Mammography screening alters the presentation of breast cancers from mainly palpable to mainly non-palpable. Randomized controlled mammography screening trials have convincingly demonstrated the following:

The randomized controlled mammography screening trials have provided the opportunity to study the mechanism by which earlier diagnosis and treatment affect the outcome of breast cancer. In these trials one group was randomized to receive an invitation to screening, but the other, randomly selected group of women (control group) was not invited. The breast cancers in women invited to screening were diagnosed on average at an earlier phase than the self-detected tumors in the control group. Screening has a significant impact on all three first-generation prognostic factors: tumor size, axillary node status, and histologic malignancy grade. In a high-quality service screening program 15–20 % of the cancers will be in situ and more than 50 % of the invasive carcinomas will be <15 mm in diameter. Early detection also results in significantly fewer cases with axillary lymph node metastases and also prevents worsening of the malignancy grade in a certain percentage of the tumors. Since two components of the TNM classification, tumor size and node status, will improve significantly in women invited to screening, the incidence of Stage II and more advanced cancers will decrease in this same group of women (see Fig. 2.3).

There is parallelism between the incidence of advanced cancers and the breast cancer-specific mortality rate in any given population, since most breast cancer deaths occur in women whose tumor was at an advanced stage at the time of detection [10, 12, 30–32]. Thus decreasing the incidence rate of advanced tumors through screening will result in a corresponding decrease in breast cancer mortality in this same group of women. In the Swedish Two-County Trial the advanced cancer rate began to fall starting from year four and onwards in women invited to screening, as did the breast cancer death rate. Both of these declines were a consequence of early detection and treatment in an earlier phase (see Fig. 2.4).

The effect of tumor size on long-term survival (28 years) in the Swedish Two-County Trial is presented in Fig. 2.5. The beneficial impact of screening is reflected in the excellent long-term survival of women with in situ and 1–14 mm invasive breast cancer. The 28-year survival according to axillary node status and distant metastases for all tumor sizes demonstrates the profound prognostic impact of these parameters (see Fig. 2.6). These survival rates are from the era prior to the widespread use of chemotherapy for primary breast carcinoma; none of the women with <20 mm node-negative tumors received chemotherapy in the Swedish Two-County Trial (1977–1985). Women with lymph node metastases had significantly poorer survival than those without lymph node metastases. The cumulative survival of women aged 40–69 years according to the histologic grade of the invasive cancers is shown in Fig. 2.7.

Many breast tumors display intratumor heterogeneity, containing two or more histologic types and phenotypes (see Fig. 2.8a–d) [33–35]. Early detection through screening prevents many small, well- or moderately differentiated tumors from developing into more poorly differentiated, larger tumors. The evidence that the histologic malignancy grade worsens as the breast cancer progresses comes from the analysis of both clinical [36] and screening data [37]. The clinical research of Tubiana et al. demonstrated that “during their growth tumors progress towards higher grades” [36]. Duffy et al. used data from a randomized controlled mammography screening trial to perform a more precise measurement of progression by comparing the tumor characteristics in the control group, where the tumors were allowed to grow until clinically detectable, with the tumor characteristics in the group of women invited to mammography screening, where screening aimed at arresting tumor growth [37]. This comparison required the removal of the prevalence screen tumors from both groups in order to eliminate length bias. These two sets of tumors were then equivalent in all aspects except that the tumors in the group invited to screening were diagnosed, on average, earlier. Comparison of tumor size, node status, histologic malignancy grade, and detection mode showed that the proportion of cancers with positive nodes and a higher malignancy grade increased with increasing tumor size [3, 37]. There were also significantly fewer node-positive and poorly differentiated cancers among women invited to screening. There could be two competing explanations for these results: (1) The malignancy grade remains unchanged as the tumor grows, and screening has mostly detected well- and moderately differentiated tumors (length bias sampling). (2) The malignancy grade tends to worsen as the tumor grows, and tumor progression does indeed occur. If this were to happen, one would see a deficit of poorly differentiated tumors in the incident cancers of a group of women invited to screening compared with an uninvited group. When Duffy et al. eliminated the length bias cases from both groups, i.e., the prevalent screen, and could thus compare the incident cancers in those invited and those not invited to screening, they observed that the tumors were significantly smaller and there was a significant deficit of poorly differentiated tumors in the invited group. This demonstrated that screening prevented the deterioration of the malignancy grade of some of the tumors. In summary, tumor progression (worsening of the malignancy potential through the process of dedifferentiation) has been shown to occur in both clinical and screening studies [36, 37].

Detailed analysis of the breast cancer cases from women invited and not invited to a randomized controlled trial demonstrated that the rate of poorly differentiated breast cancer increases in all age subgroups with increasing tumor size, but in premenopausal women this process of dedifferentiation occurs more rapidly, earlier in the preclinical detectable phase, and to a greater extent than in postmenopausal women. All these factors in combination make it necessary that women are invited to screening at a frequency which takes into account the varying tumor growth rates according to different histologic tumor types and women’s age [29, 38]. Reversion to less frequent screening, as recommended by the US Public Services Task Force (USPSTF), would tend to increase the number of advanced (more frequently poorly differentiated and node positive) cancers at the time of treatment and increase fatality from breast cancer [39].

The recent publications by Esserman et al. maintain that screening does not decrease the incidence of advanced breast cancers [27, 28]. This is contrary to the published evidence [10, 12, 30–32, 40]. The claim of Esserman et al. that “tumor biology does not change over time” [27] reflects unfamiliarity with appropriate statistical analysis of clinical [36] and screening trial data [37] and fails to account for certain fundamental observations in breast tumor biology, including the consequences of intratumor heterogeneity [35].

Our primary goal is to reduce mortality from breast cancer. Mammography screening and the associated improvements in diagnosis and therapy have enabled us to reduce the breast cancer mortality in women attending screening regularly by 40–50 % [19]. Despite this accomplishment, women are still dying from breast cancer. Investigation into the characteristics of the cancers that are still causing breast cancer death requires assessing the extent of the disease as a measure of the tumor burden. Two comprehensive whole-breast histologic studies examined the unifocal, multifocal, and diffusely infiltrating nature of breast cancer [41, 42]. The term multifocality includes (a) multiple in situ cancer foci without invasion, (b) a solitary invasive carcinoma associated with multiple in situ foci, and (c) multiple invasive breast cancer foci with or without associated in situ cancer. The invasive cancers with or without an associated in situ component are responsible for breast cancer death; the relative frequency of unifocal/multifocal/diffusely infiltrating invasive breast cancers is approximately 68/27/5 % [109]. In which of these groups is breast cancer most fatal? The fatality ratio of unifocal breast cancers (with or without associated in situ foci) is 9.1 % and most (74 %) of these fatal cancers were >2.0 cm in size. In the era of mammography screening enhanced by the use of multimodality breast imaging, unifocal tumors can be detected and successfully removed before they reach the size of 2.0 cm. The fatality ratio in multifocal and diffusely infiltrating invasive breast cancers is 20 and 26 %, respectively, considering all sizes of tumors [43]. Multifocality is an important, independent negative prognostic factor (see Fig. 2.9) and its harmful effect becomes more significant with increasing tumor size. Weisenbacher and coworkers have arrived at the same conclusion [44].

The highly significant size-related survival difference also applies to multifocal breast cancers, suggesting that a combination of imaging methods that enables detection of multifocal cancers with a lower tumor burden (when the largest tumor focus is <15 mm) will result in a lower fatality rate (see Fig. 2.10). The multimodality approach (mammography, automated breast ultrasound, and especially breast MRI) will detect multifocal cancers having a lower tumor burden, and will correspondingly lower the fatality rate. This emphasizes the importance of using breast MRI to determine the presence and extent of multifocal disease. The use of breast MRI in multifocal and diffusely infiltrating invasive breast cancers is invaluable in describing the true extent of the disease. This is an important part of treatment planning to prevent incomplete resection of breast cancer at primary surgery. Incomplete resection of invasive cancer foci is associated with a poor outcome: “For patients who underwent second surgery, the finding of a residual invasive carcinoma was associated with increased risk for distant recurrence (22.8 vs. 6.6 %; HR 3.5; 95 % confidence interval, 1.8–7.4; P < .0001).” These same authors concluded, “there is a need to improve techniques for the presurgical and/or intraoperative determination of margins” [45].

Modern, high-resolution breast MRI has the capability of describing the true extent of the disease in the vast majority of cases, far exceeding that of earlier MRI technology, on which most currently available reviews are based. The COMICE trial, which used 2.4/4.0 mm slice thickness (as opposed to the current practice of 0.7–1.0 mm), was a multicenter trial in which 45 centers supplied an average of only 18 cases each during the 5-year accrual period starting in 2002. This study’s failure to detect an impact of preoperative MRI upon reoperation rate may reflect the outdated technology and the extremely low average rate of patient accrual per site, reflecting limited experience in breast MRI interpretation. High-resolution breast MRI was practically nonexistent prior to 2007. For these reasons the COMICE trial results [46], the meta-analysis by Houssami et al. [47], and other earlier studies may have lost their relevance to current breast MRI practice.

The reliance upon local recurrence as a measure of success or failure of breast cancer treatment is subject to serious limitations. Fatality often occurs without local recurrence and the term “local recurrence” as used in the literature does not discriminate among recurrences in unifocal, multifocal, and diffusely infiltrating breast cancers. One classification system uses a cutoff point of 4.0 cm to separate “extensive” from “non-extensive” breast cancer [42]. Using this arbitrary cutoff point, “A disease extent ≥4 cm was shown to be an independent marker for local recurrence; the cumulative 10-year local relapse rate for the group with a disease extent ≥4 cm was 20.5 %, and for the rest 6.7 % (p value = 0.003)” [48].

The seriousness of multifocal and diffusely infiltrating breast cancers has not been generally appreciated for two main reasons. First, the current TNM classification system does not account for multifocality, using only the size of the largest invasive focus as the major descriptive factor. This seriously underestimates the actual tumor burden of multifocal tumors. We have proposed a quantitative evaluation of tumor burden in terms of total tumor volume and surface area [43].

Second, the current practice of histopathology of breast specimens has serious limitations; “conventional techniques may not reflect the extent of neoplasia when the neoplasia is impalpable or grossly indistinct as in the case of dense breast tissue.” Additionally, “complete specimen examination is rarely performed in clinical practice.” “In a typical 8-cm diameter lumpectomy specimen, assuming four conventional pathology margin sections are removed in a single plane, only 16 % of the circumference is examined microscopically” [49] (see Fig. 2.11). “People blame MRI instead of the limitations of conventional pathology and a failure of small section pathology to correlate with MRI and mammography.” (Lee Tucker, M.D., F.A.P.C., personal communication 2012).

We recommend that large-section histopathology should be standard for all breast cancer surgical specimens, as it also provides better correlation with breast imaging (see Fig. 2.12a–m).

Breast cancer originates either from the epithelial cells lining the acini within the terminal ductal lobular unit (TDLU) or from the cells lining the milk ducts (see Fig. 2.13). The majority of breast cancers originate from the TDLUs, not from the ducts. Figure 2.14 shows the relative distribution of the crushed stone-like and powdery microcalcifications on the mammogram, both of which are the mammographic presentations of in situ tumor growth which arise from and are localized within the TDLUs. Despite the fact that these in situ tumors arise within the lobules and not from the ducts, they are paradoxically termed “ductal” carcinoma in situ. When this population of cancer cells invades the surrounding breast tissue, forming a stellate or circular/oval-shaped tumor mass, the invasive tumor is also erroneously called “ductal.”

In situ breast cancers are usually detected at mammography. There are more than ten distinctly different mammographic presentations of in situ cancer subtypes (see Figs. 2.15a –2.15c and 2.16a–x), but current terminology bundles them all under the same name: DCIS. This simplification unfortunately leads to misunderstanding and confusion. Additionally, the term DCIS is a misnomer, since the vast majority of in situ carcinomas do not arise from the milk ducts and are not situated within these ducts.

Breast cancers actually arising within the major milk ducts have a histopathologic appearance (see Fig. 2.17a, b) very similar to that of metastatic prostate cancer (Fig. 2.17c–e) and metastases of breast cancer to the axillary lymph node(s) (Fig. 2.18a, b). Although the histopathologic appearance shown in Figs. 2.17c–e and 2.18a, b will be termed by pathologists as invasive cancer, i.e., when found in the prostate or in the axillary lymph node(s), the similar histopathologic appearance is termed “DCIS” when found in the breast. The unpredictable clinical course and also the occasional fatal outcome of these cases indicate that, contrary to its name “ductal carcinoma in situ” of the breast, the special breast cancer subtype originating from the major ducts may behave as an invasive cancer and can prove fatal (see Figs. 2.19, 2.20, and 2.21) [50].

Taking the logical and consistent nomenclature that is used to describe prostate cancer and using it to describe breast cancer as well can resolve these terminological inconsistencies and the resulting confusion. Our proposed terminology emphasizes the site of origin of the cancer: acinar adenocarcinoma of the prostate (AAP) would correspond to acinar adenocarcinoma of the breast (AAB), in which the cancer originates from the TDLU. Similarly, ductal adenocarcinoma of the prostate (DAP) would correspond to ductal adenocarcinoma of the breast (DAB), in which the breast cancer originates from the major milk ducts. The striking difference between the long-term outcome of breast cancers of similar size originating from the TDLUs (AAB) and the cancers originating from the major ducts (DAB) justifies the radical change in terminology (Figs.2.19 and 2.20).