| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Present address: Kaiser Permanente Division of Research, 2000 Broadway,
Oakland, CA 94612.
2 Pharmaceutical Outcome Research and Policy Program, University of Washington
School of Pharmacy, Center for Cost & Outcomes Research, Seattle, WA
98195.
3 Department of Radiology, University of Washington Medical Center, Seattle
Cancer Care Alliance, Seattle WA 98195.
4 Center for Health Studies, Group Health Cooperative of Puget Sound, Seattle WA
98101.
5 Department of Epidemiology, University of Washington School of Public Health
and Community Medicine, Seattle, WA 98101.
6 Cancer Prevention Research Program, Fred Hutchinson Cancer Research Center,
Seattle, WA 98109-1024.
Received February 13, 2004;
accepted after revision June 30, 2004.
Address correspondence to N. Oestreicher
(Nina.Oestreicher{at}kp.org).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. We prospectively followed 61,688 women aged 40 years or older who had undergone at least one screening examination with mammography and CBE between January 1, 1996, and December 31, 2000, for 1 year after their mammogram for invasive cancer. We computed the incremental sensitivity, specificity, and positive predictive value of CBE over mammography alone for combinations of age and breast density (predominantly fatty or dense).
RESULTS. Mammography sensitivity was 78% and combined mammography–CBE sensitivity was 82%, thus CBE detected an additional 4% of invasive cancers. CBE detected a minority of invasive cancers compared with mammography for all age groups and all breast densities. Sensitivity increased from adding CBE to screening mammography for all ages, from 6.8% in women ages 50–59 with dense breasts to 1.8% in women ages 60–69 years with fatty breasts. CBE generally added incrementally more to sensitivity among women with dense breasts. Specificity and positive predictive value declined when CBE was used in conjunction with mammography, and this decrement was more pronounced in women with dense breasts.
CONCLUSION. CBE had modest incremental benefit to invasive cancer detection over mammography alone in a screening program, but also led to greater risk of false-positive results. These risks and benefits were greater in women with dense breasts. The balance of risks and benefits must be weighed carefully when evaluating the inclusion of CBE in a screening examination.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
There have been few investigations of factors that affect the performance of CBE with and without mammography, relative to mammography alone, when evaluated in the same group of women [5, 8]. Comparisons between the relative performance of CBE and mammography are difficult to interpret when the screening techniques are performed in different study populations. Nonetheless, as single screening techniques, it is known that the performance of CBE and mammography is influenced by patient age and breast density [9–14]: The sensitivity of mammography is somewhat diminished in young women and in women with dense breasts [9–12, 14], whereas CBE performs better in these groups [13]. However, to our knowledge, the incremental performance of CBE over mammography alone and the performance of CBE relative to mammography have not been studied by examining age and density concurrently in a single study population.
We report the results of a prospective study in which mammography and CBE were used to screen a large population of women in the context of a formal screening program. Our purposes were to examine the influence of age and breast density on the potential added contribution of CBE to invasive cancer detection over mammography alone and the performance of CBE relative to mammography in invasive cancer detection. Performance measures included sensitivity, specificity, and predictive value of a positive examination.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
During the study period, women received a two-view mammogram and a CBE at each screening examination. Registered nurses (RNs) perform CBEs in the screening centers after being mentored and precepted by an experienced nurse to ensure the accuracy of their CBE skills. This training is a comprehensive seven-step process that takes from 80 to 120 hr to complete. In addition to the onsite training, RNs take the American Cancer Society CBE Certification Training. CBEs were performed in a separate room from the mammography and, in most cases, after the mammogram. The screening mammograms were batch interpreted subsequently, and therefore the mammography results usually were not available to the nurse who performed the CBE. Any clinical concerns arising from the CBE were discussed with the radiologist, referring provider, or both. We applied for and were granted permission from the Institutional Review Board of GHC to obtain and analyze these data.
Study Population
Women were eligible for study if they were enrolled in the BCSP and had at
least one screening examination including mammography and CBE (index screening
examination) between January 1, 1996, and December 31, 2000. Women with breast
implants or a history of breast cancer were ineligible for study. Eligible
women were followed up for 12 months after their index screening examination
and before their next BCSP visit for diagnosis of invasive breast cancer. We
identified women with invasive breast cancer by linking the BCSP database with
the Seattle–Puget Sound Surveillance, Epidemiology and End-Results
(SEER) cancer registry, a population-based registry covering the 13 counties
of western Washington State. We restricted the study to women who remained
continuously enrolled at GHC for at least 12 months following their index
screening examination or who had died from any cause during the 12-month
period after the index screening examination. This restriction allowed us to
follow women for a full year to identify breast cancer occurrence that might
have been missed by screening, in order to inform our sensitivity,
specificity, and positive predictive value calculations. The exception that
allows inclusion in the study for women who died during that year is needed
because women with breast cancer might not have survived the full year.
A total of 62,467 women (653 women with invasive breast cancer, 61,814 women without breast cancer) met the eligibility requirements. We excluded 291 women because they were missing a mammographic assessment, 139 women because they were missing a CBE assessment, and 349 women because they were missing information on breast density. Thus, data from 61,688 women, 574 with invasive breast cancer and 61,114 without breast cancer, were available for analysis.
Data Collection
Mammographic assessments were made based on the six-category BI-RADS
criteria developed by the American College of Radiology
[15]. Mammograms with a
BI-RADS assessment of 1 (normal), 2 (normal, benign findings), or 3 (probably
benign) were defined as negative. We considered a BI-RADS result of 4
(suspiciously abnormal) or 5 (highly suggestive of malignancy), which are
recommendations for biopsy, to be positive. We followed women forward who
received an assessment of BI-RADS 0 or whose assessment was missing on the
mammogram from their index screening examination and obtained the assessment
from their next mammogram with a non-zero assessment within 12 months after
their index screening examination.
The assessments for CBE were either negative or indeterminate. Since 1988, BCSP nurses have been required to call nonnegative findings indeterminate. We interpreted an indeterminate finding as a positive assessment.
GHC radiologists recorded mammographic breast density at the time of reading each mammogram at the index screening examination. Density was coded using the four-category BI-RADS terminology [15]. This system provides density ratings for each breast based on the breast composition of fat and fibroglandular tissue. We defined dense breasts as ratings of heterogeneously dense or extremely dense, and we defined predominantly fatty breasts as ratings of almost entirely fat or scattered fibroglandular densities. For women with breast cancer, we used the density rating from the contralateral breast. For women without breast cancer, if the densities in the right and left breast were equal, we used this density. We used the density value for a randomly selected breast laterality if the densities in the right and left breasts were not equal (n = 130).
Screening and Cancer Outcomes
For mammographic sensitivity calculations, subjects were women with
invasive breast cancer diagnosed within 12 months of their index screening
examination (n = 574). These subjects comprised the denominator of
the sensitivity calculation. For these women, the index screening examination
was the screening examination before diagnosis. For mammographic specificity
calculations, subjects were women without a diagnosis of breast cancer within
12 months of their index mammogram (n = 61,114). These subjects
comprised the denominator of the specificity calculation. For these women, the
index screening examination was identified from a randomly selected screening
mammogram during the study period. For positive predictive value calculations,
subjects were women with a positive assessment on CBE (n = 1,454) or
mammography (n = 1,013).
We examined the performance of mammography alone (i.e., regardless of CBE result), CBE alone, and mammography and CBE together for sensitivity, specificity, and positive predictive value calculations. For the performance of mammography and CBE together, a positive examination was defined as positive on either technique and a negative examination was therefore negative on both. Thus, for sensitivity, we evaluated the percentage of women with invasive cancer who had positive results on mammography, CBE, and mammography or CBE. For specificity, we evaluated the percentage of women without breast cancer who had negative results on mammography, CBE, and mammography and CBE. For positive predictive value, we evaluated the percentage of women with invasive cancer among women with a positive mammographic assessment, a positive CBE assessment, and a positive mammographic or CBE assessment. The incremental effect of CBE on sensitivity, specificity, and predictive value was the difference between the performance of mammography and CBE combined versus mammography alone. We categorized women as those with positive findings on CBE alone (i.e., positive on CBE and negative on mammography) versus all others to determine whether age or breast density influenced the incremental effect of CBE over mammography alone.
Statistical Analyses
We used unconditional logistic regression analysis to analyze the
association of age and breast density with the likelihood of detection by CBE
alone, separately for women with and without breast cancer. We categorized age
into five 10-year groups and breast density into two groups (heterogeneously
dense and extremely dense vs mostly fat and scattered fibroglandular
densities). Age and density were both included in the same model—that
is, each was controlled for the other, and the interaction between them (as a
single continuous interaction term) was tested.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
|
Specificity was 99% for mammography, 98% for CBE, and 97% for both techniques (Table 2). Mammographic specificity was higher than CBE specificity for every category of age and breast density. This difference was greater in women with dense breasts. The addition of CBE to mammography screening reduced the specificity for every category of age and breast density by 1–3%, and this reduction was greater among women with dense breasts (2–3%) than fatty breasts (1–2%). Women with dense breasts were nearly twice as likely to receive a false-positive result from CBE alone (adjusted OR = 1.85; 95% CI, 1.40–2.46), and this result was highly significant (p < 0.001). There was not a statistically significant effect of age or interaction between age and density.
Positive predictive value was 44% for mammography, 8% for CBE, and 20% for the two techniques combined. Positive predictive values were higher for mammography than CBE for every category of age and breast density. The addition of CBE to mammography screening reduced the positive predictive value for every category of age and breast density by 8% to 35%, and this reduction was greater among women with dense breasts (16–35%) than those with fatty breasts (8–24%).
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Our results suggest that mammography yields a far greater benefit to invasive cancer detection than CBE, even in the group of women in which mammographic sensitivity tended to be somewhat diminished: young women with dense breast tissue, where mammographic sensitivity was still nearly threefold higher. Our results also suggest CBE has the greatest incremental benefit in women ages 50–59 years with dense breasts and tended to make a greater contribution to sensitivity in women with dense breasts. These findings may have been due to the relatively poor performance of mammography and relatively improved performance of CBE in women with dense breasts [12, 13]; however, even in these groups, the effect of CBE alone is small (5–7%). Our estimates of CBE benefit appear to be lower than those of Kolb et al. [8], who found CBE alone detected an additional 2–15% more tumors in predominantly fatty breasts and extremely dense breasts, respectively. Trials and demonstration projects [6, 16–19] have assessed the proportion of cancers detected by CBE alone among cancer detected by screening and have found incremental benefits in the range of 3% [18] to 82% [6], although several of these reports were based on early and less sensitive mammographic technology [6, 16, 17].
CBE also had a greater detrimental effect on specificity and positive predictive value in women with dense breasts and, therefore, modest benefits to sensitivity must be weighed against these potential costs of including CBE as a screening technique. These decrements ranged from a modest 1–3% for specificity, but were more dramatic (8–35%) for predictive value. Women who benefit most from increased tumor ascertainment when CBE is added to a screening program—that is, women with dense breasts—also are those who experience the most harm from CBE in terms of more biopsies due to false-positive results and the poorer predictive value of abnormal findings in the screening examination. Because of adding CBE to a screening program, these women may experience more of the negative effects of screening, such as the additional morbidity and anxiety from additional diagnostic evaluations.
Both the U.S. Preventive Services Task Force [7] and the American Cancer Society [20] recommend that clinicians inform their patients of the potential risks and benefits of including CBE in their screening examination. With this information, a patient and her physician may carefully weigh the trade-offs between sensitivity (increased breast cancer ascertainment) and specificity (increased number of women with false-positive results sent to biopsy) based on their risk status and preferences for the risks and benefits associated with CBE screening. These trade-offs differ depending upon a patient's breast density, and potentially, age, so that CBE can be targeted to the groups in which it has the highest net benefit. Furthermore, even more caution is required because there currently is no evidence to suggest the ultimate value of the early detection of cancers by CBE alone on breast cancer mortality.
Our analysis of women ages 40–49 years was limited to those women at high risk of breast cancer, because only high-risk women in their 40s were invited to the screening program. Therefore, our results may not be generalizable to normal-risk younger women. It is difficult to determine how screening performance might differ between high- and normal-risk women, as there is a limited literature available to answer this question [21–26]. Some of these studies suggest that mammography may be slightly less sensitive in women with a family history of breast cancer, compared with those without [23–25], although a recent study suggested that women with a family history were more likely to have their cancer detected by screening in general (mammography and CBE) [26], although the study was conducted in women age 50 years and older. In terms of the incremental value of CBE over mammography alone, studies have not been in agreement about the probability of breast cancer detection by mammography, as compared with CBE in women with a family history, with one indicating a higher likelihood of mammographic detection [22] and the other a lower likelihood [27], although these studies were not conducted within a screening program. Not surprisingly, due to the increased prevalence of breast cancer in those women with a family history, the positive predictive value of mammography was higher in this group [24, 26].
The sensitivity of mammography may be overestimated, and the incremental value of CBE may be underestimated in this study because the CBE results were sometimes available to the radiologists who interpreted the mammograms. It is also possible that in our study there were differences in CBE training and expertise between nurses and physicians, which may have influenced CBE performance. However, the Canadian National Breast Screening Study did not find significant differences in the performance of their physician-versus nurse-examiners [28]. In addition, some analyses were based on small samples and may have led to unstable estimates of mammography or CBE performance. For example, there were only 15 high-risk women ages 40–49 years with predominantly fatty breasts. Small group size also may have affected positive predictive value calculations, although results of increasing values with age were generally in agreement with the expectations of increasing values as the age-specific prevalence of invasive breast cancer increases.
Several aspects of the study setting were advantageous. The use of an organized, population-based screening program, combined with follow-up of women for 1 year after their screening examination by linkage to a cancer registry, allowed near complete ascertainment of breast cancer cases not detected by CBE or mammography from which to calculate false-negative rates. This database included information on over 130,000 mammograms and subsequent cancers, allowing stratification by age and breast density, and reflected the performance of CBE and mammography in a naturalistic practice setting.
Our findings suggest that women may receive different amounts of benefit from having CBE included in a mammography screening program, depending upon their age and breast density. There was no group of women in our study population for whom CBE detection approached that of mammography and, overall, the added benefit was 4%. Given the small added benefit of CBE, it is unlikely CBE added to screening programs would result in a decrease in mortality rates. Nevertheless, other benefits of CBE are possible, such as increased awareness of early breast cancer detection and increased compliance with screening mammography recommendations. The potential and real benefits of CBE should be reviewed and balanced with the possible harms associated with false-positive examinations.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Robson and K. Offit Management of an Inherited Predisposition to Breast Cancer N. Engl. J. Med., July 12, 2007; 357(2): 154 - 162. [Full Text] [PDF]
|
|
|
|
 |
|
 A. A. J. Roelofs, N. Karssemeijer, N. Wedekind, C. Beck, S. van Woudenberg, P. R. Snoeren, J. H. C. L. Hendriks, M. Rosselli del Turco, N. Bjurstam, H. Junkermann, et al. Importance of Comparison of Current and Prior Mammograms in Breast Cancer Screening Radiology, January 1, 2007; 242(1): 70 - 77. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. N. Feigin, D. M. Keating, P. M. Telford, and M. A. Cohen Clinical Breast Examination in a Comprehensive Breast Cancer Screening Program: Contribution and Cost Radiology, September 1, 2006; 240(3): 650 - 655. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
