. There was no evidence of departure from the assumption of proportional hazards in any model fit. Compared with nulliparous women, the unadjusted HRs were 2.3 (95% CI, 1.3 to 3.8; P = .002), 1.7 (95% CI, 1.1 to 2.6; P = .03), and 0.9 (95% CI, 0.6 to 1.5; P = .8) for each of the groups described above, respectively. Modeling as a continuous function of time since last birth and diagnosis found that the HR for ever having given birth was 2.4 (95% CI, 1.5 to 3.8; P < .001), and this decreased by 10% (95% CI, 5% to 14%; P < .001) for each additional year between last birth and diagnosis of breast cancer.
Multivariate Cox regression analysis showed that mortality was associated with the number of involved axillary nodes, tumor grade, and PR status (all P < .05), but not age at diagnosis, ER status, tumor size, or adjuvant treatment. Compared with nulliparous women, the adjusted HR for mortality for those who had given birth within 2 years before diagnosis was 1.9 (95% CI, 1.1 to 3.2; P = .02); for those who had given birth between 2 and 5 years before, it was 1.3 (95% CI, 0.8 to 2.1; P = .3); and for those who had given birth more than 5 years before, it was 0.9 (95% CI, 0.5 to 1.4; P = .5). Following adjustment for these identified confounders, parous patients with breast cancer were estimated to be almost twice as likely to die, as compared with nulliparous patients (HR, 1.9; 95% CI, 1.2 to 3.1; P = .007), with the risk decreasing by 8% (95% CI, 4% to 13%; P < .001) for each additional year between diagnosis and last birth. Table 3 shows that whether analysis was by categories of time between last birth and diagnosis, or as a continuous variable, the decrease in the associated HRs with time between last birth and diagnosis, after adjusting for all prognostic factors, was almost identical to that seen after adjusting for number of involved axillary nodes alone. The estimates of risk associated with recent childbirth were not affected by adjustment for ER status of the tumor, breastfeeding after the most recent childbirth, duration of breast feeding and time between cessation of last breast feeding episode and breast cancer diagnosis. Further multivariate analyses that included age at menarche, number of births, age at first birth, age at last birth, and use at any time of oral contraceptives found that none of these influenced the strength of association between time since last birth and breast cancer prognosis.
In this study, the temporal proximity of prior childbirth to subsequent diagnosis of breast cancer was an independent prognostic factor for mortality. Patients with breast cancer who had had a recent birth before their cancer diagnosis had higher mortality than nulliparous patients with breast cancer. The higher mortality in the former group was only partly accounted for by number of involved axillary nodes, minimally by tumor grade and PR status, and was not affected by age at diagnosis, ER status, tumor size, or adjuvant treatment. The risk of death decreased by 8% for every additional year between last birth and breast cancer diagnosis, even after adjusting for these other factors. This study was derived from a population-based cohort in which 69% of patients with early-onset breast cancer cases have been studied. Our findings would have been spurious, or at least exaggerated, if those women who had both a recent childbirth and a poorer outcome were more likely to participate. The opposite may be more plausible. We know of 2% of eligible women who died before they could participate, so there is evidence that at least some of the women with poorer outcome were less likely to participate. It is also possible that having had a recent childbirth may be associated with being less likely to participate due to the competing demands, though contrary arguments could also be made. Therefore, there is nothing obvious to suggest that our results are biased away from the null, and we may even be underestimating the deleterious effect of having a recent childbirth on prognosis. As summarized in Table 4, many other studies have found similar results [8-12]. Only two of the previous studies were large and population-based [10,12], and all used arbitrary groupings for time between most recent childbirth and subsequent breast cancer diagnosis. Using similar groupings, the results of our study are consistent with those of others. Only one study [9] had looked at time since last birth as a continuous variable and it found that the risk of dying decreased by 15% (relative risk, 0.85; 95% CI, 0.74 to 0.97; P = .01) for each additional year between the most recent previous pregnancy and breast cancer diagnosis, compared with 8% in the current study.
We found that women who had given birth within 2 years before their breast cancer diagnosis were significantly more likely to have involved axillary nodes and to have ER-negative tumors. Most previous studies have not investigated whether there were statistically significant differences in tumor characteristics between groups. However, our findings are consistent with those of Daling et al [12], who showed that women who gave birth within 2 years of their breast cancer diagnosis were more likely to have node-positive, PR-negative, high-grade tumors. Thus, women who develop breast cancer in the setting of a recent prior childbirth are more likely to have tumors with a poorer profile of standard histopathological prognostic factors. The worse prognosis of women who developed breast cancer in the setting of a recent prior childbirth was only partially explained by their tumor characteristics, suggesting that other unmeasured tumor or host factors related to recent pregnancy may have influenced outcome. Other tumor characteristics that might mediate the effect include molecular factors such as p53, p27, cyclin D, and cerbB2. These have not yet been examined in our study, but such analysis is underway. Daling et al found that although the tumors of women with a recent birth were more likely to be p53-positive (but no more likely to be erbB2-positive), accounting for p53 status had virtually no effect on their estimate of risk [12]. The most obvious host factor potentially mediating the adverse prognostic effect of a recent childbirth is the hormonal milieu of pregnancy, which might result in stimulation of aggressive pre-existing breast cancer clones. Increased circulating estrogen and progesterone levels in pregnancy might thus be expected to specifically stimulate ER-positive clones. However, in our analysis, the negative prognostic effect of a recent childbirth was not affected by adjusting for the ER status of the subsequent breast cancer. Some studies have suggested a role for prolactin in breast cancer promotion and growth [14], and that hyperprolactinemia is a predictor of poorer outcome [15,16]. Prolactin levels are elevated in women who breastfeed, and we speculated that this might account for the more aggressive tumors seen in women who had had a recent childbirth before their breast cancer diagnosis. However, we found no association between duration of breastfeeding with the last pregnancy, or time between cessation of breastfeeding and diagnosis of breast cancer with mortality. High fasting insulin levels have recently been identified as an independent adverse prognostic factor in breast cancer [17]. Pregnant women develop relative insulin resistance, resulting in high circulating insulin levels [18-20]. Further research is required to determine whether circulating insulin levels mediate the effect of recent childbirth on mortality seen in this study. Melatonin is a hormone involved in circadian rhythms and sleep that may also have oncostatic properties [21]. Exposure to light at night decreases production of melatonin. Observational studies have consistently shown an association between night-shift work and increased breast cancer risk [22,23]. There are, to our knowledge, no data regarding the influence of exposure to light at night on the prognosis of breast cancer; however if melatonin has oncostatic properties, it might be anticipated that decreased melatonin levels might also influence prognosis. The women who had had a recent pregnancy in the current study are likely to have had increased exposure to light at night while tending to the child as compared with those who had not, and may have, therefore, had reduced melatonin levels. This hypothesis requires further testing. It might also be hypothesized that the worse prognosis seen in patients who had had a recent childbirth is due to delay in diagnosis, perhaps associated with breastfeeding. It may be more difficult for a woman to palpate a small mass in a breast that is engorged due to breastfeeding. However, in our study, tumor size did not influence the estimates of risk of mortality associated with childbirth history. Thus, we have no evidence that the worse outcomes seen in women with a recent childbirth are due to delay in diagnosis of the primary tumor. The current study adds to the increasing body of evidence stating that having a recent pregnancy influences the biologic behavior of breast cancer in young women, resulting in a more aggressive phenotype. The tumor or host factors mediating this effect require further research. In the meantime, clinicians should be aware that young women diagnosed with breast cancer within a few years following a recent childbirth may have a worse outcome than might be suggested just by assessing the standard histopathological prognostic factors of their cancer. Whether more intensive adjuvant treatment might improve the prognosis of such patients is currently not known. Randomized clinical trials of adjuvant breast cancer treatments should record childbirth history so that this issue can be assessed.
The authors indicated no potential conflicts of interest.
We thank Mandy Muir and Pauline Jacklin for data collection and Maggie Angelakos for data management.
This study was specifically supported by the National Health and Medical Research Council of Australia (#145604). The Australian Breast Cancer Family Study was supported by the National Health and Medical Research Council of Australia, the New South Wales Cancer Council, the Victorian Health Promotion Foundation, and the United States National Cancer Institute, National Institutes of Health (under RFA #CA-95-003), as part of the Breast Cancer Family Registries, and through cooperative agreements with the Fox Chase Cancer Center, Huntsman Cancer Institute, Columbia University, Northern California Cancer Center, Cancer Care Ontario, and the University of Melbourne. M.R.E.McC. was supported in New Zealand by the Inkster Ross Memorial Fund of the University of Otago. Poster discussion presented at the Annual Meeting of the American Society of Clinical Oncology, May 31–June 3, 2003, Chicago, IL. The content of this manuscript does not necessarily reflect the views or policies of the National Cancer Institute or any of collaborating centers in the Breast Cancer Family Registries, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government or the Cancer Family Registries. Authors' disclosures of potential conflicts of interest are found at the end of this article.
1. Colditz GA. Epidemiology of breast cancer: Findings from the nurses' health study. Cancer 71:1480–1489, 1993 (suppl 4)[Medline] 2. Bruzzi P, Negri E, La Vecchia C, et al: Short term increase in the risk of breast cancer after full term pregnancy. BMJ 297:1096–1098, 1988
3. Lambe M, Hsieh C, Trichopoulos D, et al: Transient increase in the risk of breast cancer after giving birth. N Engl J Med 331:5–9, 1994 4. Albrektsen G, Heuch I, Kvale G: The short-term and long-term effect of a pregnancy on breast cancer risk: A prospective study of 802,457 parous Norwegian women. Br J Cancer 72:480–484, 1995[Medline] 5. Leon DA, Carpenter LM, Broeders MJ, et al: Breast cancer in Swedish women before age 50: Evidence of a dual effect of completed pregnancy. Cancer Causes Control 6:283–291, 1995[CrossRef][Medline] 6. McCredie MR, Dite GS, Giles GG, et al: Breast cancer in Australian women under the age of 40. Cancer Causes Control 9:189–198, 1998[CrossRef][Medline] 7. Russo J, Tay LK, Russo IH: Differentiation of the mammary gland and susceptibility to carcinogenesis. Breast Cancer Res Treat 2:5–73, 1982[CrossRef][Medline] 8. Mohle-Boetani JC, Grosser S, Whittemore AS, et al: Body size, reproductive factors, and breast cancer survival. Prev Med 17:634–642, 1988[CrossRef][Medline] 9. Guinee VF, Olsson H, Moller T, et al: Effect of pregnancy on prognosis for young women with breast cancer. Lancet 343:1587–1589, 1994[CrossRef][Medline]
10. Kroman N, Wohlfahrt J, Andersen KW, et al: Time since childbirth and prognosis in primary breast cancer: Population based study. BMJ 315:851–855, 1997 11. Olson SH, Zauber AG, Tang J, et al: Relation of time since last birth and parity to survival of young women with breast cancer. Epidemiology 9:669–671, 1998[Medline]
12. Daling JR, Malone KE, Doody DR, et al: The relation of reproductive factors to mortality from breast cancer. Cancer Epidemiol Biomarkers Prev 11:235–241, 2002
13. Dite GS, Jenkins MA, Southey MC, et al: Familial risks, early-onset breast cancer, and BRCA1 and BRCA2 germline mutations. J Natl Cancer Inst 95:448–457, 2003 14. Welsch CW, McManus MJ: Stimulation of DNA synthesis by human placental lactogen or insulin in organ cultures of benign human breast tumors. Cancer Res 37:2257–2261, 1977[Medline] 15. Holtkamp W, Nagel GA, Wander HE, et al: Hyperprolactinemia is an indicator of progressive and poor prognosis in advanced breast cancer. Int J Cancer 34:323–328, 1984[Medline] 16. Bhatavdekar JM, Patel DD, Shah NG, et al: Prolactin as a local growth promoter in patients with breast cancer: GCRI experience. Eur J Surg Oncol 26:540–547, 2000[CrossRef][Medline]
17. Goodwin PJ, Ennis M, Pritchard KI, et al: Fasting insulin and outcome in early-stage breast cancer: Results of a prospective cohort study. J Clin Oncol 20:42–51, 2002 18. Catalano PM, Tyzbir ED, Roman NM, et al: Longitudinal changes in insulin release and insulin resistance in nonobese pregnant women. Am J Obstet Gynecol 165:1667–1672, 1991[Medline] 19. Holmes RP, Holly JM, Soothill PW: A prospective study of maternal serum insulin-like growth factor-I in pregnancies with appropriately grown or growth restricted fetuses. Br J Obstet Gynaecol 105:1273–1278, 1998[Medline] 20. Stanley K, Fraser R, Bruce C: Physiological changes in insulin resistance in human pregnancy: Longitudinal study with the hyperinsulinaemic euglycaemic clamp technique. Br J Obstet Gynaecol 105:756–759, 1998[Medline]
21. Hansen J: Light at night, shiftwork, and breast cancer risk. J Natl Cancer Inst 93:1513–1515, 2001
22. Davis S, Mirick DK, Stevens RG: Night shift work, light at night and risk of breast cancer. J Natl Cancer Inst 93:1557–1562, 2001
23. Schernhammer ES, Laden F, Speizer FE, et al: Rotating night shifts and risk of breast cancer in women participating in the Nurses Health Study. J Natl Cancer Inst 93:1563–1568, 2001 Submitted July 10, 2003; accepted December 2, 2003. This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2004 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
5 years) used by Daling et al [
2 test. All statistical analyses were performed using STATA 7.0 (STATA statistical software, release 7.0; Stata Corporation, College Station, TX), and all tests and P values were two-tailed. 
