Diet

Alcohol consumption is the best-established dietary factor associated with increased risk for breast cancer, although there is considerable interest in other dietary components.⁸ Initial studies correlated fat intake and decreased intake of fruit and vegetables with an increased risk for breast cancer, but more recent prospective studies have failed to confirm those observations.^9–12 One randomized study (Women’s Healthy Eating and Living) in breast cancer patients of a low fat/high fruit and vegetable diet did not demonstrate a decreased risk of breast cancer recurrence.¹¹ However, a second randomized study (Women’s Intervention Nutrition Study) did demonstrate a lower risk of recurrence with a decreased intake of dietary fat associated with modest weight loss in breast cancer survivors.¹²

Dietary phytoestrogens, such as those found in soybeans, have a chemical structure that is similar to that of 17β-estradiol and can bind to the ER to compete with estrogen. Their consumption may have a weak protective effect against breast cancer. More recent studies suggest an inverse association of soy intake and breast cancer in Asian populations (in which soy intake is generally high from birth) but not in non-Asian populations.¹³ Although there is significant interest in the association of vitamin D intake and breast cancer risk, there are insufficient data to draw conclusions regarding potential benefit at this time.¹⁴

Ionizing Radiation

The accumulated knowledge about radiation-related breast cancer risk in women derives mainly from epidemiological studies of patients exposed to diagnostic or therapeutic radiation and of the Japanese atomic bomb survivors. Low-dose radiation exposure to the breast is carcinogenic, and risk increases as a linear function of increasing dose. The risk is particularly high in the developing breast (i.e., before the age of 20 years), and is minimal for women exposed after menopause.¹⁵ Therapeutic radiation, particularly extended field radiation for Hodgkin lymphoma, has been associated with an increased risk of breast cancer. The age of exposure is important with the highest breast cancer risk seen in women undergoing mantle radiation around puberty.¹⁶

Exogenous Hormones

The NHS originally was set up to determine whether oral contraceptive usage was associated with an increase in breast cancer. An analysis of data from that study and more than 50 additional studies has confirmed that current users of exogenous hormones have a slightly increased risk of developing breast cancer.¹⁷ However, the mortality associated with this risk appears to be offset by a decreased risk of colon and ovarian cancer, such that there is no increase in overall cancer mortality associated with their use.⁶ The Women’s Health Initiative randomized study of hormone replacement therapy (HRT) in postmenopausal women demonstrated an increased risk of invasive breast cancer with the combination of estrogen plus progesterone therapy,¹⁸ but not with estrogen alone.¹⁹ Other studies have also demonstrated that breast cancer risk depends on type of HRT (estrogen alone vs. combination) and time of initiation (with an increased risk seen when HRT was started soon after menopause).²⁰ Age-adjusted incidences of invasive breast cancer decreased in the United States between 2002 and 2003 by 7%, reflecting an absolute decrease of approximately 14,000 breast cancer cases (Fig. 91-1). This decrease was limited to women 50 years of age or older, occurred primarily in ER-positive tumors, and coincided closely with a dramatic decrease in the use of HRT following the publication of the Women’s Health Initiative results that demonstrated no cardiovascular benefit for HRT.⁴

Reproductive Factors and Endogenous Hormones

Breast cancer risk consistently has been correlated with an earlier age of menarche, later age of menopause, nulliparity, and late age of first birth, all of which determine the cumulative number of ovarian cycles. This finding is consistent with the correlation between estrogen levels and breast cancer risk. In postmenopausal women, an increased risk of breast cancer has been demonstrated in those with estradiol levels in the highest quartile.²¹

Breast Cancer Genome

The elucidation of genetic aberrations in cancer genomes (mutations, deletions, and amplifications) has led to a better understanding of cancer pathophysiology and is providing targets for rational drug development. With the determination of the DNA sequence of the normal human genome, efforts have shifted toward a more comprehensive analysis of cancer genomes. The first report of a comprehensive sequence analysis of 21,000 of the best studied and annotated human genes was published, including data from 11 breast and 11 colon cancer samples.⁹⁸ This screen rediscovered cancer genes known to be associated with breast cancer (e.g., p53 and BRCA1) as well as genes not previously characterized as associated with cancer. This impressive effort suggests that an individual breast cancer harbors an average of about 90 mutant genes, of which a minority (approximately 12) are expected to contribute to the neoplastic process. Unfortunately, the functionally important mutations often differ among tumors (i.e., no tumor had more than six mutated cancer genes in common with any other tumor). However, when the genes were analyzed by functional groups, mutations in signal transduction pathway and transcription factor genes were found in nearly all breast tumor samples. As detailed below, rapid advancements in sequencing technology and bioinformatics have greatly expanded the information available.

Molecular Profiling in Breast Cancer

Simultaneous genome-wide determination of the relative abundance of RNA species (i.e., transcriptional profiling) combined with novel bioinformatic approaches has led to the development of molecular classifications based on continuous expression ranges of thousands of genes (as opposed to a binary determination of discrete factors, i.e., ER/PR and HER2). Tumors can be classified either according to transcripts that tend to segregate together according to underlying biological differences (unsupervised clustering) or sorted by a given end point such as prognosis or response to therapy (supervised clustering). The first of these studies identified an “intrinsic gene list” of 534 genes that defined five biological subtypes of breast cancer including the basal type (HER2-positive, cytokeratins 5 and 17 positive), the “triple negative” type (HER2-, ER-, and PR-negative), two luminal types (so-called luminal-A and luminal-B), and the normal-like type (Fig. 91-2).⁹⁹ Studies by others have confirmed large-scale gene expression differences between ER-positive and ER-negative tumors, and have shown additional molecular subsets within these categories.¹⁰⁰ This molecular classification of breast cancer is robust in that it is observed regardless of microarray platform type, and the various subtypes correlate with overall prognosis.^101,102 Luminal-type cancers have the most favorable long-term survival, whereas HER2-positive and basal-like cancers may be more sensitive to chemotherapy but have the worst overall prognosis. Thus breast cancer is not a single disease with heterogeneous ER and HER2 expression but appears to be composed of three to five molecularly and clinically distinct subtypes.

A similar approach can be taken by assaying DNA amplifications and deletions across the genome.¹⁰³ Such data recently have been mapped against both the “intrinsic gene set” and clinical outcome data, and have the potential to identify high-level DNA amplification that, similar to HER2, may be useful as therapeutic targets.¹⁰⁴ However, it should be noted that although these classifications are robust across a population, there is no standardized assay of measurement for assigning a molecular class to a new individual case. Furthermore, much of the molecular heterogeneity overlaps with conventional histopathology and is captured by determination of ER, PR, HER2 status, and tumor grade.¹⁰⁰

In addition, although the intrinsic gene set and its derivations correlate with prognosis, this correlation does not easily translate into the specificity and sensitivity required for clinical decision making. One of the greatest clinical challenges has been determining the value of adjuvant chemotherapy for individual women with early-stage breast cancer, where many patients are treated but only a few directly benefit.

Several multigene prediction scores that correlate with outcome and provide more accuracy than traditional histopathological markers have been retrospectively derived from microarray data, and, in some cases, prospectively validated on additional datasets. The best known examples are a 70-gene predictor set (Fig. 91-3)^105,106 and a 21-gene predictor set,^107,108 and both are commercially available. The 21-gene test consists of 16 target transcripts and five reference transcripts, including ER, PR, and HER2, and a “proliferation” group of five genes (which includes Ki-67 and cyclin B₁); and a carefully constructed weighted formula is used to determine the overall “recurrence score.”^107,108 Although this recurrence score provides a continuous risk variable, for clinical use it is pared down to three categories of low, intermediate, and high (of note, given the weighting schema, it is unlikely that a HER2-positive tumor would be classified as “low risk”). It is interesting that these two gene sets as well as the “intrinsic gene set” described earlier and yet another “wound healing” signature all seem to perform similarly when tested against the same clinical data set, even though they share few to none of the same genes.¹⁰⁹ The ability of the 70-gene assay to determine which ER-positive node-negative patients are most likely to benefit from adjuvant therapy is being prospectively tested in the Microarray in Node-Negative Disease May Avoid Chemotherapy (MINDACT NCT00433589) trial. Similarly, the Trial Assigning Individualized Options for Treatment (Rx) (TAILORx) NCT00310180) study, which is testing the utility of adjuvant chemotherapy in patients who have an “intermediate risk” according to the test, has completed enrollment; low-risk patients received hormonal therapy only, whereas all high-risk patients were treated with both chemotherapy and hormonal therapy. The RxPONDER study (NCT01272037) is addressing a similar question in patients with node-positive disease, where patients with a recurrence score <=25 are being randomized to endocrine therapy alone or also receive chemotherapy. In the meantime, the Oncotype assay has been incorporated into the National Comprehensive Cancer Network (NCCN) algorithm for node-090 negative, ER-positive breast cancer patients.

Disseminated and Circulating Tumor Cells

From 20% to 45% of patients with primary operable breast cancer and 70% of patients with metastatic disease have disseminated tumor cells (DTCs) that can be detected in bone marrow¹¹² or in lymph nodes.¹¹³ Historically, DTCs usually have been evaluated by IHC staining for various cytokeratin markers. Although their existence has been associated with a worse prognosis,¹¹² the need for a relatively invasive procedure (a bone marrow biopsy) and the lack of sufficient specificity and sensitivity, as well as method standardization, have prevented the routine use of such information in clinical decision making.¹¹⁴ However, it is now possible to measure circulating tumor cells (CTCs) accurately by using an automated immunobead enrichment followed by pancytokeratin staining. Using this approach, patients with metastatic breast cancer and five or more CTCs per 7.5 mL of blood are predicted to have a significantly worse disease-free survival (2.1 months vs. 7 months) and overall survival (8.2 months vs. 18 months) than patients with fewer CTCs.¹¹⁵ This measurement may enable routine, minimally invasive assessment of disease progression and response to therapy. Furthermore, the ability to isolate CTCs allows the study of molecular markers such as HER2 amplification and their relationship to the primary tumor in real time. This approach has the potential not just to monitor patients but also to select specific therapy based on a molecular diagnosis obtained at the time of therapy (as opposed to relying on the original diagnostic biopsy). The clinical utility of CTCs is now being prospectively tested in a trial randomly assigning patients with 5 or more CTCs per 7.5 mL after 1 month of therapy to continue the same course or change systemic therapy.

Breast Cancer Stem Cells

Cancer cells have two somewhat contradictory traits. Although they originate from a single clone, they often display marked genetic, biological, and morphologic heterogeneity. Historically, the prevailing hypothesis was that genomic instability contributed to continuing evolution and emergence of different and ultimately more virulent subclones. This theory has been invoked to explain both progression (in terms of metastatic spread) and emerging resistance to therapy. Recently, an alternative explanation was provided by ground breaking work elucidating the “cancer stem cell” model (Fig. 91-4).¹¹⁶ Initially demonstrated in hematologic reconstitution and malignancies, cancer stem cells, although constituting a small minority of the cancer cell mass, retain the capacity for asymmetric division in which they both self-renew and give rise to more differentiated daughter cells. The latter, although forming the bulk of the tumor, are themselves incapable of endless replication. Given that stem cells are long-lived and slow to replicate, this would explain both the accumulation of genetic lesions over time as well as resistance to therapy targeting dividing cells. It would also explain the frequently observed concordance between primary and metastatic cells. For a long time this model for epithelial cancers has been theoretical, as it was technically difficult to identify these cells. However, such cells have been isolated from both normal mammary glands and breast cancers, based on the presence and absence of specific surface markers. Thus in mice, a single^Lin+D29 (hi)CD24⁻ cell can reconstitute a complete functional mammary gland in vivo, contributing to both luminal and myoepithelial lineages.¹¹⁷ Similarly, the ability to reconstitute human breast cancers in immunocompromised mice seems to reside within a minority CD44⁺, CD24(−/low) subclone. Notably, tumors formed by these cells gave rise to mixed populations of epithelial cells recapitulating the morphology of the parent tumor.¹¹⁸ These findings have several important implications for breast cancer therapy. As a first step, the molecular and biochemical pathways that control cancer stem cell division (such as Wnt and Notch pathways) might provide novel targets for therapy.¹¹⁹

Increased Surveillance

As discussed in the section on mammographic screening, there is significant debate about the age at which to begin mammographic screening in average-risk women. Nonetheless, ample evidence supports increased surveillance in women at high risk for breast cancer. A number of prediction models for breast cancer risk are available, and surveillance strategies are developing rapidly, particularly for women who are BRCA1 and BRCA2 mutation carriers.¹²⁰

There is general agreement that women at high risk for breast cancer are candidates for mammographic surveillance in their 40s, and in some cases, earlier initiation of mammographic screening may be appropriate. Shorter screening intervals may be necessary because of the accumulating evidence of unacceptably high rates of interval breast cancers, even with annual screening, in BRCA1 and BRCA2 mutation carriers. Other imaging modalities, such as magnetic resonance imaging (MRI), may be considered in these high-risk groups. Finally, although controversy remains about the role of breast self-examination in the general population, experts in the surveillance of high-risk women still recommend breast awareness for women with BRCA1 and BRCA2 mutations.¹²¹

Behavior Modification

Multidisciplinary centers that provide counseling for women who are at substantial risk for breast cancer are well established throughout the United States. These multidisciplinary consultations provide recommendations for a range of surveillance and interventional approaches. In addition to the available surgical and medical preventive strategies, modification of lifestyle factors, such as obesity, high-fat diet, alcohol consumption, and lack of exercise, should be discussed. It probably will be many years before it is possible to say conclusively whether lifestyle modification changes a woman’s risk. In the meantime, patients and their physicians must choose whether and how to modify lifestyle factors.

Chemoprevention

Tamoxifen was studied as a chemopreventive agent for breast cancer in four randomized prospective clinical trials. A significant reduction in breast cancer risk with tamoxifen was seen in the National Surgical Adjuvant Breast and Bowel Project (NSABP) P-1 study¹²² and the International Breast Cancer Intervention Study (IBIS-I).¹²³ Further analysis of the patients in the NSABP P-1 study showed that tamoxifen reduced the incidence of breast cancer among BRCA2 carriers by 62%, but not among BRCA1 carriers.¹²⁴ Two European chemoprevention studies performed at the Royal Marsden Hospital¹²⁵ and by the Italian Tamoxifen Prevention Study Group,¹²⁶ respectively, did not show a decrease in the incidence of breast cancer in women from high-risk families using tamoxifen.

Based in large part on compelling evidence from the NSABP study, which showed a 49% reduction in overall risk of invasive breast cancer with the use of tamoxifen, a number of review groups, including the American Society of Clinical Oncology Technology Assessment Working Group, recommended that tamoxifen should be considered to reduce the risk of breast cancer in women with a defined 5-year projected risk of 1.66% or greater.¹²⁷ However, this group of studies had considerable methodologic differences, and further studies are needed to better define the role of chemoprevention in high-risk women. Furthermore, the NSABP study was not designed to show a difference in death rate from breast cancer as a primary end point.

Two selective estrogen-receptor modulators (SERMs), tamoxifen and raloxifene, were compared in the NSABP STAR trial, and long-term follow-up of this trial demonstrated that raloxifene was slightly less effective in decreasing invasive breast cancer than tamoxifen, but was also less toxic.¹²⁸ Finally, the aromatase inhibitor, exemestane, has been shown to decrease the risk of breast cancer in a randomized placebo-controlled, double-blind trial study. In the MAP.3 trial exemestane significantly reduced invasive breast cancers in postmenopausal women who were at moderately increased risk for breast cancer. During a median follow-up period of 3 years, exemestane was associated with no serious toxic effects and only minimal changes in health-related quality of life. Longer follow-up is required to confirm these data, including possible long-term side effects of the medication.¹²⁹

For all of these agents it is not known whether they reduce the incidence of breast cancer by preventing the formation of cancer or by treating small, clinically occult cancers. The use of any of these agents requires an in-depth assessment of the risks and benefits of the specific drug in the individual patient; practical algorithms for assessing risk and benefit for tamoxifen and raloxifene chemoprevention are now available.¹³⁰

Prophylactic Mastectomy or Oophorectomy

Prophylactic mastectomy and oophorectomy are controversial preventive approaches to breast cancer. Current data show that prophylactic mastectomy is effective in preventing breast cancer in patients with either a strong family history of breast and/or ovarian cancer¹³¹ or a genetic predisposition to breast cancer.⁵⁰ Additional data show that prophylactic contralateral mastectomy can reduce the risk of breast cancer in patients with a previous diagnosis of unilateral disease.¹³² A high prevalence of premalignant lesions is seen in prophylactically removed breasts from women who are at hereditary risk for breast cancer.¹³³ Reduction of breast and ovarian cancer risk with risk-reducing surgery has been shown to reduce the mortality for these diseases.⁵⁰

The total and incremental life expectancy with prophylactic mastectomy is difficult to demonstrate. Statistical analysis has shown that depending on the assumed penetrance of the BRCA mutation, compared with surveillance alone, 30-year-old patients with early-stage breast cancer who have BRCA mutations may gain 2.9 to 5.3 years from prophylactic bilateral mastectomy and 0.6 to 2.1 years from prophylactic contralateral mastectomy.¹³⁴ A recent survival analysis suggested that prophylactic mastectomy at age 25 years and oophorectomy at age 40 years maximized survival at 70 years of age; however, substituting enhanced breast cancer screening (with MRI and mammogram) for mastectomy resulted in similar survival.¹³⁵ Given the impact on quality of life, the choice of prophylactic mastectomy is very personal. Still controversial is the gain in quality of life in these high-risk women who chose to have prophylactic mastectomy.¹³⁶

Oophorectomy was first performed as a therapeutic procedure for advanced breast cancer more than 110 years ago and as hormonal adjuvant treatment of primary breast cancer more than 50 years ago. Uncertainty remains about the appropriate use of oophorectomy in adjuvant treatment, so it is not surprising that preventive oophorectomy in high-risk women remains controversial. Again, the identification of women with BRCA1 or BRCA2 mutations has provided an opportunity to study this prophylactic procedure. Studies with prospective follow-up have demonstrated that oophorectomy is associated with a decrease in breast and ovarian cancer risk, and enhanced breast cancer-specific survival, ovarian cancer-specific survival and overall survival. ⁵⁰

Screening

The efficacy of screening for occult cancer depends heavily on the following factors: (a) tumor growth rate; (b) the sensitivity of the test related to tumor volume; and (c) the interval between screens. Growth rates of breast cancer vary, and some younger women have more rapidly growing tumors. In determining the efficacy of screening, four biases must be considered: lead-time bias, length bias, selection bias, and overdiagnosis bias. Lead-time bias is the interval that the diagnosis has been advanced by screening. Length bias concerns the timing of detection. When screening is infrequent, fast-growing tumors are not detected as early in their natural history as more slowly growing tumors. Thus the outcome of cancers detected by screening is better than that for interval cancers. Selection bias is an obvious factor. For example, women who participate in breast cancer screening may have been shown to be more health conscious; they also are more likely to obtain Pap smears, use seat belts, and not smoke.¹⁴² Thus it is likely that their outcomes would be better, even in the absence of screening. Overdiagnosis has always been a concern in screening for cancer. Detection of breast lesions of questionable malignancy affects mortality data, because these lesions would not be diagnosed without screening. It is not known how many of these very early in situ cancers would progress to become invasive malignancies.

These biases complicate the design and evaluation of screening trials. In evaluating the results of screening trials, several points must be remembered. First, the survival and case fatality rates of patients with breast cancer are not adequate end points for determining results. Second, the closest end point for studying the mortality rate in a population is an observed reduction in the incidence of advanced disease. Finally, if screening detected an excessive number of cancers that were not destined to be lethal, then the death rate among the screened women would not be altered and no benefit would be observed.

Before 1977, no formal guidelines existed for screening women for occult breast cancer. The publication of the Health Insurance Plan (HIP) of Greater New York Screening Project, conducted in the 1960s, led the NCI and the American Cancer Society to support a nationwide breast cancer screening project to evaluate further the efficacy of screening by mammography and physical examination. The Breast Cancer Detection Demonstration Project took place in 29 centers, and 280,222 women were screened over a period of 5 years.¹⁴³

Although the study is well documented, it is important to review the design and findings of the HIP landmark clinical trial. The HIP study invited women 40 to 69 years of age who were enrolled in the HIP of Greater New York to be screened for breast cancer. The HIP project was the first randomized controlled trial and included annual clinical examinations and mammography for 4 years. Using pairwise allocation, 62,000 women were randomly assigned to the study group or the control group. Of the total population, 45% were 40 to 49 years of age at study entry. Of the 31,000 women in the study group, 65% attended one or more screening sessions. Because it was not possible to identify within the control group women who would decline screening, the outcome of the control group was evaluated in relation to the total study group.

A surprising aspect of the initial evaluation of the results was the striking difference in the effectiveness of screening at different ages. At 5 years, the HIP study found a 50% decrease in mortality rate in women older than 50 years of age, but only a 5% decrease in mortality rate in women younger than 50 years of age. With follow-up, the mortality rate in women younger than 50 years of age showed a 23.5% decrease, and it became statistically significant at year 18 of follow-up. In women older than 50 years of age, the reduction in mortality became evident at approximately the fourth year of follow-up. Significantly, a reduction in mortality rate in younger women took longer than a decade to manifest. Women who refused screening or who had interval cancers had no reduction in mortality rate.

For the first time, the results of the HIP study provided strong evidence that early detection of breast cancer increased the duration of survival. Since the HIP trial was published, seven additional randomized trials of mammography and two of breast self-examination have been reported. The results of these trials have been included in a 2002 meta-analysis commissioned by the United States Preventive Service Task Force (USPSTF) (Table 91-4).¹³⁷ Between the ages of 39 and 74 years, the summary relative risk (RR) for death from breast cancer was 0.84% for the use of screening mammography. Much like the findings of some of the individual studies included in this analysis, greater reduction in risk was seen in women older than the age of 50 years. The summary RR for women 40 to 49 years old was 0.85, compared with 0.78 for those 50 years of age and older. Of interest, this same meta-analysis noted “direct evidence of effectiveness among older women, limited to two trials that included women over 65 years of age.’

Table 91-4

Results of Randomized, Controlled Trials of Mammography Among Women 39 to 74 Years of Age

Study	Age (years)	Median Follow-up (years)	Breast Cancer Deaths/Total Women		Breast Cancer Death Rate per 1000 Women		Relative Risk for Death from Breast Cancer (95% Cl)	Absolute Risk Reduction per 1000 Women	No. Needed to Invite to Screening*
			Screened Group	Control Group	Screened Group	Control Group
			n/n
MAMMOGRAPHY ALONE
Stockholm†	40–64	13.8	82/39 139	50/20 978	2.10	2.38	0.91 (0.65–1.27)	0.288	3468
Gothenburg†	39–59	12.8	62/20 724	113/29 200	2.99	3.87	0.76 (0.56–1.04)	0.878	1139
Malmö†	45–70	17.1	161/21 088	198/21 195	7.63	9.35	0.82 (0.67–1.00)	1.712	584
Swedish Two-County Trial‡	40–74	17	319/77 080	333/55 985	4.14	5.95	0.68 (0.59–0.80)	1.809	553
MAMMOGRAPHY PLUS CBE
CNBSS-1§	40–49	13	105/25 214	108/25 216	4.16	4.28	0.97 (0.74–1.27)	0.12	—
CNBSS-2¶	50–59	13	107/19 711	105/19 694	5.43	5.33	1.02 (0.78–1.33)	−0.097	—
HIP	40–64	16	232/30 239	281/30 256	5.46	6.89	0.79	1.438	883
Edinburgh#	45–64	13	156/22 926	167/21 342	6.80	7.82	0.79 (0.60–1.02)	1.020	980

CBE, clinical breast examination; CI, confidence interval; CNBSS, Canadian National Breast Screening Study; HIP, Health Insurance Plan of Greater New York.

^*Number needed to invite to screening to prevent one death from breast cancer 13 to 20 years after randomization.

^†Nyström L, Andersson I, Bjurstam N, et al. Long-term effects of mammography screening: updated overview of the Swedish randomized trials. Lancet 2002;359:909–919.

^‡Tabár L, Vitak B, Chen HIH, et al. The Swedish Two-Country Trial twenty years later. Updated mortality results and new insights from long-term follow-up. Radiol Clin North Am 2000;38:625–651.

^§Miller AB, To T, Baines CJ, Wall C. The Canadian National Breast Cancer Screening Study-1: breast cancer mortality after 11 to 16 years of follow-up. A randomized screening trial of mammography in women age 40 to 49 years. Ann Intern Med 2002;137:305–312.

^¶Miller AB, To T, Baines CJ, Wall C. Canadian National Breast Screening Study-2: 13-year results of a randomized trial in women aged 50–59 years. J Natl Cancer Inst 2000;92:1490–1499.

Shapiro S, Venet W, Strax P, Venet L. Current results of the breast cancer screening randomized trial: the health insurance plan (HIP) of greater New York study. In: Day NE, Miller AB, editors: Screening for breast cancer. Toronto: Hans Huber; 1988. pp. 3–15.

^#Alexander FE, Anderson TJ, Brown HK, et al. 14 Years of follow-up from the Edinburgh randomized trial of breast-cancer screening. Lancet 1999;353:1903–1908.

Data from Humphrey LL, Helfand M, Chan BK, et al. Breast cancer screening: a summary of the evidence for the U.S. Preventive Services Task Force. Ann Intern Med 2002; 137:347–360.

The controversy about the efficacy of screening for women 40 to 49 years of age is ongoing because breast cancer is less frequent in this age group, and there is a greater likelihood that mammographic abnormalities will prove benign on biopsy. To further address this controversy, the Cancer Screening Evaluation Unit in Sutton, United Kingdom, conducted a trial called Age, directed at women ages 40 to 41 years, comparing annual screening mammogram with routine medical care. In selecting this age group, the trial targeted women who would still be younger than 50 years through 10 years of follow-up screening. This trial did demonstrate “a reduction in breast cancer mortality” in the intervention group, in relative and absolute terms, which did not reach statistical significance.¹⁴⁴

Because of the continued controversy about the value of mammography, especially in younger and older women, a 2009 systematic review was undertaken to update evidence since 2002 and it focused especially on women of average risk aged 40 to 49 years and 70 years or older.¹³⁸ For women aged 39 to 49 years, eight trials provided data and the pooled RR for breast cancer mortality for women randomly assigned to mammographic screening was 0.85 which corresponded to a number needed to invite to screening to prevent 1 breast cancer death (NNI) of 1904. For women aged 50 to 59 years six trials provided data that showed a pooled RR of 0.86 with an NNI of 1559, and for women aged 60 to 69 years two trials gave a RR for breast cancer mortality of 0.68 with an NNI of 337. A single trial of women aged 70 to 74 years gave a RR of 1.12 with broad confidence intervals and inability to provide an NNI. Other discussion centers around the optimal frequency of mammography, as well as the value of clinical and breast self- examination.

Table 91-5 lists the current screening guidelines for mammography for the American Cancer Society, NCCN, and the USPSTF.¹⁴⁵

Patient Compliance

Over the last decade, aggressive public education programs conducted by the American Cancer Society, the NCI, and public interest groups have led to a dramatic increase in mammographic screening in the United States. A particular concern was its marked underuse by minority women, although the use of mammography has increased dramatically and the racial gap has virtually disappeared. A recent study concluded that ethnicity was “not associated” with seeking and obtaining a screening mammogram, once socioeconomic factors are considered.¹⁴² Interventions associated with an increased use of screening included providing women with information, especially by a health care provider urging compliance, and vouchers for free mammography. Media intervention has little effect on compliance and limited long-range benefit, but educational interventions targeting primary care physicians should become a priority.

Sensitivity of Mammography

Mammography is more effective in detecting occult malignancy as patients age and breast tissue is replaced with fat. Breast density most directly affects the risk of false-negative and false-positive interpretations; it is affected by several factors. Age is the most significant factor, because premenopausal women are more likely to have breasts that are more glandular and thus more dense. Mammographic sensitivity may be increased by scheduling evaluation during the first 2 weeks of the menstrual cycle, corresponding with the follicular phase. HRT in postmenopausal women may increase breast density.

The sensitivity, or accuracy, of mammography also is affected by the experience of the radiologist. Features with the highest positive predictive value for carcinoma were spiculated margins, irregular shape, linear calcification, and segmented or linear calcification distribution. In general, the positive predictive value of a Breast Imaging Reporting and Data System (BI-RADS) category 5 lesion is greater than 80%, whereas that of a category 4 lesion approaches 30% to 40%. The last decade has seen the development of full-field digital mammography. A recent multicentric trial comparing conventional with digital mammography showed an improvement over conventional film mammography in the detection of breast cancer in young, premenopausal, and perimenopausal women, and in women who have dense breasts. However, there was no significant difference in diagnostic accuracy between digital and film mammography in the population as a whole or in the other predefined subgroups.¹³⁹

Other Methods of Screening

To overcome some of the variables that affect the predictive value of mammography, a number of new technologies are being explored. Emerging techniques for breast screening, such as MRI, tomosynthesis, scintimammography, and ultrasound are under study in selected women who present difficulties in screening with mammography.¹⁴⁶ Data increasingly suggest that women with very dense breasts and those with BRCA1 and BRCA2 mutations are more likely to have clinically occult carcinomas detected by other screening methods than by mammograms. Women who meet the criteria for hereditary breast cancer should be considered for additional screening as well. A number of options exist and could include the following: (a) initiation of screening 10 years before the onset of a first-degree relative’s breast cancer, or at 25 years of age; (b) the addition of MRI screening; (c) shorter screening intervals (e.g., every 6 months); and (d) the addition of ultrasound screening. A woman with certain BRCA1 or BRCA2 mutations may have a risk as high as 20% by 40 years of age, and MRI appears to be the most sensitive technique for screening these very young women. The role of clinical examination and breast self-examination in high-risk women seems more limited. Computer simulation modeling suggests that annual MRI at age 25 years with delayed alternating digital mammography at age 30 years is probably the most effective screening strategy in BRCA mutation carriers.¹⁴⁷ Although MRI shows improved sensitivity, it has lower specificity than mammography and may lead to additional fine-needle aspiration (FNA) or core biopsy in these young women at high risk.

Screening in the Elderly Patient

The life expectancy for women (all races) in the United States increased from 77.4 years in 1980 to 80.4 years in 2002 (see http://www.cdc.gov/nchs/fastats/lifexpec.htm). Biological parameters suggest a higher prevalence of higher-risk tumors in younger women, with lower-risk tumors (better differentiated and hormone receptor-rich) occurring in older women. Tabar and colleagues¹⁴⁸ showed that for a given tumor size, the likelihood of nodal involvement is lower in older women than in younger women, and these data in aggregate have led some, including the USPSTF, to suggest that it may be appropriate to increase the screening interval to 2 years in older women and stop screening mammography at age 75 years, although there are insufficient data to suggest an age when routine screening can be abandoned. At 75 years of age, the average additional life expectancy for a white or a black woman in the United States is 12 years or more. Even so, fewer elderly women are being screened annually, even those with a history of breast cancer. Although the benefits from screening appear to be seen in all age groups, and even in patients with comorbid disease, the magnitude of the benefit appears to diminish with age and with the severity of comorbid illnesses. Indeed it seems reasonable to forego screening mammography for women whose life expectancy is less than 5 years. Data from the U.S. National Health Interview Survey show an overall drop in the overall rate of mammography screening, from 70% in 2000 to 66% in 2005.¹⁴⁹

Masses

Benign masses typically are well defined, with sharp margins, and have little effect on the surrounding breast architecture. Fibroadenomas, papillomas, intramammary lymph nodes, and cysts are the most common causes of benign mammographic density (Fig. 91-5). Malignant breast densities classically have irregular borders that blend into the surrounding tissue and often appear to infiltrate the breast background tissue with a stellate appearance. Usually, there is some distortion of adjacent breast stroma. The mammographic abnormality that has the highest rate of malignancy is a mass density with associated calcification (Figs. 91-6 through 91-8). Examples of questions that should be asked include the following:

Calcifications

Malignant calcifications typically are linear, or small (<1 mm) in diameter, nonuniform in size, and clustered. Between 20% and 25% of clustered microcalcifications are positive for cancer on biopsy. Benign calcifications usually are larger and coarser, and are often round with smooth margins. Table 91-6 lists various types and distributions of calcifications. Benign causes of microcalcification include involuting fibroadenoma, arteriosclerosis, sclerosing adenosis, fat necrosis, and previous mastitis with ductal calcium deposits.

Ultrasound

Ultrasound is of greatest use when the mammographic findings are equivocal and when examination shows multiple areas of abnormality (Fig. 91-9). Ultrasound is very accurate (>95%) in diagnosing breast cysts. Cysts have well-demarcated, smooth margins and an echo-free center (Fig. 91-10), and usually are rounded and thin-walled and produce distal shadowing. Clear cysts require no further evaluation. Complex cysts that contain evidence of tissue or debris may be aspirated to clarify whether they are simply cysts or represent cystic degeneration of a tumor.

Ultrasonic image texture analysis is an effective way to distinguish benign from malignant breast lesions. This technique is a simple way to reduce the number of biopsies performed on benign lesions without missing additional cancers. Doppler flow analysis may provide additional information about solid lesions. Malignant tumors show increased blood flow compared with benign tumors, resulting in a characteristic blood flow signal.

Management of the Nonpalpable Mammogram Abnormality

The health care provider often is faced with a patient who is asymptomatic and has a normal breast examination, and an abnormal mammogram. The mammographic criteria for biopsy are fairly well accepted and include the following:

If these findings are seen on screening mammograms, diagnostic mammograms—including magnification views—are required to confirm the abnormality. Often, mammographic changes are not as clear as those listed in Table 91-6. The mammography report may state “cannot exclude an early breast cancer,” or “clinical correlation is recommended.” In these situations, all previous mammograms must be obtained for comparison. If the mammographic abnormality is believed to warrant histologic evaluation, several options exist: needle core biopsy (NCB) under ultrasound guidance if the abnormality is easily visualized by ultrasound; NCB with a dedicated stereotactic mammography unit; and wire localization and guided excisional biopsy with specimen radiography. Clearly, the expense of tissue sampling is considerably less if stereotactic NCB is used and proves reliable. Candidates for NCB include patients with highly suspicious mammographic findings, those with multiple suspicious lesions in the same breast, and those with lesions of low suspicion.

The procedure requires the patient to lie prone and to remain immobile for 30 to 45 minutes. NCB is not suitable for lesions that are close to the skin or for patients with very small or thin breasts (compressed to <3 cm). NCB is not optimal in the evaluation of radial scars or areas of diffuse microcalcification. Studies evaluating the efficacy of NCB indicate that a biopsy needle larger than 14 gauge is preferred, because it significantly increases the rate of concordance with surgical biopsy. Approximately one-half to two-thirds of patients diagnosed by stereotactic cores as having ductal carcinoma in situ (DCIS) ultimately may be found to have invasive cancer. Similarly, when core biopsy shows atypical ductal hyperplasia, open surgical biopsy shows DCIS in 30% to 50% of cases. Thus excisional biopsy is indicated in the following situations: (a) for a nondiagnostic core biopsy, (b) when atypical ductal hyperplasia is found, (c) when a papillary lesion is present, (d) when DCIS is present, (e) when the patient has a radial scar, or (f) when calcifications cannot be sampled. All patients with benign histologic features should have a 6-month follow-up and a two-view mammogram, and a system should be in place to ensure compliance.

When lobular carcinoma in situ (LCIS) is found in a NCB, in situ or invasive cancer can be found in about 25% of cases at surgery.¹⁵² On biopsy, approximately two benign tumors are found for each cancer that is discovered, but some authors argue that biopsy should be performed when the chance of finding cancer reaches 10%. Driving the rate of positive biopsy findings even lower, with the goal of decreasing the incidence of interval cancers, would significantly increase the overall cost of screening. The refinement in technology for the use of NCB suggests that this may be a way to reduce cost and provide an alternative to short-interval follow-up mammography. As with wire-localized biopsy, however, careful evaluation of the criteria for the use of NCB is required. Performing NCB biopsy on all patients would clearly defeat the goal of reducing cost. It is estimated that approximately 10% of women who undergo screening require either diagnostic biopsy or short-interval (4- to 6-month) follow-up. False-negative rates for NCB are reported to be as low as 2%, a rate similar to that for wire-localized–guided excisional biopsies. No cost savings will occur if every negative stereotactically guided NCB is followed by excisional biopsy.

The techniques used for wire-localized excisional biopsy are shown in Figures 91–12 through 91–15. Wire-localized excisional breast biopsy is highly accurate and has in good to excellent cosmetic results. The key to success is the accuracy of the needle or wire placement, which should go directly into the area to be removed. Other techniques of preoperative localization using radioisotopes, radioactive seeds, and cryo assistance, have been described with possible surgical advantages in obtaining more of the lesion in the surgical specimen and higher rate of negative margins.¹⁵³ The surgical specimen is sent to the radiology department, where it is filmed and compared with the mammogram to ensure that the suspicious lesion or microcalcifications are contained within the specimen. For patients with a benign biopsy finding, repeat mammogram is obtained 4 to 6 months later to reestablish a baseline and confirm clearing of the suspicious abnormality.

Management of the Magnetic Resonance Imaging Abnormality

With the increasing use of MRI in women at high risk for breast cancer, radiologists and breast surgeons often are faced with an MRI lesion not visible with mammography and ultrasound. A focal ultrasound in an MRI suspicious lesion fails to reveal a sonographic correlate in up to 77% of cases,¹⁵⁴ and MRI-guided percutaneous, vacuum-assisted biopsy should be performed.¹⁵⁵ Open surgical biopsy with MRI needle localization is the other option to diagnose a suspicious MRI lesion. Because the nonmagnetic wire used for the MRI localization is thinner than the wire used for ultrasound or stereotactic localizations, possible transection of the wire during surgery may make removal of the area more challenging. Because the presence of the lesion in the specimen cannot be confirmed by x-ray, postexcisional MRI may be performed 1 month after the procedure to confirm removal of the lesion.

2010 Update of the TNM Staging System

The seventh edition of the American Joint Committee on Cancer staging system for breast cancer includes important changes in each of the TNM elements (Table 91-8).¹⁵⁶ Changes in the tumor (T) classification include: (a) an attempt to estimate size of DCIS and LCIS; (b) use of microscopic measurement if a small invasive cancer was entirely submitted in one paraffin block; use of the clinical measurement to determine T at presentation in patients treated with neoadjuvant treatment, and use of gross and microscopic histology to describe the T size after neoadjuvant treatment; (c) the description of simultaneous ipsilateral carcinomas whether in the same quadrant or in separate quadrants; and (d) the classification of Paget disease based on the underlying cancer (e.g., Tis, T1, …) or as Tis (Paget) if no underlying cancer. Changes in nodes (N) classification include: (a) the use of the new stage IB to describe small T1 tumors with node micrometastases (N1mi); (b) the description of isolated tumor cell (ITC) clusters or single cells; (3) the deletion of the modifier (sn) modifier if six or more sentinel nodes are identified on gross pathological examination; and (4) the expansion of ITCs clusters to include small clusters of cells up to 0.2 mm, or if nonconfluent or nearly confluent, clusters of up to 200 cells in a single histologic lymph node cross section. As for metastases (M), the new M0(i+) category is defined by DTCs in bone marrow, as CTCs, or if incidentally observed in other tissues as clusters of up to 0.2 mm. Of note, however, M0(i+) patients with no clinically or radiographically detectable metastases continue to have their staging defined by T and N.