Biology of Malignant Breast Tumors and Ultrasonography
Breast cancer is not merely characterized by features that can obviously be detected by clinical examination, medical imaging or visual evaluation of a tumor specimen, but rather by distinct intrinsic attributes. Essential tumor characteristics, like histology grade, hormone receptor status and HER2 expression, have a biological, proteomic or genetic background. Therefore, the characterization of tumors has moved from the macroscopic over the microscopic to the molecular dimension.
In the first instance, it has to be considered if it is plausible that differences in the molecular attributes of breast cancer can have an impact on the sonomorphology of the tumor. Ultrasound is principally capable of visualizing macroscopic qualities of a mass and thus detecting differences in the gross appearance. However, our results demonstrate that even sub-microscopic features of a tumor may modify its appearance in ultrasound. Understandably, ultrasound cannot directly detect intrinsic parameters of the tumor and it cannot be the aim to predict these parameters by imaging methods. Nevertheless, the typical ultrasound features of malignant breast masses may vary in distinct tumor types. Knowledge about these variations would help the examiner to avoid the false classification of breast lesions. Depending on the chosen groups, we detected a various number of different ultrasound features.
The common features of malignant breast tumors are described in specialized books. Furthermore, the American College of Radiology (ACR) has published reference guidelines on the categorization of breast tumors according to their ultrasound characteristics. However, breast cancer cannot be regarded as a single disease and according to histological, immunohistochemical or genetic features, several subtypes can be distinguished. Although there has never been a detailed and systematic approach before, we found data in the literature that focuses on certain (sono-) morphologic features of distinct subtypes and that will be discussed in the following sections.
ILC may be occult in both mammography and ultrasound, and breast-MRI may have certain advantages in the detection of this tumor type. However, the sensitivity of ultrasound seems to be higher than mammography (93.9% versus 79.8%). In 2005, Watermann et al. published data on the ultrasound features of ILC cancer. They found that an irregular shape, indistinct margins and posterior acoustic shadowing were described significantly more often in ILC than in other tumor types (88% versus 67%, p < 0.001; 95% versus 76%, p = 0.001; 84% versus 58%, p = 0.001, respectively). These findings partly comply with our own results (85% versus 71%, p = 0.116; 97% versus 94%, p = 0.700; 59% versus 45%, p = 0.164), as we found corresponding tendencies, but did not reach a level of statistical significance. Nevertheless, we can support the theory that histological differentiation modifies the ultrasonographic appearance of breast cancer. To our interpretation, the posterior acoustic features are of special importance in ILC. We found that posterior acoustic enhancement is observed significantly less often in ILC (3% versus 16%, p = 0.023) and either shadowing or mixed features or no features is significantly more frequently observed. The detectability of ILC is often impaired in both mammography and ultrasound as this tumor type has a diffuse and frequently multicentric growth pattern and does not present as a mass. However, a slight architectural distortion with a related posterior acoustic shadowing may be the only hint for this tumor type in ultrasound. The sonographer should be aware of the distinct ultrasound features of ILC in order to avoid false-negative diagnosis. We did not focus on rare histological types of breast malignancies (e.g. mucinous, medullary, tubular, mixed forms, metastases) as these entities only represented a small number of cases in our study. However, we want to emphasize, that these tumors, in particular, tend to exhibit imaging characteristics that are unique and that may be different compared to IDC and ILC. Ultrasound features of these subtypes are described elsewhere.
Focusing on the HER2 status, architectural distortions were observed significantly more often in HER2 positive tumors than in HER2 negative tumors (91% versus 78%). Gene amplification and/or protein over-expression of HER2 results in a more aggressive phenotype with increased cell proliferation, motility and tumor invasiveness, accelerated angiogenesis, and reduced apoptosis. These biological behaviors imply a rapid infiltration and destruction of the surrounding tissue and, consequently, influence both the macroscopic growth pattern of the tumor and the appearance on ultrasound. The resulting architectural distortions are a reliable predictor for malignancy and occur in the majority of HER2 positive tumors (91%). Therefore, HER2 positive tumors may be regularly detected and classified as probably malignant by ultrasound. In the literature, we found no conclusive data that could be compared to our results.
Apparently, the HR status with the associated biological background has a strong impact on the expression of sonographic features. In an earlier analysis of 281 women, Aaltomaa et al. correlated HR status with histological variables and mitotic indices. The authors described a relation to nuclear grade, tumor necrosis, tumor circumscription, inflammatory cell reaction, intraductal growth pattern and tubule formation. The authors concluded that HR negativity implies an increased proliferation rate and a number of malignant histological features in breast lesions. These histological features may explain the variation in the ultrasound characteristics between HR positive and HR negative cancers concerning a lobulated or microlobulated margin, an echogenic halo, the posterior acoustic feature and changes in the Cooper's ligaments. Just recently, Aho et al. published data on 101 breast tumors. The authors concluded, that posterior acoustic shadowing was more often associated with ER positive tumors (90.9% versus 9.1%) and PR positive tumors (72.7% versus 27.3%). This corresponds well with our results (47% versus 20%), although the differences are less accentuated in our case series, which can be explained, as we analyzed the global HR status and did not differentiate between ER and PR.
The most frequently observed echo pattern in breast cancer are hypoechoic tumors (86%). We found no variables that influence the distribution of echogenicity. Our results comply with reports in the literature that also found no difference in the groups with respect to histological size, grade, axillary metastases, hormone receptor status and lymphovascular invasion.
Virtually independent from the cut-off, stratification by age revealed that younger patients were more likely to exhibit round or oval tumors and a lobulated or microlobulated margin. On the other hand, they were less likely to show an echogenic halo, architectural distortions and disruption of the Cooper's ligaments. Consequently, the tumors may lose some of the typical criteria for malignancy in young patients and may be misjudged as benign lesions if the sonographer is not fully aware of this behavior. This phenomenon cannot be explained by the patient's age alone, but must be considered as a coincidental effect that is mainly triggered by the distinct tumor biologies that are common in young patients. Recently, Bullier et al. published data on 97 cases of breast cancer in women under 40 years old. The authors concluded, that young women have more luminal B/Her2+ phenotypes and that the appearance of cancers is correlated with their biological profiles. Consequently, sonomorphology is mainly driven by the tumor type and not by patient's age. We support this interpretation as our results suggest the same interrelation.
The main limitation of our study is that there was only one observer and image analysis was based on a single, previously acquired still image. Although the observer was blinded, this circumstance may impair some of the results. However, for further studies we would propose to include multiple observers and provide multiple images of each tumor or even video loops. Furthermore, the considerable number of groups and observed variables may boost statistical errors: Following stratification, we performed about 310 comparisons and calculated the statistical significance between the various groups. However, with an error rate of 5%we could expect no more than 16 positive results by chance. Finally, we found 93 differences that were statistically significant. Therefore, we consider that most of the positive results reflect objective differences between the groups. Nevertheless, it has to be considered that some of the apparent differences between groups may be invalid for statistical reasons.
Discussion
Is It Plausible, That the Tumor Biology Has an Impact on the Sonomorphology?
Breast cancer is not merely characterized by features that can obviously be detected by clinical examination, medical imaging or visual evaluation of a tumor specimen, but rather by distinct intrinsic attributes. Essential tumor characteristics, like histology grade, hormone receptor status and HER2 expression, have a biological, proteomic or genetic background. Therefore, the characterization of tumors has moved from the macroscopic over the microscopic to the molecular dimension.
In the first instance, it has to be considered if it is plausible that differences in the molecular attributes of breast cancer can have an impact on the sonomorphology of the tumor. Ultrasound is principally capable of visualizing macroscopic qualities of a mass and thus detecting differences in the gross appearance. However, our results demonstrate that even sub-microscopic features of a tumor may modify its appearance in ultrasound. Understandably, ultrasound cannot directly detect intrinsic parameters of the tumor and it cannot be the aim to predict these parameters by imaging methods. Nevertheless, the typical ultrasound features of malignant breast masses may vary in distinct tumor types. Knowledge about these variations would help the examiner to avoid the false classification of breast lesions. Depending on the chosen groups, we detected a various number of different ultrasound features.
Data From the Literature
The common features of malignant breast tumors are described in specialized books. Furthermore, the American College of Radiology (ACR) has published reference guidelines on the categorization of breast tumors according to their ultrasound characteristics. However, breast cancer cannot be regarded as a single disease and according to histological, immunohistochemical or genetic features, several subtypes can be distinguished. Although there has never been a detailed and systematic approach before, we found data in the literature that focuses on certain (sono-) morphologic features of distinct subtypes and that will be discussed in the following sections.
Histological Tumor Type
ILC may be occult in both mammography and ultrasound, and breast-MRI may have certain advantages in the detection of this tumor type. However, the sensitivity of ultrasound seems to be higher than mammography (93.9% versus 79.8%). In 2005, Watermann et al. published data on the ultrasound features of ILC cancer. They found that an irregular shape, indistinct margins and posterior acoustic shadowing were described significantly more often in ILC than in other tumor types (88% versus 67%, p < 0.001; 95% versus 76%, p = 0.001; 84% versus 58%, p = 0.001, respectively). These findings partly comply with our own results (85% versus 71%, p = 0.116; 97% versus 94%, p = 0.700; 59% versus 45%, p = 0.164), as we found corresponding tendencies, but did not reach a level of statistical significance. Nevertheless, we can support the theory that histological differentiation modifies the ultrasonographic appearance of breast cancer. To our interpretation, the posterior acoustic features are of special importance in ILC. We found that posterior acoustic enhancement is observed significantly less often in ILC (3% versus 16%, p = 0.023) and either shadowing or mixed features or no features is significantly more frequently observed. The detectability of ILC is often impaired in both mammography and ultrasound as this tumor type has a diffuse and frequently multicentric growth pattern and does not present as a mass. However, a slight architectural distortion with a related posterior acoustic shadowing may be the only hint for this tumor type in ultrasound. The sonographer should be aware of the distinct ultrasound features of ILC in order to avoid false-negative diagnosis. We did not focus on rare histological types of breast malignancies (e.g. mucinous, medullary, tubular, mixed forms, metastases) as these entities only represented a small number of cases in our study. However, we want to emphasize, that these tumors, in particular, tend to exhibit imaging characteristics that are unique and that may be different compared to IDC and ILC. Ultrasound features of these subtypes are described elsewhere.
HER2 Status
Focusing on the HER2 status, architectural distortions were observed significantly more often in HER2 positive tumors than in HER2 negative tumors (91% versus 78%). Gene amplification and/or protein over-expression of HER2 results in a more aggressive phenotype with increased cell proliferation, motility and tumor invasiveness, accelerated angiogenesis, and reduced apoptosis. These biological behaviors imply a rapid infiltration and destruction of the surrounding tissue and, consequently, influence both the macroscopic growth pattern of the tumor and the appearance on ultrasound. The resulting architectural distortions are a reliable predictor for malignancy and occur in the majority of HER2 positive tumors (91%). Therefore, HER2 positive tumors may be regularly detected and classified as probably malignant by ultrasound. In the literature, we found no conclusive data that could be compared to our results.
Hormone Receptor Status and Triple Negativity
Apparently, the HR status with the associated biological background has a strong impact on the expression of sonographic features. In an earlier analysis of 281 women, Aaltomaa et al. correlated HR status with histological variables and mitotic indices. The authors described a relation to nuclear grade, tumor necrosis, tumor circumscription, inflammatory cell reaction, intraductal growth pattern and tubule formation. The authors concluded that HR negativity implies an increased proliferation rate and a number of malignant histological features in breast lesions. These histological features may explain the variation in the ultrasound characteristics between HR positive and HR negative cancers concerning a lobulated or microlobulated margin, an echogenic halo, the posterior acoustic feature and changes in the Cooper's ligaments. Just recently, Aho et al. published data on 101 breast tumors. The authors concluded, that posterior acoustic shadowing was more often associated with ER positive tumors (90.9% versus 9.1%) and PR positive tumors (72.7% versus 27.3%). This corresponds well with our results (47% versus 20%), although the differences are less accentuated in our case series, which can be explained, as we analyzed the global HR status and did not differentiate between ER and PR.
Echo Pattern
The most frequently observed echo pattern in breast cancer are hypoechoic tumors (86%). We found no variables that influence the distribution of echogenicity. Our results comply with reports in the literature that also found no difference in the groups with respect to histological size, grade, axillary metastases, hormone receptor status and lymphovascular invasion.
Patient's Age
Virtually independent from the cut-off, stratification by age revealed that younger patients were more likely to exhibit round or oval tumors and a lobulated or microlobulated margin. On the other hand, they were less likely to show an echogenic halo, architectural distortions and disruption of the Cooper's ligaments. Consequently, the tumors may lose some of the typical criteria for malignancy in young patients and may be misjudged as benign lesions if the sonographer is not fully aware of this behavior. This phenomenon cannot be explained by the patient's age alone, but must be considered as a coincidental effect that is mainly triggered by the distinct tumor biologies that are common in young patients. Recently, Bullier et al. published data on 97 cases of breast cancer in women under 40 years old. The authors concluded, that young women have more luminal B/Her2+ phenotypes and that the appearance of cancers is correlated with their biological profiles. Consequently, sonomorphology is mainly driven by the tumor type and not by patient's age. We support this interpretation as our results suggest the same interrelation.
Limitations of Our Study
The main limitation of our study is that there was only one observer and image analysis was based on a single, previously acquired still image. Although the observer was blinded, this circumstance may impair some of the results. However, for further studies we would propose to include multiple observers and provide multiple images of each tumor or even video loops. Furthermore, the considerable number of groups and observed variables may boost statistical errors: Following stratification, we performed about 310 comparisons and calculated the statistical significance between the various groups. However, with an error rate of 5%we could expect no more than 16 positive results by chance. Finally, we found 93 differences that were statistically significant. Therefore, we consider that most of the positive results reflect objective differences between the groups. Nevertheless, it has to be considered that some of the apparent differences between groups may be invalid for statistical reasons.
SHARE
