A distinguishing feature of these imaging modalities is that they recover actual tissue property distributions that provide functional information about the tissue being interrogated. A crucial aspect of this multi-modality project is the investigation of the relationship between the imaged properties and actual tissue pathology. Toward this end, the following section describes initial pathology studies that have been conducted in parallel with imaging modality development. A general overview of breast physiology is followed by a more in-depth discussion of morphologic criteria that may correlate with contrast mechanisms operative in alternative imaging modalities. The discussion is somewhat technical, so non-medical readers may wish to skip directly to Section 4, "Unifying the Four Modalities."
The adult female breast is a large, modified sebaceous gland that consists mostly of fat, fibrous septa, and glandular structures. The weight range for a "normal," mature female breast is 30 grams to over 500 grams, depending on the woman's body habitus. The breast typically comprises 15 to 25 lobes that are divided into multiple lobules, each containing 10-100 terminal milk-secreting alveoli. Numerous tiny milk-transporting ductules combine to form a single lactiferous duct that exits each lobule. About 15 to 25 such ducts converge at the nipple. The composition of the breast varies from individual to individual and with age and other factors. Pregnancy, lactation, menstruation, and menopause all introduce characteristic changes in breast physiology. For example, in postmenopausal involution of the breast, the lobular and alveolar structures regress and the vascularity of the intervening connective tissue is reduced. Eventually only small, occasional islands of functional breast parenchyma remain, surrounded by dense, scarred connective tissue .
We have only a limited understanding of the biological and physiological bases of image contrast for malignancy and their relation to biological and molecular markers of cancer progression or regression that are predictive of therapeutic response and, ultimately, outcome. The interpretation and significance of variables imaged using alternative modalities will only be appreciated if the electromechanical properties being measured (e.g., electrical conductivity and permittivity, optical absorption and scattering, and mechanical elasticity and compressibility) are correlated with the biological characteristics of the tissue being imaged. Measures of tissue microvascular-ity such as mean vessel density and area may correlate with hemoglobin con centration and oxygen saturation, as indirectly measured by NIS. The ratio of functional breast parenchymal epithelium to surrounding dense connective tissue stroma (epithelium-to-stroma ratio, E:S) may correlate with tissue hardness, elasticity or compressibility as measured by MRE. Variable interfaces between tissue types are also likely to influence the electromagnetic properties associated with modalities such as EIS and MIS, which are sensitive to such morphologic attributes of the local cell population as volume fraction, membrane integrity, water content, and ionic concentrations. Unfortunately, biological correlates with tissue water content are difficult to evaluate since the tissue must be routinely processed (formalin-fixed, dehydrated, and paraffin-embedded) in order not to compromise the pathologist's ability to make a definitive tissue diagnosis.
A range of electromechanical and biological values for normal breast tissue must first be established to ensure a meaningful comparison to diseased breast tissue. To this end, we have completed a study employing a computer-aided program specifically developed to reproducibly assess microvascula-ture and tissue type interfaces in benign and malignant breast tissue.
The benign diagnostic categories comprised (1) breast tissue with normal histology, (2) fibrocystic disease, and (3) a common benign neoplasm (fi-broadenoma). The malignant neoplasm category comprised invasive carcinomas. Fourteen patients who underwent breast-reduction surgery with sampling from both breasts provided tissue with normal breast histology. Twenty-one patients (16 of whom also underwent breast reduction surgery with sampling from both breasts, 5 with unilateral biopsies) provided tissue with fibrocystic disease of variable severity (mild, moderate, or severe). Nineteen patients provided tissue with a benign fibroadenoma from one breast, each lesion classified according to the degree of stromal hyalinization or scarring in the tumor. Seventeen patients provided tissue with an invasive, malignant carcinoma from one breast .
Mean vessel density (MVD, percentage of each unit area that consists of transected vessels), mean vessel area (MVA, average cross-sectional area of an individual vessel), and vessel orientation (correlated with shape of observed cross-section) were the morphologic criteria chosen to assess tissue microvascularity. The criteria chosen to evaluate tissue hardness, elasticity, and compressibility were the epithelium-to-stroma ratio (E:S), the degree of severity of fibrocystic disease, the degree of stromal hyalinization or scarring in the benign neoplasms (fibroadenomas), the infiltrative patterns of the malignant neoplasms (carcinomas), and the type of tissue interfacing with the neoplasms (fatty, fibrofatty, fibrous, fibrocystic changes).
Computerized image-processing techniques can be used to select regions of interest in tissue samples for analysis. First, hematoxylin and eosin are term LinG - live, informative, Non-cost and Genuine !
used to stain routinely processed (i.e., formalin-fixed, paraffin-embedded) tissue sections. The segmentation of specific regions of interest (vessels and epithelium) is facilitated if these regions are stained to distinguish them from the surrounding tissues. Outlining of vessels can then be achieved using a specific immunologic marker of the endothelial cells that line the vessels (i.e., a commercially available CD31 stain). The epithelial component of the tissue can be distinguished from the surrounding connective tissue stroma using a specific immunologic marker of epithelial cells (cytokeratin 5D3).
Using these immunohistochemical techniques, we analyzed more than 100 breast specimens for MVD, MVA, shape,* and E:S across the four diagnostic categories, namely (1) normal, (2) fibrocystic disease, (3) benign neoplasms (fibroadenomas), and (4) malignant neoplasms (invasive carcinomas). Representative micrographs illustrating tissue types, computer processing, and staining are given in Figures 1-4. Vessel analysis of the neoplasms was compared peripherally and centrally. Adjusted i-tests assessed the effects of fibroadenoma stromal hyalinization or scarring and fibrocystic disease severity. Measurement reproducibility for the three benign diagnostic groups was assessed using Spearman correlation coefficients.
Was this article helpful?