Abstract
Breast cancer is a leading cause of death among women in the United States, but good evidence equates early detection with reduction in breast cancer mortality. X-ray screening mammography remains the most sensitive noninvasive technique for detecting early tumors when women are asymptomatic and cancers may still be noninvasive. Limitations that reduce both sensitivity and specificity are inherent in all medical testing and mammography has been reported to miss 10%-15% of breast cancers. Federal standards, along with marked improvements in technology have led to improved regulation, inspection, and compliance requirements. Refinements continue to be made in film/screen combinations, but the ability to obtain the high contrast required to resolve fine structures with only minor density differences is limited by a narrow dynamic range. Recent technologic developments include Computer-Aided Detection (CAD) and Digital Mammography, which work on the principle that cancers cause specific patterns of abnormality and are immune to fatigue, illness, and distraction. The ImageChecker (R2 Technology, Inc, Los Altos, CA), currently the only FDA approved CAD system, was installed at the Breast Imaging Center of Ochsner and three comprehensive studies were performed by R2 Technology, Inc. The technology is promising.
Breast cancer is a leading cause of death among women in the United States. In 1997, over 180,000 new cases were diagnosed and almost 44,000 deaths were attributed to breast cancer (1). Good evidence equates early detection with reduction in breast cancer mortality (2). X-ray screening mammography remains the most sensitive noninvasive technique for detecting early tumors (3, 4), though periodic enthusiasm develops for other nonradiation imaging methods of cancer detection such as thermography, diaphanography (light scanning), whole breast ultrasound, and magnetic resonance imaging (MRI). While additional methods can refine screening mammography or clinically based impressions, they do not approach the reliability, low cost, and efficiency of conventional mammography as a mass screening test for large populations (5, 6).
Limitations that reduce both sensitivity and specificity are inherent in all medical testing. Traditionally, mammography has been reported to miss 10%-15% of breast cancers (7). The “misses” can be attributed to three major types: intrinsic growth patterns of select cancers, technical factors (equipment performance, film processing, technologist positioning, and breast compression), and the failure of radiologists to detect very subtle signs of malignancy (8, 9).
Little can be done to improve the detection of malignancies that do not create mammographic alteration. These malignancies may present clinically and, when a clinical abnormality is suspected, diagnostic mammographic and sonographic techniques are vigorously applied to hone the clinical impression. Often, clinically suspicious findings that may have been invisible on screening mammography can be confirmed by highly specified imaging (10,11), though a small number may remain occult to imaging techniques. No clinically suspicious finding should be ignored or downplayed in the face of a negative mammogram—especially a screening mammogram, which is only for an asymptomatic woman.
In 1987, a watershed year for mammography, the voluntary American College of Radiology (ACR) Mammography Accreditation program was instituted to address technical and interpretive factors influencing mammography “misses” (12). Simultaneously, marked improvements in equipment and film/screen combinations were occurring in the competitive marketplace (13). The accreditation program requires each machine at a facility to meet or exceed performance standards. Each step in the imaging chain was standardized, and a specific registry test was devised to certify technologists to perform mammography. Interpreting physicians were required to have specific training and continuing education, and to maintain skill by evaluating a proscribed number of mammograms quarterly (14). The voluntary program was quite successful but, when Medicare began to pay for screening mammograms, mandatory compliance to performance standards naturally followed. This took form through the federal Mammography Quality Standards Act (MSQA) of 1992, mandating standards similar to those of the ACR. The regulatory power was delegated to the Food and Drug Administration in 1993, which has further developed regulation, inspection, and compliance requirements (15).
Mammographic equipment, technical performance, and radiologists' skills have improved during this time but radiologists must still balance the need for a high sensitivity for mammographic abnormalities (high cancer detection) with high specificity (reduction in the number of false-positives leading to call-backs and biopsies for findings that ultimately are not cancer). Observational lapses by radiologists persist and, surprisingly, are not strongly related to experience (16, 17). The tendency to err is all too human and probably cannot be eliminated entirely. This has spurred research into developing aids for the radiologist. Notable among these are Computer-Aided Detection (CAD) and Digital Mammography (18, 19).
Double reading by two observers has been shown to improve radiologists' performance, but it is usually logistically problematic as well as time consuming. Even double reading does not eliminate all lapses, and it is not routinely practiced in the United States. CAD can function as a second reader that is not prone to fatigue, illness, and distraction. These devices work on the principle that cancers cause specific patterns of abnormality (i.e. calcifications or dense areas with radiating tissue patterns). By programming specific parameters, image analysis can be performed by a sophisticated computer system. The mammogram is digitized for computer analysis, and areas with features of cancer are flagged on a video monitor. This allows the radiologist to double check findings and may help reduce false-negative interpretations. With time and experience, false-positive interpretations may also be reduced (20).
The ImageChecker (R2 Technology, Inc, Los Altos, CA) was installed at the Breast Imaging Center on Ochsner's main campus as part of a multicenter study. Three comprehensive studies were performed by R2 Technology, Inc prior to FDA approval in 1998 for breast cancer screening. It is currently the only FDA approved CAD system. A prospective study of approximately 40,000 cases did not demonstrate a statistically significant increase in patient work-up rates. ImageChecker correctly identified over 80% of the cases in the second study, a multicenter retrospective evaluation of the sensitivity of the system in identifying actionable abnormalities in screening mammograms 9–24 months before the cancer was actually detected. The third study measured overall sensitivity in mammograms with biopsy-proven cancer (21). Detection of cancers may never be 100%, but assistance from devices such as ImageChecker can reduce the 10%-15% miss rate.
Refinements continue to be made in film/screen combinations, but the ability to obtain the high contrast required to resolve fine structures with only minor density differences is limited by a narrow dynamic range (exposure range). This is especially problematic for dense breast tissue, but great optimism has developed for the burgeoning field of digital mammography, which records and reproduces images separately in independently optimized processes. The dynamic range can be made quite wide to allow all areas of the breast to be displayed with ideal contrast. Work is being done with solid state image recorders (CCD arrays) that display the complete breast. Theoretical advantages include the following: minimized repeat images due to exposure errors, digital storage (time and space savings), digital data transfer, and rapid availability (22). Digital images would likely be used in conjunction with CAD. The technology has not met all expectations at this point but is promising.
The American Cancer Society, American Medical Association, and many other specialty organizations recommend annual mammography for women beginning at age 40, even if they have no specific risk factors for breast cancer (23). Screening mammography can reduce breast cancer mortality through early detection, when women are asymptomatic and cancers may still be noninvasive.
- Ochsner Clinic and Alton Ochsner Medical Foundation