Although mammography is the gold standard for the detection of early breast cancer, according to data from the Breast Cancer Detection Demonstration Project, the false-negative rate for screening mammography may be as high as 8% to 10%.1 Since then, others authors have suggested an even higher rate (up to 25%) of missed breast cancers on screening mammography.2,3 Studies have suggested that double reading4-6 or the use of computer-aided diagnosis7,8 may increase the accuracy of mammographic interpretation. Double reading of screening mammography has been found to improve the sensitivity for breast cancer detection by 5% to 15%.4-6

Causes of missed breast cancers can relate to a variety of factors, including those related to the patient, technical issues, perception, and interpretation. In a review of 320 cases of breast cancer, Bird et al2 found that 24% were missed at screening by one reader. Of the missed cancers, 61% were visible in retrospect and not called, and 18% were interpreted incorrectly. Only 5% were accounted for by technical errors.2 Careful attention to quality control, positioning, comparison with prior studies, and the meticulous use of additional views and adjuvant studies are necessary to minimize the false-negative rate of mammography. This article will discuss the causes of missed cancers and guidelines for reducing the false-negative rate.

Patient factors

Inherent breast density can certainly compromise the ability to detect a mass, particularly a noncalcified, non-distorting lesion. In the interpretation of the mammogram of a patient with dense parenchyma, particular attention to any areas of architectural distortion and careful assessment for faint microcalcifications are needed. A patient with dense parenchyma, a negative mammogram, and a palpable mass needs ultrasound for further lesion evaluation.

Other patient factors relate to difficulties in positioning and visualizing posterior breast tissue. This may occur in patients who are very tense, but also occurs in patients who have had strokes, who have shoulder problems, and in those who are otherwise debilitated.

Lesions that are located in areas that are difficult to image with mammography are inherently challenging to diagnose. Lesions located at the chest wall, particularly those high in the upper inner quadrant near the sternum or in the posterior-inferior aspect of the breast, are difficult to visualize.

Technical factors

Proper positioning and image contrast are absolute requirements for mammography. The technologist must adhere to positioning standards in order to maximize the amount of tissue included on the films.9 In order to include an area of palpable abnormality on the images, the examination must be tailored appropriately and the necessary views must be performed. Palpable areas should be identified with radiopaque markers, and the radiologist should verify that any posteriorly located palpable lesion is included in the field of view.

For densities seen on the mediolateral oblique (MLO) view only, a mediolateral (ML) view is helpful to determine if the lesion is real and whether it is located medially or laterally. Lateral lesions project lower on the ML view than on the MLO view, and medial lesions show the reverse displacement. Exaggerated craniocaudal (CC) views are also helpful to demonstrate a posteriorly located lesion seen on the MLO view only. The use of a spot compression view alone may be misleading in the assessment of a density seen on one view only. A true lesion may look less apparent on spot compression as other overlying tissue is displaced from it. Additional off-angle projections (i.e., ML or rolled CC views) should be utilized to complete the evaluation of a possible lesion.

The technologist must optimize image contrast and avoid underpenetrated or overpenetrated images. Proper placement of the photocell is necessary to achieve correct optical density on the images. Careful attention to daily processor quality control is also necessary to optimize contrast.

Perception factors

Two major causes of missed breast cancers reflect radiologist errors, namely, perception and interpretation problems. In perception errors, the lesion is included in the field of view and is evident, yet the radiologist fails to observe it. The lesion may or may not have subtle features of breast cancer, rendering it less visible than more obvious lesions.

Types of lesions that are more easily not perceived are small non-spiculated masses (figure 1), areas of asymmetry and architectural distortion, or small clusters of faint amorphous calcifications. Bird et al2 found most cancers were missed because of perception problems, and the common types of lesions missed were noncalcified indistinct or spiculated densities. Goergen et al10 found that missed cancers were statistically significantly of lower density and were seen on only one of two views more often than were detected cancers.

There are several steps that should be taken routinely to avoid missed perception of a lesion. Films should always be placed as mirror images, MLOs together and CCs together. The radiologist's search pattern should include comparing portions of the images side by side to look for any focal density. A suggested search pattern emphasizing a mirror image approach is shown in Figure 2. This search pattern enhances the ability to identify focal asymmetry density and small masses. The identification of a focal density on one view should prompt a search in the same arc (measured from the nipple) on the other view (figure 3). The posterior aspect of the breast in the retroglandular region is an area sometimes not observed by the non-expert mammographer, and attention must be applied to this zone specifically (figure 4).

Architectural distortion may be the only sign of breast cancer in a dense breast. The parenchyma must be scrutinized for any disruption of the orientation of the elements or for an area of focal pulling or tethering of the tissue (figure 5). Architectural distortion must be further evaluated unless it represents a documented post-surgical scar.

The observation of microcalcifications optimally requires careful evaluation of each film with a magnifying glass, as well as the naked eye. Optimal reading conditions necessitate high luminance viewboxes with obscuration of extraneous light. Magnification views should be performed to evaluate questionable faint microcalcifications and the morphology of observed microcalcifications.

Another perception problem that is extremely important in its impact on patient management is the diagnosis of multicentric breast cancer. Multicentric breast cancer (two or more or cancers in more than one quadrant) is a contraindication to breast conservation therapy. The observation of an obvious finding (benign or malignant) may cause the "happy eye syndrome," misleading the radiologist into not looking carefully for other lesions (figure 6).

Contralateral synchronous breast cancer occurs in 0.19% to 2.0% of patients,11 so careful attention must also be paid to the opposite breast. When one lesion that is suspicious for cancer is observed, the next step should be the search for other suspicious areas in both breasts that would require biopsy (figure 7).

Interpretation errors

Interpretation errors by the radiologist occur when a lesion with suspicious features is mischaracterized as a benign or probably benign lesion. This situation may occur because of lack of experience, fatigue, or radiologist inattention.12 Other causes, though, are that the radiologist does not use all the necessary views to assess the lesion characteristics or that the lesion is changing slowly and comparison with older examinations was not made.

When a seemingly circumscribed mass is identified, assessment of its borders should be made with spot compression. A round mass may have margins that are indistinct or microlobulated that become apparent only with spot compression, and these findings warrant biopsy. Microcalcifications require magnification views in order to assess their morphology and number accurately. Decisions about the characteristics of a lesion identified on screening are made based on diagnostic mammographic images and not the screening views alone.

Decisions about lesion location should not be made without verification of the position on multiple views. For example, one should not assume that a lesion overlying the pectoralis muscle in the superior aspect of the MLO view is a lymph node. If its features are not that of a node, additional views are performed to determine its location and significance (figure 8). Ultrasound may be a helpful adjunct in the evaluation of a posteriorly located mass seen on one view only.

Cancers that present with subtle signs of carcinoma are most challenging to diagnose. Such subtle signs include relatively circumscribed masses, focal asymmetric densities, architectural distortion, and small groups of punctate or amorphous calcifications. Well-circumscribed cancers are very uncommon, but do exist. Swann et al13 found at least a partial halo sign around 25 of 1000 breast cancers. A well-defined mass found on a baseline mammogram is considered to be probably benign and is followed at an early interval; increase in size of a non-cystic circumscribed mass warrants further intervention (figure 9).

Asymmetric densities are frequently present on mammography and are associated with a relatively low positive predictive value of malignancy. However, asymmetries that are developing, that are associated with microcalcifications or architectural distortion, or that are palpable are more worrisome and are often biopsied. Invasive lobular carcinoma accounts for approximately 8% to 10% of breast cancers, and is a type of cancer easily missed. Common presentations of invasive lobular carcinoma are an asymmetric density, architectural distortion, or with negative mammography.14

As more women undergo screening mammography on a routine basis, the possibility that a slowly changing cancer goes undetected may increase. The doubling time for breast cancer ranges from 44 to 1869 days.15 Low-grade malignancies may not change to an obvious degree between annual interval mammograms. Malignant calcifications have been reported to be stable for as long as 63 months. However, comparison with older studies 3 or more years before may demonstrate the change in size of the lesion. A mammographic abnormality should be judged by its worst features, not its most benign features. Therefore, a lesion that has suspicious features, but seems to be stable for 1 or even 2 years, may still need to be biopsied (figure 10).

Conclusion

Radiologists must make every effort to reduce the false-negative rate of mammography without unnecessarily increasing the false-positive rate. Perception and interpretation problems account for many missed breast cancers. One must pay careful attention to film review; searching for developing densities, focal asymmetries, and masses; and perform complete diagnostic evaluation of potential lesions detected. AR

Acknowledgement

The author gratefully thanks Ms. Louise Logan for her preparation of the manuscript.

References

1. Baker LH: Breast cancer detection demonstration project: Five-year summary report. CA Cancer J Clin 32:194-225, 1982.

2. Bird RE, Wallace TW, Yankaskas BC: Analysis of cancers missed at screening mammography. Radiology 184:613-617, 1992.

3. Harvey JA, Fajardo LL, Innis CA: Preview mammograms on patients with impalpable breast carcinoma: Retrospective vs blind interpretation. AJR Am J Roentgenol 161:1167-1172, 1993.

4. Bird RE: Professional quality assurance for mammography screening programs. Radiology 177:587, 1990. Letter.

5. Thurfjell EL, Lernvall KA, Taube AAS: Benefit of independent double reading in a population based mammography screening program. Radiology 191:241-244, 1994.

6. Anderson EDC, Muir BB, Walsh JJ, Kirkpatrick AE: The efficacy double reading mammograms in breast screening. Clin Radiology 49:248-251, 1994.

7. Chan HP, Doi K, Vyborny CJ, et al: Improvement in radiologists' detection of clustered microcalcifications on mammograms: The potential of computer-aided diagnosis. Invest Radiol 15:1102-1110, 1990.

8. Wu Y, Giger ML, Coi K, et al: Artifical neural networks in mammography: Application to decision making in the diagnosis of breast cancer. Radiology 187:81-87, 1993.

9. American College of Radiology

10. Goergen SK, Evans J, Colen GPB, MacMillan JH: Characteristics of breast carcinomas missed by screening radiologists. Radiology 204:131-135, 1997.

11. Kinne DW: Management of the contralateral breast. In: Harris JR, et al (eds): Breast Diseases. pp. 620-621. Philadelphia: JD Lippincott,1987.

12. Huynh PT, Jarolimek AM, Daye S: The False-negative mammogram. Radiographics 18:1137-1154, 1998.

13. Swann CA, Kopans DB, Koerner FC, et al: The halo sign and malignant breast lesions. AJR Am J Roentgenol 149:1145-1147, 1987.

14. Mendelson EB, Harris KM, Doshi N, Tobon H: Infiltrating lobular carcinoma mammographic patterns with pathologic correlation. AJR Am J Roentgenol 153:265-271, 1989.

15. Fournier DV, Weber E, Hoeffken W, et al: Growth rate of 147 mammary carcinomas. Cancer 45:2198-2207, 1980.

16. Lev-Toaff AS, Feig SA, Saitas VL, et al: Stability of malignant breast microcalcifications. Radiology 192:153-156, 1994.