In this chapter, we review the imaging characteristics of invasive breast cancer, focusing primarily on their various morphologic and kinetic findings on MRI. Additionally, we discuss the heterogeneous nature not only of the imaging findings, but also the clinical, histologic and molecular characteristics of invasive breast cancer as they apply to patient outcome. Microarray-based, high-throughput gene expression profiling methods have been applied to the investigation of breast cancer. These efforts are aimed towards improved understanding of the molecular basis of tumor biological features such as histologic grade, metastatic propensity, and identification of tumor genetic signatures that are associated with prognosis and therapeutic response. We show the imaging characteristics of invasive breast cancer and characterize their findings on MRI according to the BI-RADS Atlas- MRI. The most common histologic type of invasive breast cancer is now classified as Invasive Breast Carcinoma of No Special Type (IBC-NST), a change from the prior edition of the WHO Classification of Tumors of the Breast where it was designated Invasive Ductal Carcinoma Not Otherwise Specified (IDC-NOS). We review the MR imaging findings for invasive cancers grouped by both histology and their molecular signatures. These findings are shown in the 86 case examples that accompany this chapter.
As future research unfolds, it is possible that the imaging phenotype of invasive breast cancer, particularly as applied to DCE-MRI with associated advanced computer analysis, may well provide even more independent prognostic and predictive markers, complementing existing biomarkers, and thus improving patient care.
MR Imaging of invasive cancer – molecular signatures of breast cancer – invasive carcinoma no special type (NST) – invasive lobular carcinoma – histologic subtypes of invasive cancer – tubular subtype – mucinous subtype – medullary subtype – papillary subtype – luminal A subtype – luminal B subtype – HER2/neu-enriched subtype – triple-negative and basal-like sub-type – inflammatory breast cancer (IBC) – uncommon tumor subtypes
8 Image Interpretation: Invasive Cancer
In this chapter, we will review the magnetic resonance imaging (MRI) characteristics of the various subtypes of invasive cancer and correlate the MR lesion phenotype and kinetic characteristics with histology and established biomarkers.
Invasive breast cancer is a heterogeneous disease often harboring various histologic subtypes within a main tumor mass (intratumoral heterogeneity) or within separate tumor satellite lesions (intertumoral heterogeneity). Histologic heterogeneity may be found even within the morphologic types of invasive cancer, such as ductal, lobular, and other less common subtypes.1For many decades, breast cancer treatment has been primarily guided by the histologic classification of cancers using tumor grade, morphology, and the tumor–node–metastasis (TNM) staging method, based on cancer size, nodal status, and the presence or absence of distant metastases. The TNM classification predates the era of modern imaging and identifies the size of the largest invasive tumor focus as the main descriptive factor, discounting tumor multifocality in the overall assessment.2Tumors vary in both grade and histopathology. Grade is based on cellular differentiation: the higher the grade, the more “poorly differentiated” the cancer cells. It is well known that imaging phenotypes based on morphologic analysis of invasive cancers seen on mammography, ultrasound, and MRI can provide prognostic information.
In recent years, molecular subtyping of breast cancer has complemented the standard histologic classification to encompass treatment stratification based on prognostic indicators.2,3Microarray-based, high-throughput gene expression profiling methods have been applied to the investigation of breast cancer. These efforts are aimed toward improved understanding of the molecular basis of tumor biological features such as histologic grade, metastatic propensity, and identification of tumor genetic signatures that are associated with prognosis and therapeutic response. Although molecular subtyping for every patient with breast cancer could potentially enable clinicians to guide therapy more effectively, the costs of genetic analysis are high because of the need for technical expertise, and specialized equipment for individual sample processing. Molecular subtype analysis is not currently feasible for all patients and other markers have been established. Immunohistochemical (IHC) analyses are used to substitute for genetic profiling and are widely used in clinical practice. IHC markers of breast cancer include estrogen (ER) and progesterone (PR) receptors, and HER-2/neu overexpression. They have the advantage of rapid testing at a lower cost than formal genetic analysis; however, their results have been less robustly predictive of patient outcomes. A report by the IMPAKT 2012 Working Group on molecular subclasses of breast cancer showed that concordance for molecular subtype classification, between IHC and formal genetic analysis, ranges from 41 to 100% within varying subtypes.4Despite increased costs, tumor stratification based on tumor biology and gene expression profiles derived from DNA microarrays can now be made available to patients and are slowly being integrated into clinical practice.
The original influential studies published by Perou et al5,6defined a molecular subtype classification of breast cancer into four main categories: luminal A, luminal B, HER-2/neu positive, and basal-like. These groups are distinguished by distinct patterns of genomic additions and deletions, providing prognostic information and influencing systemic treatment decisions. The main distinction between these molecular subtypes is at the ER level and secondarily on the level of HER-2/neu. Of these molecular subtypes, two are ER positive, the luminal A (human epidermal growth factor receptor 2 negative [HER2–]) and luminal B (HER2+) groups, and two are ER negative, the basal (HER2–) and ERBB2 (HER2+) groups. These transcriptome-based subtypes
Transcriptome is the set of all messenger RNA molecules in one cell or a population of cells.
correlate well with other histological and clinical features; luminal A and B lesions generally are of a lower grade with a more favorable prognosis when compared to basal and ERBB2 subtypes, which are often higher grade with a worse prognosis. Additionally, rapidly growing tumors overexpress genes such as Ki-67, a marker of cellular proliferation known to correlate with increased mitotic indices at histopathology. Prognostic information, obtained from the tumor of each individual patient using tumor genetic signatures, or IHC markers, can guide therapy and provide selected treatment for each individual patient. Why are imaging biomarkers important? Imaging is quantitative, can sample the entire cancer, measure intratumoral heterogeneity, and complement other predictive and prognostic indicators. Serial imaging is noninvasive and can monitor tumor response during therapy.
8.3 MRI of Invasive Cancer
The vast majority of invasive cancers present as masses, noting that only a small subset of invasive cancers exhibit non–mass enhancement as the primary imaging finding. The most common histologic type of invasive breast cancer is now classified as invasive breast carcinoma of no special type (IBC-NST), changed in the fourth edition of the World Health Organization (WHO) Classification of tumors of the breast from invasive ductal carcinoma not otherwise specified (IDC-NOS).1The term “ductal” is no longer included in the new definition, the rationale being that the term “ductal” conveys unproven histogenetic assumptions, and that IDC-NOS does not comprise a uniform group of cancers. The IBC-NST lesions constitute about 75 to 80% of all breast cancers; invasive lobular carcinomas (ILCs) contribute another 10 to 15%; and special subtypes, tubular, medullary, papillary, and mucinous constitute the majority of the remaining malignant lesions. Invasive tumors may be further defined by their distribution and extent and are categorized as unifocal (Fig. 8‑1; Fig. 8‑2), multifocal/multicentric (Fig. 8‑3; Fig. 8‑4; Fig. 8‑5; Fig. 8‑6), and diffuse (Fig. 8‑7).
Multifocal cancer is usually defined as disease confined to one quadrant, whereas multicentric disease may involve more than one quadrant. This distinction may be difficult to apply in clinical practice, whereas confluent tumor involvement of most of the breast tissue is easily defined as diffuse disease. Inflammatory carcinoma is distinguished by its clinical presentation rather than by a distinct histologic subtype.
8.3.1 BI-RADS Atlas—MRI
The BI-RADS Atlas7defines a mass as a three-dimensional space-occupying structure, with a convex-outward contour and is further categorized by shape, margin, and internal enhancement characteristics. High spatial resolution techniques allow optimal morphologic analysis with superior definition of both shape and margins. A mass may or may not displace or otherwise affect the surrounding normal breast tissue, and may be associated with other enhancing or nonenhancing findings (Table 8‑1; Table 8‑2).
Table 8.1 Mass descriptors
Internal enhancement characteristics
Dark internal septations
Source: ACR BI-RADS Atlas MRI: Second edition.
Table 8.2 Nonenhancing findings
Ductal high-contrast signal on T1w
Posttherapy skin thickening and trabecular thickening
Postoperative collections (seroma, hematoma)
Signal void from foreign bodies, clips, etc.
Source: ACR BI-RADS Atlas MRI: Second edition.
An oval mass is elliptical or egg-shaped and includes lobulation, with two or three undulations (Fig. 8‑8; Fig. 8‑9; Fig. 8‑10). A round mass is spherical, ball-shaped, circular, or globular in shape (Fig. 8‑11) and an irregular lesion’s shape is neither round nor oval and usually implies a suspicious finding (Fig. 8‑12).
The margin is the edge or border of a lesion. The descriptors of margin, in addition to the descriptors of shape, are important predictors of benignity or malignancy. A circumscribed margin (changed from smooth in the prior lexicon) is sharply demarcated with an abrupt transition between the lesion and surrounding tissue (Fig. 8‑13). The entire margin must be well defined in its entirety to be qualified as “circumscribed.” A not circumscribed margin may be categorized as irregular, edges that are either uneven or jagged (Fig. 8‑14), or spiculated, lines radiating from a mass implying a suspicious finding (Fig. 8‑15; Fig. 8‑16; Fig. 8‑17; Fig. 8‑18). When reporting a mass with both irregular shape and margin, the MRI report should indicate that there is “a mass of irregular shape and margin.” In general, circumscribed masses are indicative of benign lesions in all imaging methods in contradistinction to noncircumscribed masses, which are suspicious for carcinoma. Margin analysis is highly dependent on optimal spatial and temporal resolution for accurate diagnosis at MRI.
Internal Enhancement Characteristics
Internal enhancement describes the enhancement pattern within an abnormally enhancing structure, and is an important reflection of lesion biology.8Enhancement patterns are characterized as homogeneous, a confluent uniform enhancement of a mass, and heterogeneous, a nonuniform enhancement of a mass with variable signal intensity. Homogeneous enhancement is suggestive of a benign process; however, some small cancers can exhibit homogeneous enhancement, and careful margin assessment with high spatial and temporal resolution is essential for an accurate diagnosis of benignity (Fig. 8‑19). Heterogeneous enhancement is generally more characteristic of malignant lesions (Fig. 8‑20).8Rim enhancement is more pronounced at the periphery of a mass, and is commonly seen in high-grade malignancies. Cysts can become inflamed and enhance peripherally, but are usually bright on T2-weighted (T2w) sequences confirming internal fluid, unless they are very small or contain proteinaceous material (Fig. 8‑21). Fat necrosis (oil cysts) and normal postoperative seromata may exhibit smooth peripheral enhancement. Rim enhancement of a solid mass is a suspicious finding (Fig. 8‑22; Fig. 8‑23). The MRI features with the highest predictive value for malignancy are lesion mass types with irregular shape, irregular or spiculated margins, and marked enhancement.9Dark internal septations are seen as nonenhancing lines within an enhancing mass. These dark septations are seen more frequently at higher spatial resolution (3.0 Tesla) on either T2w precontrast or T1-weighted (T1w) postcontrast series and are suggestive of a fibroadenoma if other morphologic and kinetic characteristics support benignity (Fig. 8‑24; Fig. 8‑25). Myxoid fibroadenomata are highly cellular with moderate internal mucoid material exhibiting hyperintense signal on T2 in some cases. These circumscribed masses usually show dark internal septations and heterogeneous enhancement on T1w series (Fig. 8‑26, Fig. 8‑27). Fibroadenomata are frequently seen on breast MR studies, often occult at mammography, and must be confidently recognized as benign lesions whenever possible. Nonenhancing masses with benign morphology are benign.
These masses are generally benign and include lesions such as lipomata, fibroadenolipomata (hamartoma) (Fig. 8‑28, Fig. 8‑29), and lymph nodes (Fig. 8‑30; Fig. 8‑31). The key to accurate diagnosis is identification of fat signal within the lesion on a non–fat-suppressed high-resolution T2w or T1w series. Fat necrosis may present as a rim-enhancing mass exhibiting varying enhancement (Fig. 8‑32). History of prior trauma or surgery may explain the enhancing finding, and verified central fat content on non–fat-suppressed images may confirm the diagnosis (Table 8‑3).
Table 8.3 Fat-containing lesions
Lymph nodes: normal or abnormal
Postoperative seroma/hematoma with fat
Source: ACR BI-RADS Atlas MRI: Second edition.
T2w Precontrast Mass Findings
High-resolution T2w sequences provide important information and, when evaluating breast masses, can improve specificity of diagnosis. Comparing the morphology and signal intensity on T2w and T1w images can be helpful. High T2w signal may signify extensive necrosis seen most frequently in high-grade lesions, or in mucinous/loose myxoid stroma seen in mucinous carcinoma. Cystic or microcystic tumor components are rare and can be found most frequently in papillary cancers. Intratumoral fat-containing tissue usually indicates a benign diagnosis. The presence of peritumoral, stromal, subcutaneous, or prepectoral edema augurs an aggressive malignancy type.
8.4 Kinetic Curve Assessment
Kinetic analysis of suspect lesions should be assessed after complete evaluation of the MRI findings. The first postcontrast series is generally selected for kinetic analysis because abnormal tissue is usually most intense and distinct from normal background parenchymal enhancement at this time point. Kinetic information is typically expressed as a time intensity curve (TIC), plotting the signal intensity of the most suspicious enhancement finding on a pixel-by-pixel basis, depicting the enhancement rate over time. The TIC can be manually calculated by placing a region of interest (ROI) of at least three pixels within a lesion or can be automatically created by CAD systems (dedicated software generating lesion color maps and TIC graphs). The TIC depicts the initial phase of enhancement within the first 2 minutes following injection or until peak enhancement is reached, and the delayed phase of enhancement depicts the TIC after 2 minutes or after the peak enhancement is reached and is used to describe the curve shape. Malignancies generally enhance rapidly in the initial phase of contrast enhancement with contrast washout in the delayed phase. Evaluation of both morphologic and kinetic data is essential for diagnosis.
8.4.1 Initial Phase
Initial enhancement is determined by comparing signal intensity in the precontrast image to the first postcontrast image acquired. An intensity increase of less than 50% is classified as “slow,” 50 to 100% is classified as “medium,” and greater than 100% enhancement is classified as “fast.”
8.4.2 Delayed Phase
Delayed enhancement is divided into three main categories:
Persistent curves (type 1) are defined as showing ≥10% of the initial enhancement with continuously increasing enhancement throughout the delayed phase.
Plateau curves (type 2) are equal to the initial enhancement and remain constant in their signal intensity once peak enhancement is reached, usually after 2 to 3 minutes.
Washout curves (type 3) are defined as showing ≤10% of the initial enhancement with continuous decreasing signal intensity after peak enhancement is reached (Fig. 8‑33).
Kinetic Curve Assessment
In general, benign lesions exhibit persistent curves, and malignant lesions exhibit washout curves, although there is considerable overlap between the kinetic curves that depict malignant and benign lesions. Diagnosis should only be made after consideration of both the morphologic and kinetic features of an enhancing lesion. Kinetic features reflect underlying lesion biology and the efficacy of advanced computerized analytic methods, using both morphologic and kinetic data, will be discussed in Chapters 12 and 13.
8.5 Histologic Subtypes of Invasive Cancer
As previously noted, the most common type of invasive breast carcinoma (75–80%) is classified as IBC-NST. This category incorporates all breast adenocarcinomas that fail to exhibit specific histologic characteristics that would warrant classification as one of the special types. The WHO classification recognizes the existence of at least 17 distinct histological special types.1Invasive malignancies exhibit distinctive morphologic, kinetic, and molecular characteristics that are reflected on mammography, ultrasound, and MRI. The imaging characteristics of the most common special types of invasive cancer will be reviewed.
8.5.1 Invasive Lobular Carcinoma
Invasive lobular carcinoma accounts for 10 to 15% of all invasive cancers, is known to be difficult to diagnose both on clinical examination and on mammography, and is usually larger at initial presentation, more frequently exhibiting multifocal/multicentricity than IDC. Palpable thickening and skin or nipple retraction may be the most common clinical findings rather than detection of a discrete breast mass. The growth pattern of uniform, small, round tumor cells, typically infiltrating in a single file without mass formation, may limit conspicuity at mammography. The reported sensitivity of mammography ranges from 34 to 92%10; however, even when detected, size underestimation is common. Lack of desmoplastic reaction and absence of associated ductal carcinoma in situ (DCIS) with visible microcalcification also contribute to difficulties in lesion perception. ILC is diagnosed, in part, by absence of e-cadherin expression, a gene involved in cell–cell cohesion, thought to account for the singular growth pattern.
The most common presentation on mammography is that of a spiculated or ill-defined mass; however, findings of asymmetry and architectural distortion are more commonly found in ILC lesions than in IDC lesions. In some cases, a sheet-like infiltrating pattern of the tumor may result in decreased natural breast elasticity in the affected breast, limiting breast compression at mammography so that the breast appears to be smaller than the contralateral breast. This finding is known as the “shrinking breast” sign. The affected breast may appear to be of normal size, however, on clinical examination “thickening” of the breast may be evident.10,11,12Ultrasound has a higher sensitivity for ILC than mammography (68–98%) and findings usually present as an irregular, hypoechoic mass with indistinct or spiculated margins and acoustic shadowing.13
The heterogeneous nature of ILC is well established, and histologic subtypes of ILC lesions have been described. In 1982, Dixon and colleagues categorized 103 ILC lesions as classical, solid variant, alveolar variant, or mixed histologic subtypes, the classic and mixed types being the most common.14The classic type of ILC is described as a single filing growth pattern, with “peri-parenchymal” distribution and diffuse multifocal invasion. The solid variant consists of a “sheet-like” pattern or irregular-shape nests of cells, whereas the alveolar type is described as globular aggregates of 20 or more cells. These three subgroups exhibit small noncohesive regular cells with round or oval nuclei, whereas the fourth mixed type demonstrates cohesive cells with nuclear pleomorphism. Recently, subclassification has been reported using histological features and IHC15; however, no clear differences in long-term patient outcome between the different histologic subtypes was found in either study.14,15
Two studies have reported on the mammographic imaging appearance of ILC subtypes and histopathology.16,17In 2014, Tabar and colleagues16reported on 428 consecutive cases of ILC, diagnosed in the screening era, from 1996 through 2010, and compared patient outcome of these cases with a cohort of ILC cases diagnosed and treated in the prescreening era 25 to 30 years earlier.7A classification of the mammographic features of ILC lesions was made with an approximate correlation to the earlier histologic classification of Dixon and colleagues14(Table 8‑4). ILC subtypes vary in their imaging characteristics and lesion size at diagnosis. The alveolar subtype is often extremely difficult to detect at mammography even though it is often palpable and multifocal, because it consists of many individual, tiny, scattered 2- to 3-mm cancer foci, often involving an entire quadrant but without a discrete tumor mass. Tumor sizes differed significantly in this study according to their mammographic subtypes (p < 0.001). Large tumor sizes (≥30 mm) were overwhelmingly more likely to be found in architectural distortion cases (76%) than in spiculated masses (21%), round masses (25%), and equivocal asymmetric densities (34%). Patient outcome was shown to be related to mammographic appearance when long-term survival of women with ILC lesions was compared in the two groups (1960–1970 vs. 1996–2010). Outcome for women with spiculated and circular/oval-shape lesions diagnosed during the screening era was dramatically improved; however, no change in survival was found in women with the classical architectural distortion subtype, despite the advent of screening mammography in the latter group and the introduction of new therapeutic regimens.
Table 8.4 Mammographic features of ILC subtypes
Solid variant pattern
Round- or oval-shape mass
Solitary or multifocal spiculated mass
Abbreviation: ILC, invasive lobular carcinoma.
Magnetic Resonance Imaging
The MRI findings of ILC reflect the morphologic findings seen on mammography and ultrasound; however, determination of disease extent is more accurate at MRI, and ILC lesions, found to be occult on mammography and ultrasound, are almost always visible on MRI. An irregular or spiculated-shape mass, with heterogenous internal enhancement, is a common presentation of ILC, although focal and “sheet-like” regional nonmass enhancement may also be found.12,18,19,20ILC rarely presents as a round mass, an important distinction compared to IDC lesions, which may often present as round masses with not circumscribed margins, and are generally grade 2 lesions. Mann and colleagues reported that a round-shape mass was identified in only 1 of 143 ILC lesions.18ILC has been described as a “stranding” pattern of tumor enhancement, associated with multiple, small, enhancing foci, by some investigators, possibly reflecting the single-file pattern of tumor growth.14,21Reports on the kinetic features of ILC generally conclude that in the initial phase of enhancement, ILC enhances more slowly than IDC, although peak enhancement measures may be similar. In the delayed phase of enhancement, ILC exhibits washout curves less frequently and a larger proportion of the tumor shows persistent delayed-phase curves than are seen in IDC.22
Multimodal images of the four main presentations of ILC are shown.
The classical subgroup of ILC: architectural distortion (Fig. 8‑34).
Additional ipsilateral malignant lesions detected only on MRI were found in 32% of patients, in a study of preoperative assessment of ILC extent, resulting in changed management in 28% of cases.18Another study of preoperative assessment of ILC showed management change in 49% women, 40% needing more extensive surgery and 9% needing less extensive surgery (Fig. 8‑43, Fig. 8‑44).23Patients with an ILC diagnosis are more likely to have positive surgical margins after lumpectomy than patients with IDC lesions. Reduction in reexcision rates of ILC lesions by use of preoperative MRI has been shown in a study of 267 patients undergoing breast conserving surgery (BCS). Investigators reported a 9% reexcision rate in the MRI group compared to a 27% reexcision rate in the group without preoperative MRI.24MRI has been shown to be an essential imaging method in the preoperative assessment of women with a new diagnosis of ILC.
8.5.2 Invasive Duct Carcinoma
Tubular carcinoma is an uncommon, low-grade subtype of invasive carcinoma accounting for fewer than 2% of all invasive breast cancers, its incidence increasing with the use of screening mammography. These cancers are generally small and node negative with a favorable prognosis, the overall 10-year disease-free survival rate for tubular cancer being greater than 90%. Tubular carcinoma contains a central fibroelastotic core that entraps glandular proliferating elements, exhibiting round, ovoid, or angulated tubules formed by a single layer of small, regular cells with little nuclear pleomorphism. A tubular component greater than 90% is a requirement for a histologic diagnosis of tubular carcinoma.25,26
The hallmark finding, on all modalities, is that of spiculation, which corresponds to reactive stroma surrounding the tumoral mass, seen on ultrasound as an echogenic “halo.” Small tubular cancers usually present on MRI as spiculated or irregular masses, with heterogeneous enhancement and persistent kinetics (Fig. 8‑45; Fig. 8‑46, Fig. 8‑47). Minimal enhancement or initial enhancement rates less than 100% are commonly found. This enhancement pattern can be explained by the slow progression of the contrast material within the dense fibrous and elastotic stromal components. The morphologic features of radial scar (complex sclerosing lesion) often mimic those of tubular carcinoma and their kinetic behavior is often variable. Although visible on MRI, the distinction between these two lesions is usually not possible if lesion enhancement is present, and pathologic examination of the surgically excised tumor is generally needed.