Supplementary Screening Sonography in Mammographically Dense Breast: Pros and Cons

The potential use of sonography for breast cancer screening was first evaluated in the 1980s. Until the early 1990's, however, the studies found that sonography was not useful for screening because of inadequate detection of smaller cancers and excessive false-positive biopsy rates (11). With advances in imaging equipment and techniques of sonography, in 1995, Gordon and Goldenberg (12) first reported the ability of sonography to depict nonpalpable, mammographically occult solid masses, with 3% of these masses being cancers (prevalence, 0.3%). Since that time, investigators have reanalyzed the use of screening sonography in populations of women with dense breasts, which are summarized in Table 1 (10).

Table 1
Results with Screening Breast Sonography

• Click for larger image

The studies consistently showed that in women with mammographically dense breast, sonography was able to detect a substantial number of cancers, with supplemental cancer detection of 0.3-0.5% by sonography alone. In one study including women with fatty breasts (13), 11 of 16 cancers diagnosed at sonography were detected in Breast Imaging Reporting and Data System (BI-RADS) density three and four breast tissue and the remaining five in BI-RADS density one and two breast tissue. Moreover, the data indicated that most breast cancer detected at screening sonography and previously not detected was at an early stage (Table 1). In the study by Kaplan (14), all six cancers were stage 0 or 1 and four of them were smaller than 1 cm in size. Corsetti et al. (15) reported that the proportion of early-stage cancer was significantly higher in the sonography-only detected (65%) than the mammography detected cancers (36%, p = 0.001). It is well established from randomized controlled trials of mammographic screening that earlier detection of cancer results in a decrease in mortality which parallels the reduction in size distribution of cancers depicted and closely parallels the reduction in rates of node-positive breast cancer (1).

As mentioned before, breast density is one of the factors that points out false-negative findings in mammography. Furthermore, mammographically dense breast tissue has been identified as an independent marker strongly associated with breast cancer risk and in particular with a higher risk of interval cancer (16). Supplemental screening imaging such as sonography and MRI can be used to detect those mammographically occult cancers in dense breast. Compared with MRI, sonography is relatively inexpensive, requires no contrast, is well tolerated by patients, and is widely available for equipment, which is attractive as a supplement to mammography. Even more, sonographically guided biopsy of lesions is fast and easy to readily perform (4).

The benefit of detection with supplemental sonography in mammographically occult cancers can increase with increasing grades of breast density (10, 13, 17). For the sonography of dense breast, most breast cancers are relatively hypoechoic within a background of hyperechoic fibroglandular tissue, which would make lesions more conspicuous and detectable. In contrast, about one-third of breast cancers are only mildly hypoechoic with respect to fat, and these very subtle, mildly hypoechoic or isoechoic lesions are more likely to be indistinguishable from surrounding fat in fatty breast (18).

In the studies of screening sonography in women with mammographically dense breasts, three to four breast cancers per 1000 women were detected by sonography only, and generally at an early stage (Table 1). This suggests that sonography as an adjunct screening test may provide further benefit in screening women with dense breast tissue. However, most studies were performed at single center and sonographic interpretations might be biased because readings were not truly blinded to mammographic findings and in many cases, performed sonography only when the mammogram was negative (17). To address those limitations, a prospective multicenter trial of screening sonography for women with intermediate and high breast cancer risk (American College of Radiology Imaging Network [ACRIN] protocol 6666) (19) was performed. The sequence of performing mammography and sonography was randomized to help control the biases of recruiting women with vague mammographic abnormalities, and examinations were performed by radiologists who were masked to results of the other examination. As a result, sonography was associated with a 55% increase in diagnosing breast cancer compared with mammography alone. Adding a single screening sonography to mammography yielded an additional 4.2 cancers per 1000 high-risk women. The study provided evidence for the importance of supplemental breast sonography screening. Furthermore, these results were consistent and generalizable across 21 international centers, so that standardized scanning and interpretive criteria in this trial proved to be practicable for independent performance and interpretation (16).

Although the addition of a single screening sonography to mammography increased the diagnostic yield, it also substantially increased the number of false positives. The positive predictive value for lesions subjected to biopsy on the basis of screening sonography findings ranged from 8% to 18% (mean, 9%) (Table 1). Compared with the positive predictive value of 25% to 40% recommended for lesions undergoing biopsy on the basis of screening mammographic findings, screening sonography may be expected to result in 2.5 to 4.0 times as many false-positive biopsy findings per cancer detected (11). Although it is one of the advantages of sonography that sonographically guided biopsy is relatively simple and less invasive, minimizing unnecessary patient anxiety and costs is a critical goal of any new screening procedure (17). In the ACRIN protocol 6666 trial, the false-positive rates for mammography plus sonography (10%) and sonography alone (8%) were higher than that of mammography alone (4%), even if the radiologist investigators in this trial were all specialists in breast imaging who met experience requirements and completed qualification tasks in sonography (19, 20).

As for cancer characteristics, the vast majority of cancers seen only on sonography were invasive (94%) because ductal carcinoma in situ (DCIS) is difficult to see on sonography (Table 1). Tabar et al. (21) found that microcalcifications were the predominant mammographic feature in 85% of cases of high-grade DCIS and 69% of cases of intermediate- and low-grade DCIS. The interaction of ultrasound with normal breast tissue produces tiny bright echoes known as speckle, which simulate faint microcalcifications and make detection of true microcalcifications difficult (22). By projecting ultrasound beams from several angles, compound imaging will diminish speckle, but also reduces resolution (11). In contrast, invasive cancers which often present as hypoechoic masses on sonography, would be more conspicuous and detectable within a background of dense fibroglandular tissue (23). However, it is uncertain whether the detection of DCIS is required or whether the detection of node-negative invasive breast cancer is sufficient for a screening test (19).

Breast sonography is limited by operator dependence. The detection of an abnormality at sonography is completely dependent on the perception and skill of the person performing the procedure (24). Unlike most other radiologic examinations, a lesion not detected while performing breast sonography will generally not be documented; and additionally, having representative normal images is not proof that a mass was not present (25). This is a limitation of ultrasound in general (24). Given that sonography is operator-dependent, intra- or interoperator variability is to be expected in screening sonography. Although Berg et al. (25) found substantial reliability for reporting lesion size, location, and key features, as was moderate agreement for lesion management with whole-breast sonography, participating examiners were physicians specialized and highly skilled in breast sonography and they performed sonography using a standardized scanning protocol and interpretive criteria. Conversely, without the implementation of a standardized technique and interpretation, as well as training and qualification for screening sonography, the result in general practice could vary from the aforementioned study. Difficulty with reproducibility potentially limits the ability to accurately monitor the sonography-only masses that are thought to probably be benign and suitable for surveillance with follow-up examinations (24).

Screening sonography is so time-consuming and would require a much larger number of qualified breast radiologists than currently available. In the ACRIN protocol 6666 trial (19), a breast screening sonography takes an average of 19 minutes of physician time. In comparison, breast radiologists will complete less than three screening sonographic studies per hour. However, a breast radiologist, if involved in batch reading of screening mammograms, will read about 50 studies per hour (26). Moreover, the time of 19 minutes does not include comparison to prior studies, discussion of results with patients, nor the creation of a final report (19). At this time, however, there is insufficient evidence to address the efficacy of technologist-performed screening sonography (27). Furthermore, many incidental lesions will be detected during screening sonography which require a biopsy or follow-up. Whether the additional cost of screening with sonography will be cost-effective should be determined (28).

To overcome the lack of uniformity or reproducibility and the shortage of qualified personnel limiting the wide implementation of screening sonography, automated whole breast sonography may be helpful. In a study of automated whole breast sonography added to mammography in asymptomatic women with dense breasts and/or at an elevated risk of breast cancer (29), adding automated whole breast sonography to mammography yielded an additional 3.6 cancers per 1000. This finding is consistent to supplemental yield findings of earlier studies using handheld sonography (2.7 to 4.6 per 1000), and moreover, 87% of cancer detections added by automated whole breast sonography were found in 68% of studies in women with dense breasts. However, clinical validation by further studies is needed to be implemented in breast cancer screening.

In general, the rate of cancers detected in patients screened for the first time (prevalent cancers) should be much higher than in a population that has been screened previously (incidental cancers). Thus, knowledge of the proportion of initial screening examinations to follow-up screening examinations can be extremely useful in interpreting the results (30). Although Kolb et al. (8) performed multiple rounds of screening sonography, they did not distinguish results from each round. The other series to date reported on a single prevalence screen; an estimate of the yield from annual screening sonography is not available. Compared with mammographic screening, the cancer detection rate of sonographic screening (2.7-4.6/1000 exams) was lower than would be expected from a mammographic prevalence screen (6-10/1000 first-time exams), but was comparable to a mammographic incidence screen (2-4/1000 follow-up exams). The reason for similar results of screening sonography to mammographic incidence screening could be attributed to the fact that many women in these studies probably had several previous mammographic screenings but no previous sonographic screening. Moreover, factors favoring sonographic detection are different from those favoring mammography detection (11). The data on incident screens from subsequent rounds in the ACRIN protocol 6666 trial will be forthcoming and will be important for determining stage distribution of breast cancers in subsequent rounds of screening with mammography plus sonography in this study (1, 27).

In conclusion, although mammography remains the mainstay of breast cancer screening, screening sonography in women with mammographically dense breast has been reported consistently to detect a substantial number of sonography-only cancers at an early stage. However, further study is needed to validate and implement the screening sonography in clinical practice, which is ongoing. Radiologists need to be familiar with the pros and cons of screening sonography and also be able to frame a proper concept of sonography for breast cancer screening.