SCREENING MAMMOGRAPHY

To date, the benefits from screening mammography for women 40 to 70 years of age have been proven in eight randomized controlled trials (RCTs) conducted in Europe and the United States during the past 40 years.^5–11 Reported reductions in breast cancer mortality range from 20% to 45%. Because of the relatively small numbers of women aged 40 to 49 years in the individual trials, the benefit from screening in this age range has been controversial. However, in 1997, a meta-analysis of women aged 40 to 49 years in all five Swedish trials found a 30% reduction in breast cancer deaths.¹² In addition, long-term follow-up of three trials (Health Insurance Plan Project [HIP], Gothenburg, and Malmö) each found statistically significant reductions in breast cancer mortality.^13–15 Therefore, for average risk women, most professional organizations in the United States recommend screening mammography beginning at the age of 40 years. For screening women younger than the age of 40 years, there are few RCT data. Therefore, decisions concerning screening practice in this age group must be based on less rigorous evidence and on a more individual basis. Given that younger women have a longer life expectancy and that cancers tend to grow more rapidly in younger women, earlier detection of these cancers may theoretically be advantageous. However, this must be balanced by the limitations of screening mammography in younger women, which include a lower frequency of breast cancer, reduced sensitivity of mammography, slightly increased radiation risk, and higher recall rates. Screening of younger women will be of most benefit in women at high risk, especially those known to carry the BRCA1 or BRCA2 gene mutation. Methods for estimating risk based on medical and family history include the Gail, Claus, and BRCAPRO mathematical models. Observational studies of high-risk women aged 30 to 39 years show a cancer detection rate similar to that for women aged 40 to 49 years.^16–17 In general, screening of women aged less than 40 years is restricted to high-risk subgroups (women with a 20% lifetime breast cancer risk at or before the age of 30 years or breast cancer risk at a given age equivalent to that of the average woman at the age of 40 years). These include BRCA1 or BRCA2 gene mutation carriers, women with a personal history of breast cancer, women with a prior diagnosis of atypical ductal or lobular hyperplasia, women with previous radiation therapy to the chest before the age of 30 years, and women with a strong family history of breast cancer (usually involving one or more first-degree relatives with premenopausal breast cancer or breast cancer before the age of 50 years).

LIMITATIONS OF MAMMOGRAPHY

Although there have been significant improvements in mammographic technique over the past 50 years, fundamental limitations remain. These include the low inherent contrast differences between tissue structures in the breast and the fact that mammographic detection of breast cancer (sensitivity) relies on the ability to visualize cancer through the background of overlying normal tissue (Figure 1). Mammographic specificity relies on the ability to distinguish benign from malignant breast lesions based on their margins and morphologic features.^18–19 However, malignant and benign lesions may have similar appearances, thereby reducing specificity. Data from the Breast Cancer Surveillance Consortium demonstrated that on average, screening mammography programs had a callback rate of 6.4% to 13.3% and a positive predictive value of 3.4% to 6.2%.²⁰ The mean cancer detection rate was 4.7 per 1000, and the mean size of invasive cancers was 13 mm. The main mammographic signs of breast cancer include clustered microcalcifications, masses, architectural distortions, and asymmetrical densities. Comparison to prior mammograms is essential because some cancers may only be detected by perceiving them as a subtle change from prior studies. In addition, the availability of prior mammograms for comparison can reduce the number of unnecessary callbacks for stable benign findings.²¹

FIGURE 1 A 58-year-old asymptomatic woman undergoing yearly screening mammography. A, A spiculated mass is seen in the medial left breast on the CC view, isodense to breast parenchyma. On biopsy, this proved to be an invasive ductal carcinoma. B, In retrospect, the mass is evident on the study 2 years before, but is difficult to differentiate from glandular tissue.

DIGITAL MAMMOGRAPHY AND COMPUTER-AIDED DETECTION

Screen-film mammography (SFM) has been the standard method used for breast cancer screening since the end of xeromammography some 20 years ago. Advances in screen-film technology and film-processing techniques have contributed to major improvements in the quality of mammographic images. In addition to high contrast, the strength of SFM lies in its extremely high spatial resolution, often greater than 10 line pairs per millimeter (lp/mm). This allows the detection of exceedingly small clusters of microcalcifications, one of the earliest signs of breast cancer. The limitations of SFM include the detection of subtle soft tissue lesions, especially in the presence of dense glandular tissues, and the fact that the film serves simultaneously as the image receptor, display medium, and long-term storage medium. Recent technologic advances have led to the development of full-field digital mammography (FFDM). One of the initial concerns of FFDM is its inherent lower spatial resolution of 5 to 10 lp/mm. However, several studies have demonstrated that despite the limited spatial resolution, the visibility of calcifications on FFDM is not significantly different from that on SFM.^22–24 The higher contrast resolution of FFDM may account for its comparable detection rates. In a recent multicenter trial of 49,528 women, Pisano and colleagues²⁴ reported that the overall diagnostic accuracy of FFDM was comparable to SFM as a means of screening for breast cancer. The study also found that digital mammography was more accurate in women younger than 50 years, women with radiographically dense breasts, and premenopausal or perimenopausal women.

Digital mammography has the potential to overcome the inherent limitations of SFM. By directly converting the detected x-ray photons to numerical values, the process of x-ray photon detection is decoupled from the image display. The digital images can be processed by a computer, displayed in multiple formats, and fed directly to computer-aided detection (CAD) software programs. In addition, there are logistical and financial advantages of FFDM, including faster patient throughput, no film or processing costs, the ability to transmit images, and the ability to perform telemammography.

CAD programs were developed to assist a radiologist in the interpretation of screening mammograms, so that cancer detection rates could be improved. These programs rely on neural networks to analyze the images and highlight potentially suspicious findings that may have been overlooked by the radiologist. Several studies have shown improved cancer detection by radiologists using CAD versus radiologists alone, without significantly increasing callback rates.^24–27 However, other studies have shown less promising results.^28–29 A recent study by Fenton and colleagues²⁹ reported that CAD decreased the accuracy of screening mammogram interpretation. This was not due to a decrease in cancer detection, but rather to an increase in false-positive results. To be effective, CAD programs should not increase callback rates and should not significantly prolong interpretation times for screening mammograms. It is important to remember that CAD programs are tools that must be used in an appropriate manner. They should not be used to override a suspicious finding detected by a radiologist. In general, these systems tend to be more helpful for low-volume or inexperienced readers.

MAGNETIC RESONANCE IMAGING

Dynamic, contrast-enhanced MRI of the breast has been shown to be extremely sensitive in the detection of invasive breast cancer and is not limited by the density of the breast tissue. However, because the reported sensitivity of MRI for ductal carcinoma in situ (DCIS) ranges between 45% and 100%, MRI is currently not recommended as a replacement for mammography.³⁰ A more recent single institution study suggests that MRI may have a higher sensitivity than mammography for DCIS than previously thought, particularly for high-grade DCIS.³¹ The use of MRI in the general population has been limited by its moderate specificity. Therefore, its use has been focused on studying patients in whom the yield from MRI is likely to be higher. Multiple studies have shown that MRI is a useful tool as an adjuvant to screening mammography in women at high risk for breast cancer.^32–34 The American Cancer Society (ACS) recommends annual screening MRI for women with a 20% to 25% lifetime risk for breast cancer.³⁵ This includes women with the BRCA1 or BRCA2 breast cancer genes, as well as women with multiple family members with breast or ovarian cancer, and women who have undergone mediastinal irradiation for Hodgkin’s disease. Women whose benefit from screening MRI was considered questionable by the ACS because of insufficient data included women with a personal history of breast cancer, prior biopsy yielding atypia, or extremely dense breasts on mammography. The decision to perform screening MRI in these women should be made on a case-by-case basis. Several models may be used to calculate lifetime risk for breast cancer, including the Gail, Claus, and Tyrer-Cusick models.

ULTRASOUND

Early studies of breast ultrasound for cancer screening were disappointing because of its poor detection of small cancers and excessively high false-positive rates.^36–40 With the advent in the early 1990s of technical improvements in ultrasound, including improved spatial and contrast resolution utilizing higher-megahertz (MHz) transducers, the potential of whole-breast ultrasound as a screening tool has been revisited. Several recent singleinstitution studies have demonstrated a prevalence detection rate of 3 to 4 mammographically occult cancers per 1000 women screened.^41–50 However, the biopsy positive predictive values were less than 20%, lower than accepted for mammography.⁵¹ These initial studies have focused mainly on women with mammographically higher-density breasts. Screening ultrasound appears to be more sensitive in detecting early invasive cancer, whereas mammography is more sensitive in the detection of DCIS. Of the invasive cancers, ultrasound found a higher percentage of invasive lobular carcinomas than that usually found on mammography. Therefore, whole-breast ultrasound should supplement mammographic screening, rather than replace it.

Despite these promising early results, the use of screening ultrasound in general has not been recommend by professional organizations because of concerns regarding the scientific validity of these initial studies and the lack of randomized controlled trials.

The initial studies were performed mostly by radiologists with a high ultrasound skill level and in some studies were not blinded to the mammographic findings. Therefore, the initial results of screening ultrasound may not extrapolate to its use in general clinical practice.⁵² Other concerns include the lack of standardized exam techniques, interpretation criteria, false-positive results, and unnecessary biopsies. In addition, most of the initial studies on whole-breast ultrasound have evaluated prevalence (initial) detection rates, rather than incidence (subsequent) detection rates. The benefit from subsequent yearly screening ultrasound is uncertain, but likely to be less. The American College of Radiology Imaging Network (ACRIN) is currently conducting a randomized multicenter trial evaluating whole-breast bilateral screening ultrasound in high-risk, asymptomatic women with dense breasts. The study will evaluate both prevalence (year 1) and incidence (years 2 and 3) screening ultrasound detection rates as compared with mammography. Interpretation of the examinations will be performed by radiologists trained in mammographic and ultrasound interpretation, using standardized interpretive criteria.

Although ultrasound is less sensitive than MRI, because of its lower cost and availability, it is likely that whole-breast ultrasound will play an increasing role as a supplemental screening modality in women with dense breasts. Current postprocessing algorithms, including spatial compounding and harmonic imaging, as well as newer techniques such as elastography, may help to improve the specificity of breast ultrasound and decrease the number of false-positive results and unnecessary biopsies. In addition, automated whole-breast ultrasound systems, currently being developed by several manufacturers, may help to standardize and streamline whole-breast ultrasound.

MOLECULAR IMAGING

Previous attempts of breast imaging using ^99mTc sestamibi and ¹⁸F fluorodeoxyglucose (FDG) positron emission tomography (PET) were limited by the use of relatively low-resolution whole-body systems. However, newer high-resolution dedicated breast imaging systems have been developed. Breast-specific gamma imaging (BSGI) and positron emission mammography (PEM) allow high-resolution imaging of the breast based on metabolism, rather than anatomic changes. Their use as screening tests has not been established. Variability of uptake in primary breast tumors, as well as moderate specificities, has limited their potential. Randomized control trials are needed to evaluate their possible role as an adjuvant screening tool in certain subgroups of women.