Answers are grounded in this paper and cite their sources.
Given the direct and indirect costs of mammographic breast cancer screening in North America (1,2), any evidence leading to the improved efficacy of screening mammography should be welcome. The importance of high technical standards to achieve mammographic accuracy has been recognized (3). Improved accuracy should lead to improved efficacy, namely, further reductions in mortality from breast cancer. The efficacy of breast screening is most contested in women who are 40-49 years old (4), many of whom are premenopausal. In this group, a 16% reduction in breast cancer mortality has been observed 7-10 years after the initiation of screening (5), about half of that observed in women who are 50-69 years old. Although mammographic sensitivity is indisputably lower (less accurate) in younger women, one expert has asserted (6) that “Nothing magical happens at age 50. There is no significant difference in breast tissue as seen by mammography” when women in their 40s are compared with women in their 50s. Tabar et al. (7) attributed the smaller benefit of screening women in their 40s to shorter sojourn times in some women and to the long interval between screening examinations that occurred in the Two-County Trial. This commentary will give an overview of available information about biologic and life style factors that may reduce the accuracy of screening mammography and thereby also reduce screening efficacy. Among the factors diminishing the efficacy of screening (Table 1), false-negative mammograms are of crucial importance. Delay in the detection of breast cancer must reduce the opportunity to reduce mortality. This conviction led to a study of menstruating women that found that the chances of a woman having a false-negative mammographic examination were almost doubled in the luteal phase of the menstrual cycle compared to what the chances were in the follicular phase, although only in women who had ever used exogenous hormones (presumably oral contraceptives) (8). Possible biologic mechanisms that could explain this observation are increased mammographic density associated with breast tissue changes in the luteal phase and luteal mastalgia (pain) restricting mammographic compression. More recently, White et al. (9) reported a small but statistically significant increase in mammographic density in the luteal phase. This observation is consistent with the increase in false-negative mammograms previously reported (8), with the conclusion that early breast cancer detection is more difficult to achieve in younger women (10) and with the generally accepted notion that increased density (variously described as mammographic, parenchymal, and fibroglandular) is associated with decreased mammographic sensitivity (Table 2). To identify other factors that may reduce screening efficacy by reducing mammographic accuracy in both premenopausal and postmenopausal women, it may be useful to consider the following issues: What changes occur in the breast during the menstrual cycle? What characterizes the postmenopausal breast? How do reproductive history, age, body mass, and hormone replacement therapy (HRT) impact on mammographic density? What effect does menstrual cycle phase have on mammographic interpretation? How does density affect mammographic interpretation? What are the implications of this tangled web of physiologic and exogenous phenomena? The breast changes that occur during the menstrual cycle include subjective symptoms (heat, swelling, and pain); quantifiable changes in volume and temperature; histologic changes including involution, proliferation, and apoptosis; changes in water content; and changes in magnetic resonance imaging characteristics. Subjectively, women report increased nodularity and a sense of fullness (11) and changes in breast temperature and tenderness (12) during the menstrual cycle. In women who reported breast swelling and cyclic mastalgia, Deschamps et al. (13) observed a threefold increase in risk of having mammary dysplasia that involved 50% or more of breast parenchyma. Objectively, volume changes and temperature changes have been detected. With the use of water displacement to measure changes in volume, Milligan et al. (14) found that breast volume was maximum in nulliparous women on day 25 of the cycle. The mean total change in volume was 100 mL for natural cycles and 66 mL for women using oral contraceptives. Simpson et al. (15) reported a cyclical increase in the surface temperature of the breast that peaked during the luteal phase. Microscopic changes include involution during the follicular phase and proliferation during the luteal phase with mitosis, lobule size, and the number of acini, all reaching a maximum during the latter half of the luteal phase (16). Potten et al. (17) observed that, in menstruating women, breast epithelial proliferation peaked at day 21 and was negatively associated with age. Apoptosis is maximal in the late luteal phase (18). Malberger et al. (11), using preovulatory and postovulatory fine-needle aspirates of breast tissue, reported that nuclear features of the cells collected in the luteal phase (larger, less compact, and with a more prominent nucleolus) enabled investigators to identify all of the aspirates as preovulatory or postovulatory, when aspirates from the same woman were compared. More recently, Olsson et al. (19), examining normal breast epithelium from 58 women undergoing reduction mammoplasties, found that this tissue had a statistically significantly higher proliferative rate in the luteal phase than in the follicular phase. Soderqvist et al. (20) used the proliferation marker Ki-67/MIB-1 monoclonal antibody in biopsy specimens from 25 women who underwent fine-needle biopsy twice in the same menstrual cycle. The median percentage of MIB-1-positive cells doubled in the luteal phase compared with the value observed in the follicular phase. Gompel et al. (21) observed that expression of tyrosine transmembrane receptors, such as epidermal growth factor and c-erbB-2, which mediate proliferation, is higher in the luteal phase than in the follicular phase. Dabrosin et al. (22) has shown that ornithine, an amino acid required for cellular growth and differentiation, decreases late in the menstrual cycle, coinciding with the luteal peak in proliferation and apoptosis. Even though it is well established that proliferative activity is maximal in the luteal phase, it cannot sufficiently explain the 20% increase in breast volume. Glandular tissue accounts for only 10% of the volume of a breast. Other changes that occur are water retention and increased vascularity (18,23). Anderson (18) suggested that hormonal fluctuations may increase acid mucopolysaccharides in the breast ground substance, thereby increasing the water content of breast tissue. In 1990 Fowler et al. (24) used magnetic resonance imaging to evaluate menstrual cycle-related changes in the breast. They found that breast volume, water content, and the T1 relaxation time were lowest on days 6-15, began to rise on day 15, and peaked on day 25. Between days 16 and 28, parenchymal volume, T1 relaxation time, and water content increased by 39%, 15%, and 24%, respectively. Five days after menses onset, parenchymal volume fell by 30% and water content fell by 17%. Changes of smaller magnitude were reported by Graham et al. (25) who used a 1.5-tesla magnetic resonance imager on seven subjects twice a week over at least one menstrual cycle. They concluded that “variations in breast tissue during the cycle are detectable . . . . The variation in breast water content and fibroglandular volume correlates qualitatively with the known effects of hormones on breast structures.” However, Graham et al. (25) found the variations to be sufficiently small that they are “unlikely to impact on clinical assessment of unenhanced magnetic resonance breast images.” Kuhl et al. (26) recommended that magnetic resonance imaging should be done within the second week of the menstrual cycle (days 7-13) because during this phase there will be statistically significantly fewer diagnostic problems. Muller-Schimpfle et al. (27) concluded that, during days 7-20 of the menstrual cycle, there was a statistically significantly lower parenchymal enhancement with magnetic resonance imaging than from day 21 to day 6. Thus, “confidence in excluding or detecting constant ‘real' lesions” might be increased from day 7 through day 20. Breasts of postmenopausal women have less epithelial tissue and more fatty tissue and are radiologically less dense than breasts of younger women. Anastassiades et al. (28) evaluated breast composition in 112 mastectomy specimens from women at different ages and found that, at ages 31-40 years, 54% of breasts were classified as predominantly “solid tissue” (mammary parenchyma and fibrous stroma) and none of the breasts were extremely fatty. In specimens from women older than 70 years, 46% were fatty and only 8% had solid tissue. Brisson et al. (29), who analyzed data from 55 000 women in the Breast Cancer Detection Demonstration Project, reported that menopause was associated with decreased mammographic density. For example, 23% of premenopausal women aged 50-54 years had fatty breasts compared with 30% of postmenopausal women the same age. Spicer et al. (30) believe this is because menopause is associated with reduced mitotic activity in breast epithelial cells. Boyd et al. (31) observed a 9% reduction in percent density in women who became menopausal in the period of observation. Percent density was the proportion of radiologically dense breast expressed as a percentage of the entire projected area of the breast. It is generally accepted that increased mammographic density reduces mammographic accuracy. To increase understanding of the many variables that potentially reduce screening efficacy, factors are briefly reviewed that are known to affect density, such as menstrual cycle, reproductive history, body build, diet, and hormone use. One study to date has shown an association between phase of menstrual cycle and mammographic density. Density was increased in the luteal phase (9). As long ago as 1987, Bergkvist et al. (32) concluded that reproductive history has a lifelong impact on mammographic parenchymal patterns. Univariate analysis showed increased density with increasing age at first birth, but density was highest in nulliparous women. Brisson et al. (29) observed that the percentage of women with mammographic densities (breasts described as “glandular” or homogeneously dense in contrast to atrophic, fatty, or “ intermediate”) decreased as age, body weight, and parity increased; however, the percentage with densities increased with increasing height. White et al. (9) reported an inverse relationship between body mass index and mammographic breast density. Laya et al. (33) concluded that density is directly related to height and inversely related to parity and weight. de Stavola et al. (34) in a multivariate analysis showed that the odds for high-risk Wolfe grades (i.e., density) were statistically significantly lower with increasing Quetelet's index (i.e., body mass index) and with increasing parity in premenopausal and postmenopausal women. They observed that density increased with age until menopause, where it peaked, and then declined in the absence of HRT. This was also observed by Bergkvist et al. (32). Recently, Kolb et al. (35) have shown a very clear relationship between increasing age and decreasing mammographic density in 11 220 patients: densities involving more than 50% of breast volume decreased from a prevalence of more than 60% in women who were 30-39 years old, to slightly more than 20% in women who were 50-59 years old, and to about 10% in women who were 80-89 years old. A broad picture emerges suggesting that mammographic density is increased in association with nulliparity, increased age at first birth, increasing height, and HRT use. In contrast, mammographic density is decreased with increasing parity, increasing body mass index, and increasing age. It has long been suspected that the high fat intake that characterizes the North American diet at least partly explains the higher breast cancer incidence rates relative to other parts of the world. Because “extensive dense breast tissue seen on mammography is associated with an increased risk of breast cancer,” Boyd et al. (31) recruited women who had radiologic densities in more than 50% of their breast tissue and randomly assigned them to a low-fat diet (fat = 21% of calories) or to a control group (fat = 32% of calories). Two years later, they did follow-up mammography with 817 women. They observed that the area of dense tissue (expressed in square millimeters) was influenced most strongly by menopausal status with a lesser effect from weight change. In addition, the reduced-fat diet also had a statistically significant effect: after controlling for weight loss, menopausal status, and age at entry, the intervention group (on a reduced-fat diet) showed a statistically significant (P = .03) 6.1% reduction in the area of density compared with a 2.1% reduction in the control group. Although the effect reported is not large, it was one that occurred over a 2-year period and might well increase in magnitude if the intervention continued. It suggests one strategy that might reduce mammographic density. The consequences could be improved mammography and reduced risk of breast cancer. The fact that increased body mass index is associated with a decrease in density appears inconsistent. However, the explanation for the inconsistency may lie in the relative contributions of carbohydrate, fat, and protein to total caloric intake and the as yet undetermined interaction between estrogen and dietary components. Although HRT is generally accepted as being directly associated with mammographic density, a summary of current information can be confusing because this issue has been approached in diverse ways. Some studies have compared women using HRT to women who do not use HRT (nonusers); others have done before-after comparisons (before and after onset of HRT); and still others have done before-after studies comparing women initiating HRT with women not using HRT. HRT, the agent of interest, is variably defined with some studies including very detailed data on drug formulation and others not. And finally, the outcome measure, mammographic density, can be subjectively and qualitatively rated as with the Wolfe classification (36) or objectively and quantitatively rated with actual measurement of density as a percentage of total breast area (37). Not all studies have evaluated intraobserver and interobserver agreement. Based on qualitative analyses and a comparison of nonspecified HRT users with women who did not use HRT, a higher prevalence of mammographic densities has been shown with hormone use (34,35,38). Other studies have compared mammograms before and after initiation of hormone therapy. In a series of 30 women, HRT was associated with qualitatively increased density in five women (17%), all on combined HRT. Increased density was not observed in the 14 women receiving estrogen alone (39). Stomper et al. (40) also reported qualitative increases in density after the onset of HRT that occurred in 24% of the women. The difference observed when comparing women on combined HRT versus women on estrogen alone was not statistically significant, although increased density was more prevalent in the former group. Another before-after study included 33 postmenopausal women on HRT and a comparison group of 31 nonusers. Over time, there was a subjective increase in density in nine women (27%) using hormones with no differences noted by type of therapy (41). In contrast, a study of 41 women before and after initiation of combined HRT showed that a qualitative increase in density occurred in 24% and a quantitative increase occurred in 73%. However, the quantified increase was visually detectable in only 32% of women (33). These authors speculated that progesterone may cause “greater mammographic changes” than estrogen, which is consistent with the observations of Baines et al. (8) and White et al. (9). Further confirmation of this concept is found in a Dutch study (42) of 81 women. This study reported a qualitative increase in density in 8.7% of women on estrogen alone compared with 31% of women on combination HRT. Additional support comes from a study of more than women that found that of women on combined HRT had an increase in qualitative density compared with of women on estrogen alone and of nonusers. It is that HRT increase mammographic density. an association has been observed and is consistent with the that the decreased density seen in postmenopausal women is related to of hormone increased density is a progesterone effect to be further Baines et al. (8) reported an increased risk of false-negative mammograms from women in the luteal phase of the menstrual cycle and of and compared with from women in the follicular phase and of and The increased risk was observed only in women who had ever used The luteal decrease in sensitivity is consistent with the observations of Fowler et al. (24) and Graham et al. These studies reported that water content increased in the luteal phase. The luteal decrease is also consistent with the observations of Graham et al. who reported that relative water content, as by magnetic resonance showed a with percent density, as by Further support for comes from White et al. after mammograms from menstruating women, concluded that there was a small but statistically significant increase in density in the luteal phase. Given that increased mammographic density has been reported to be associated with reduced mammographic which to a increase in false-negative mammograms investigators have the effect of factors associated with increased density, such as HRT and menopausal status, on the sensitivity of Laya et al. postmenopausal women who were more than years old and found that the current use of estrogen replacement therapy was associated with lower and sensitivity of screening mammography when current users were compared with and The mammographic with for and current users of estrogen replacement therapy were = = and = respectively. In contrast, were at and with et al. screening in more than 11 000 postmenopausal women in and did not a decrease in sensitivity in women using HRT. was in current users compared with in former and These authors concluded their are with Laya et al. because of the However, because very different were it is difficult to the et al. mammographic sensitivity in premenopausal and postmenopausal women and at a the age effect with sensitivity in women who were years old or older and for women who were younger than years old. This difference was more if the comparison was to women with fatty where sensitivity was for older women and for younger women. These the is there an reduction in sensitivity in the absence of density? it was only in women older than age years that a density effect on mammographic was In this age group, higher sensitivity was observed in women with fatty breasts than with dense A was not observed in women younger than age However, it should be noted that in this false-negative mammograms were because interval between screening were not to be of false-negative mammograms if the interval cancer was of false-negative mammograms was also by the absence of a clinical breast examination at the time of the screening which may be for younger women In to the opportunity for an of actual false-negative screening the qualitative density were by a with no report of intraobserver variation or as has been reported in other studies (33). the used for sensitivity are at and 25 the of et al. are difficult to with the that factors that increase density may explain decreased accuracy efficacy of screening mammography in younger women. Given the evidence that has been factors other than density are including variations in mammographic and long an interval between screening and more growth rates in younger women. on the that hormone therapy is to increased density and decreased mammographic et al. found that HRT for days in of mammographic in of They concluded that hormone may both sensitivity and of This conclusion is consistent with the concept that HRT normal involution in postmenopausal breasts HRT may involution to there are and imaging changes in the breast tissue during the menstrual cycle. There is also evidence to believe that the postmenopausal breast and this of involution may be by that mammographic density appears to be related to age, age at first birth, height, body weight, parity, diet, phase of menstrual cycle, menopausal status, and exogenous and and that increased mammographic density reduces mammographic accuracy and screening efficacy. is not for all women. in detecting breast cancer early is known to be improved by technical and What is not known is the combined effects and of reproductive history, age, to exogenous phase of menstrual cycle, and diet are in the of Given the high to controlling breast cancer in North American it is crucial to give to all factors that may the of many women are to their and to as as from breast cancer. It only to current at a What may women to They may to their fat intake and in reduce the risk of reduce one risk factor for breast cancer and the accuracy of their mammograms by reducing mammographic density women may to their mammograms in the follicular phase of their cycle to mammographic accuracy women may to HRT for a period before their mammograms to the of mammograms Some women HRT and to control postmenopausal One of such a be the of increased mammographic density, which is observed in about of HRT users What to be women with cyclic mastalgia, with association with increased density, a group less to benefit from mammography What are the at the cellular that are associated with or explain mammographic parenchymal there a between oral use and an increased risk of false-negative mammograms in the luteal phase of the menstrual cycle? does HRT qualitatively increase density in only about of In women on combined HRT, should mammograms be to the What factors are associated with the decrease in mammographic sensitivity in younger women with fatty breasts compared with older women observed by et al. most of all be a understanding of the relative importance of variables such as age, reproductive cycle, diet, body mass, menstrual cycle, and HRT and the of their should be done to identify the biologic for mammographic efficacy. Even be a imaging that the that mammography” can of be a that could identify women who were at increased risk of breast cancer or who had breast cancer. The tangled web still to affect breast cancer screening efficacy to affect breast cancer screening efficacy to affect mammographic accuracy to affect mammographic accuracy
No takes yet. Share an insight, caveat, or question.
Cornelia J. Baines (1999) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: