In multiple studies, mammography has been demonstrated to be an effective method of detecting clinically inapparent breast cancer. In multiple screening studies, approximately 90% of breast cancers that are able to be discovered radiographically or by palpation are identifiable by mammography. However, this means that mammogram results are negative in about 10% of screen-detected cancers. Additionally, mammography often fails to demonstrate palpable masses, particularly in women with dense breast parenchyma. In both of these instances, the mammogram is falsely negative, either by failing to detect breast cancer or failing to image palpable benign or malignant masses.

Little research has been done on mammography's failure to characterize benign entities that are apparent on physical examination. However, the failure of mammography to detect otherwise evident breast cancer has been studied extensively. A review of malpractice litigation helps to demonstrate the types of cases in which breast cancer diagnosis is missed or delayed. In a 1995 review, it was reported that more than half of the cases of missed or delayed diagnosis involved women under the age of 50 years.1 In 60% of the cases studied, the cancer was originally discovered by the patient, with a breast mass as the presenting symptom in 50%. In almost 80% of cases, the mammogram was normal or equivocal, despite the presence of a palpable mass. Unfortunately, radiologists have become the most frequent defendants in these cases, now accounting for 24% of physicians named in these cases.

Several analyses of the causes of missed diagnoses on mammography have been published. In 1992, Bird2 published a review of 77 cancers that were missed by screening mammography. As compared to cancers that were correctly diagnosed, missed cancers frequently occurred in more dense breast tissue, less frequently had malignant microcalcifications associated with the cancer, and were more likely to be evident mammographically as a developing density. Other studies have confirmed that increased breast density and tumors without associated calcifications limit the sensitivity of mammography.3,4 Small cancers have been found to be more difficult to diagnose, especially when they are lobular rather than ductal carcinomas. Additionally, those cancers that fail to incite a desmoplastic reaction are likely to be missed on mammography.

The incidence of false negativity may be higher in women with palpable breast lesions than in those whose cancers are nonpalpable. In a report of 499 women with palpable cancers, 22% had a false negative mammogram.5 In another report, the incidence of false negative mammograms in women with palpable breast cancer decreased by 50% during the 1980s, due to improvements in radiographic equipment and technique.6

Why are lesions missed on mammography? In some cases no findings are present on the mammogram. In one study, this was the case in 3% of 323 women with operable breast cancer who underwent mammography.7 On review of the mammograms of these nine women, the cancer was obscured by adjacent breast tissue in five cases, and faulty positioning of the patient was to blame in four. This is consistent with data from another series in which missed cancers tended to be located close to the chest wall in the retroglandular area of the breast.2 Few cancers are actually missed at the time of mammographic interpretation, although initial findings associated with a cancer may be similar to those usually associated with a benign entity, and therefore may not prompt aggressive action at the time the lesion first appears on mammography.

Histologies associated with missed breast cancers.

Cancers that are most likely to be missed on mammography are those that are small, not associated with calcifications, and those that do not incite a desmoplastic reaction. The classic example of this histologic type is invasive lobular carcinoma, which is difficult to diagnose both mammographically and on physical examination. The presence and extent of invasive lobular carcinoma is only diagnosed accurately in about one third of the mammograms performed to detect this disease;8 however, because invasive lobular carcinoma accounts for less than 10% of invasive breast cancers, the impact of this problem is minimized.

The limited ability to appreciate the presence of invasive lobular cancers is due to the tumor's tendency to grow along normal architectural planes in the breast, not forming a focal mass and not inciting any scirrhous reaction. Also, calcifications are usually not associated with these tumors. These same findings make the detection of this disease difficult on physical examination.9 Invasive lobular cancer can have a pattern similar to that of normal breast tissue and, therefore, careful physical examination and comparison to earlier mammograms, if available, is necessary in assessing women with asymmetric patterns of parenchyma. The presence of subtle physical findings or an interval increase in the pattern of asymmetry may require a biopsy to be performed to determine if carcinoma is present.

Not surprisingly, smaller cancers are more difficult to diagnose than large cancers. They represent more subtle findings on the mammogram and can require intense concentration to identify. Because the mammographic pattern of the breast is often filled with variations in parenchymal density, the presence of more dense microcalcifications is often more striking than isodense uncalcified masses, as calcifications provide more contrast with the background pattern of the breast than do uncalcified masses. It is therefore not surprising that the smallest mammographically visible cancers, in situ ductal carcinomas, are rarely found if they are uncalcified. In two series, 11% or less of cases of mammographically evident ductal carcinoma in situ were discovered without associated calcifications.10,11 It should be expected that invasive carcinomas, especially those that are small, will be difficult or impossible to diagnose mammographically without calcifications, unless they incite peritumoral scirrhous reaction to make them more obvious.

Improving visualization in the dense breast

Although dense breast tissue can be found in women of any age, it is most common in those under the age of 35; it will be found in approximately one third of these younger women. Glandular tissue is of the same radiodensity as an uncalcified tumor and, therefore, the tissue has the potential of silhouetting and obscuring an isodense cancer. This is a significant problem in women with dense breast tissue, as approximately 50% of breast cancers will not contain mammographically apparent calcifications.

Although mammography will never be as effective in women with dense breasts as it is for those with fatty patterns, mammogram results in these women can be improved by using optimal radiographic technique. The use of compression is particularly important in this population. Maximal compression minimizes internal scatter and thins out normal breast tissue, thereby improving the possibility of visualizing an underlying mass. Eliminating motion artifact also can improve the chance of detecting nonpalpable masses. Though grids that are used routinely in modern mammography probably add little to the imaging of a fatty breast, in women with more-dense breast tissue, scatter becomes an important factor in image degradation, and the use of a grid is therefore more important.12 "Harding" the x-ray beam, or using a rhodium filter and/or target might also improve penetration and decrease scatter;13 however, contrast is reduced when a higher energy beam is used.

In any mammographic examination, patient positioning must be optimal. It is obviously not possible for the radiologist to detect clinically occult disease in areas of the breast that are missing on the mammogram. The radiologist should be familiar with the limitations of each view, and be able to critique films from the viewbox.14 Additionally, quality imaging cannot be performed without quality equipment that is functioning well. Mammographic units, processors, darkrooms, and cassettes all should be maintained in the best condition possible. Films should be critiqued constantly to immediately recognize problems that could impact on the quality of imaging, and action should be taken as soon as possible to correct these problems. The radiologist also should appreciate that, in some women, the routine of taking only two screening views may not be adequate to image the breasts. Additional views may be required to include some tissue that cannot be seen on either the MLO or CC views. In some cases, penetration of extremely dense breast tissue will burn out fatty areas; for these women, two sets of films may be necessary to enable the appropriate radiographic technique for radiodense and radiolucent areas.

Another issue that radiologists should consider is the inappropriate use of a caveat on a mammography report stating that some cancers are missed on these examinations. In women with a fatty parenchymal pattern, the sensitivity of mammography approaches 100%. Statements about the limitations of mammography should be saved for reports on those women with dense patterns in which sensitivity is truly compromised, particularly for women with palpable abnormalities in which no findings are present on imaging studies.

The palpable mass without mammographic findings

For women in whom findings are pres-ent on physical examination but not on mammography studies, the workup should be tailored to attempt to more-completely evaluate the palpable finding. It may be valuable to place a radiopaque marker over the palpable lump to ascertain its inclusion on the mammogram and its exact location on the films. Coned views are valuable, in that they focally compress the area of the breast containing the mass, decreasing thickness of the breast at this site. This may increase the possibility of visualizing the mass by decreasing surrounding tissue that might silhouette it, and by decreasing scatter, which which will increase sharpness. Tangential views also are useful, as they partially throw the mass beyond the cone of breast tissue, allowing at least part of its contour to be visualized. Some or all of these techniques should be attempted if a palpable mass is not fully characterized on the initial mammographic views.

Sonography is perhaps the most important technique in detection of a palpable lesion that was missed on the mammogram. It is particularly useful in women with dense breasts15 (figure 1). Sonography can be used to characterize the internal pattern of masses, and is extremely valuable in defining the margins of masses. In premenopausal women, in whom new, palpable masses will often be due to cysts, sonography makes a definite diagnosis possible, eliminating the need for biopsy in these women. In younger women, sonography should be the initial imaging study used to assess a palpable lesion. However, the absence of a focal, sonographic mass in the presence of a palpable lesion does not indicate that the mass that is felt is clinically insignificant. In one study, 2 of 7 cancers that were palpable in dense breasts were sonographically imapparent.15 The absence of sonographic findings may indicate only that the mass is not a cyst and is therefore solid. Biopsy should be considered in this setting, as breast cancers can be isoechoic with breast tissue.

Improving perception of abnormal mammographic findings

Even with the best quality images, some errors in the interpretation of mammography still occur. Fortunately, these account for less than 10% of the cancers missed by mammography.16 Some infiltrating ductal carcinomas, medullary carcinomas, colloid carcinomas and intracystic papillary carcinomas will be imaged as smooth, rounded masses without calcifications, and these can easily be mistaken for benign lesions. Fortunately, these cancers are rare.

Some mammographic findings are difficult to appreciate. The quality of film interpretation can be improved by cleaning the viewboxes routinely and by using lights with the same intensity in all viewboxes used for mammography. Studies should be read in darkened rooms, with ambient light minimized. A radiopaque marker to localize palpable masses may help the radiologist to locate significant findings, and to identify subtle changes that might be associated with these lesions. Additionally, the radiologist should minimize any outside distractions while he or she interprets mammography. Attending to multiple duties at the same time decreases the quality brought to each of these. Reading mammography is difficult enough when done in isolation, it may approach the impossible when done with other tasks.

One of the most common distractions that may lead to misinterpretation of a mammogram is the presence of an obvious finding, benign or malignant. A large, readily apparent benign finding can distract the radiologist from the search for a more subtle carcinoma. As breast cancer is often multifocal, multicentric, or bilateral, identification of findings that suggest the presence of one cancer should prompt an aggressive search for possible other such lesions in either breast. The presence of multiple cancers in the same breast may change treatment options and, therefore, a complete and detailed search is vital, especially for women who opt for breast conservation. Just as importantly, it would be unfortunate to successfully treat disease in one breast while missing an additional early, curable cancer on the opposite side.

Summary

When performed with optimal technique, mammography can detect over 90% of breast cancers. Unfortunately, some cancers are not evident on mammography. However, when lesions are palpable, imaging studies can be tailored to optimize their assessment. Radiologists should be aware that interpretation of these examinations can be optimized by good viewing conditions that decrease ambient light and maximize the radiologist's attention to the task. Because multiple breast cancers are often present, identification of a suspicious lesion should prompt a search for other possible sites of cancer within the breasts. AR

References

1. Breast Cancer Study. Physician Insurers Association of America. Washington, DC, 1995.

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

3. Holland R, Hendriks JHCL, Mravunac M: Mammographically occult breast cancer: A pathologic and radiologic study. Cancer 52:1810-1819, 1983.

4. Ma L, Fishell E, Wright B, et al: Case-control study of factors associated with failure to detect breast cancer by mammography. J Natl Cancer Inst 84:781-785, 1992.

5. Edeikin S: Mammography and palpable cancer of the breast. Cancer 61:263-265, 1988.

6. Wallis MG, Walsh MT, Lee JR: A review of false-negative mammography in a symptomatic population. Clin Radiol 44:13-15, 1991.

7. Cahill CJ, Boulter PS, Gibbs NM, Price JL: Features of mammographically negative breasttumours. Br J Surg 68:882-884, 1981.

8. Rodenko GN, Harms SE, Pruneda JM, et al: MR imaging in the management before surgery of lobular carcinoma of the breast: Correlation with pathology. AJR 167:1415-1419, 1996.

9. Hilleren DJ, Andersson IT, Lindholm K, Linnell FS: Invasive lobular carcinoma: Mammographic findings in a 10-year experience. Radiology 178:149-154, 1991.

10. Stomper PC, Connolly JL, Meyer JE, Harris JR: Clinically occult DCIS detected with mammography: Analysis of 100 cases with radiographic pathologic correlation. Radiology 172:235, 1989.

11. Dershaw DD, Abramson A, Kinne DW: Ductal carcinoma in situ: Mammographic findings and clinical implications. Radiology 170:411-415, 1989.

12. Dershaw DD, Masterson ME, Mali S, Cruz NM: Mammography using an ultrahigh-strip-density, stationary, focused grid. Radiology 156:541-544, 1985.

13. Gingold EL, Wu X, Barnes GT: Contrast

and dose with Mo-Mo, Mo-Rh, and Rh-Rh target-filter combinations in mammography. Radiology 195:639-644, 1995.

14. Bassett LW, Hirbawi IA, DeBruhl N, Hayes MK: Mammographic positioning: Evaluation from the view box. Radiology 188:803-806, 1993.

15. Dershaw DD, Eddens G, Liberman L, et al: Sonographic and clinical findings in women with palpable breast disease and negative mammography. Breast Dis 8:13-17, 1995.

16. Kalisher L: Factors influencing false negative rates in xeromammography. Radiology 133:297-301, 1979.