Fax: (802) 656-8892
Pathologic findings from the Breast Cancer Surveillance Consortium†
Population-based outcomes in women undergoing biopsy after screening mammography
Article first published online: 12 JAN 2006
Copyright © 2006 American Cancer Society
Volume 106, Issue 4, pages 732–742, 15 February 2006
How to Cite
Weaver, D. L., Rosenberg, R. D., Barlow, W. E., Ichikawa, L., Carney, P. A., Kerlikowske, K., Buist, D. S. M., Geller, B. M., Key, C. R., Maygarden, S. J. and Ballard-Barbash, R. (2006), Pathologic findings from the Breast Cancer Surveillance Consortium. Cancer, 106: 732–742. doi: 10.1002/cncr.21652
The views expressed in this article are solely those of the authors and do not necessarily represent the official views of the NCI or the federal government.
- Issue published online: 3 FEB 2006
- Article first published online: 12 JAN 2006
- Manuscript Accepted: 1 SEP 2005
- Manuscript Revised: 1 AUG 2005
- Manuscript Received: 18 MAY 2005
- National Cancer Institute (NCI) as part of the NCI Breast Cancer Surveillance Consortium. Grant Numbers: UO1-CA70013, U01-CA86076, U01-CA63731, U01-CA86082, U01-CA63740, U01-CA69976, U01-CA70040
- breast neoplasms;
To the authors' knowledge, a comprehensive analysis of pathology outcomes after screening mammography, as it is practiced clinically in the U.S. general population, has not been performed.
Breast Cancer Surveillance Consortium data from 1996–2001 were used to identify pathology specimens that were obtained within 1 year of screening mammograms performed on 786,846 women ages 40–89 years. The pathology results were classified as invasive carcinoma, ductal carcinoma in situ (DCIS), or benign. The associations between overall pathology outcomes and invasive tumor size and lymph node status were analyzed by age and mammography assessment category.
The rates of both recommending and performing a biopsy varied little with age. The 1,664,032 screening mammograms were followed by 26,748 known biopsies (1.6%) and 8815 diagnoses of breast carcinoma (0.53%). Overall, 81% of carcinomas were invasive, and 78% of those were pathologically lymph node-negative tumors, in contrast to the 66% prevalence observed in the Surveillance, Epidemiology, and End Results (SEER) data during the same period. Most invasive tumors measured between 0 mm and 10 mm (35%) or between 11 mm and 20 mm (36%) in greatest dimension, and 92% and 78% were lymph node-negative tumors, respectively: Biopsy results that were classified as malignant increased with age (P < 0.0001) and were most likely to follow Breast Imaging, Reporting, and Diagnosis System Category 5 and 4 assessments, respectively. Ductal hyperplasia (19.6%), fibroadenoma (18.5%), and other benign findings (56.1%) were the most common benign diagnoses.
Pathologically negative lymph nodes were more prevalent in this mammographically screened population than in the overall SEER population. The prevalence of invasive carcinoma, DCIS, and benign findings presented herein establish a range of expected biopsy outcomes for women after screening mammography in the general U.S. population. Cancer 2006. © 2006 American Cancer Society.
Breast carcinoma incidence has increased dramatically over the last 25 years. When it is age-adjusted to the 2000 U.S. standard population, a relatively rapid 3.7% annual increase from 1980–1987 has been followed by a much slower increase of 0.4% per year through 2001, although the latter trend has been observed primarily in women age > 50 years.1 Despite the increase in incidence, mortality from breast carcinoma has declined 2.3% per year since 1990.1, 2 These trends have been explained largely by improvements in both detection and treatment. To further study breast carcinoma detection and outcome, the National Cancer Institute (NCI) initiated the Breast Cancer Surveillance Consortium (BCSC) in 1994. The BCSC is comprised of seven population-based mammography registries that are linked to pathology registries (five sites) and state cancer registries (all sites), including three Surveillance, Epidemiology, and End Results (SEER) programs.3 One strength of the BCSC is the availability of both benign and malignant pathology data at five of the seven sites. Cancer outcomes have been linked previously to mammography screening4–7; however, to our knowledge, little has been published to date concerning overall benign and malignant pathology outcomes within population-based samples of women undergoing screening mammography in the U.S.8
One of the major objectives of the BCSC is to report how breast carcinoma screening relates to disease stage at diagnosis. Substantial data have been collected to date regarding pathology outcomes, including disease stage at diagnosis, all of which have been collected since the initiation of the Mammography Quality Standards Act, which resulted in higher standards for mammography imaging.9
The current analysis was conducted to report the histology distribution of benign and malignant outcomes after screening mammography in a large, geographically diverse population of women. Data regarding invasive carcinomas were examined further by tumor size and lymph node status. To our knowledge, the current analysis of BCSC data is the first to establish prevalence rates for invasive carcinoma, ductal carcinoma in situ (DCIS), and benign findings in women undergoing biopsy after screening mammography in the U.S. and examines how those rates vary by age and American College of Radiologist (ACR) Breast Imaging, Reporting, and Diagnosis System (BI-RADS™) assessment category.10 To our knowledge, the current analysis presents the first report of national-level pathology data regarding the rates of invasive carcinoma within a mammographically screened population in the U.S. We also examined rates of invasive carcinoma for the same period using the SEER Program data to provide a comparison with cancer incidence among a general population of women, approximately 30% of whom, based on national data, may not have undergone screening mammography.11
MATERIALS AND METHODS
The BCSC is comprised of seven community-based mammography registries located in California (San Francisco), Colorado (Denver), New Hampshire, New Mexico, North Carolina, Vermont, and Washington (Seattle).3 Mammography data are linked with data from population-based cancer and pathology registries to establish mammography outcomes. Deidentified data are submitted annually to a Statistical Coordinating Center for research purposes. All registries have received Institutional Review Board approval and use similar quality-control and confidentiality standards.12 Only the five registries that collect both pathology data and cancer registry data were included in the current analysis. We also limited the data to women ages 40–89 years who had undergone screening mammography in the calendar period 1996–2001.
Although all mammograms (diagnostic and screening) are captured in the BCSC, for this analysis, we focused only on routine screening mammography, as indicated by the radiology facility. To eliminate potentially misclassified screening examinations, we excluded screening mammograms that followed previous imaging within 9 months, that were unilateral breast examinations, or that were from women who had breast augmentation or a previous history of breast carcinoma. Mammographic data included breast density, ACR BI-RADS interpretation category, and recommendations for further imaging or work-up. During the study period, the following ACR BI-RADS categories were included: Category 1, negative; Category 2, benign finding; Category 3, probably benign finding; Category 0, additional imaging evaluation needed; Category 4, suspicious abnormality; and Category 5, highly suggestive of malignancy.10 These categories are ordered by the degree of suspicion that malignancy may be present.
Demographic and risk factor information also was collected prospectively at the time of the screening mammogram. Typically, a woman completes a self-report questionnaire prior to each mammographic examination. The questionnaire asks about previous mammography, family history of breast carcinoma, personal history of previous breast conditions, and some demographic information. Although questionnaires vary slightly across the sites, most items have been standardized. Details of data collection may be found on the BCSC website (available at URL: http://breastscreening.cancer.gov).
Each screening mammogram was assigned an “initial” and “final” assessment based on recorded assessment, recommendations, and imaging workup subsequent to the screening mammogram. For examinations that had an initial assessment of “0,” the final assessment was the first non-0 assessment in the follow-up period. For examinations that had an initial assessment of “3” with a recommendation for immediate workup, the final assessment was the first assessment in the follow-up period that was not a “0” or “3.” Follow-up ended when no examination occurred > 90 days and ≤ 180 days after the initial screening mammogram. Nonetheless, not all Category 0 mammograms (20%) or Category 3 mammograms with a recommendation for immediate workup (70%) were resolved, and some of these initial interpretation codes remained unchanged. For all other initial assessments, the final assessment was the same as the initial assessment. We used a single assessment per mammography examination. If assessments differed between breasts, then we used the higher assessment according to the hierarchy listed earlier, ordered by the degree of suspicion that malignancy may have been present.
Pathology and Breast Carcinoma Data
Pathology and breast carcinoma outcomes were assessed within 1 year of the screening mammogram. If the woman had a new screening mammogram within that 1 year of follow-up but > 9 months after the initial screening mammogram, then the follow-up period was terminated just before the new screening mammogram. All five pathology registries collected both benign and malignant pathology, but not all benign results were recorded in the data. In particular, if more severe pathology was found, then less severe benign results may not have been recorded. Two of the sites collected International Classification of Diseases for Oncology, second revision codes directly,13 whereas the remaining three sites coded directly into the pathology categories described below. In addition, all five mammography registries were linked to either a regional SEER Program cancer registry or to a state cancer registry to identify invasive breast carcinoma or DCIS. This allowed the collection of additional information on breast carcinomas, although cancer registries would not have information regarding benign results. Pathology specimens were classified according to the following decreasing hierarchy: invasive carcinoma, DCIS, lobular carcinoma in situ (LCIS), atypical hyperplasia, ductal hyperplasia, fibroadenoma, or other benign finding. Nonepithelial malignancies (lymphoma, sarcoma) and phyllodes tumors (benign or malignant) were included in the other benign findings category. Tissue specimens that exhibited more than one pathology diagnosis were classified by the most severe pathology finding. If > 1 pathology diagnosis occurred within 1 year of the screening mammograms, then the first was used. However, if more severe pathology was found within 60 days of the original pathologic finding, then the specimen was classified into the more severe category, although it retained the date from the first pathologic finding. This follows SEER and state cancer registry data coding conventions.14 For women with invasive carcinoma, pathology data within 60 days from diagnosis was utilized to establish tumor size and lymph node status. Cancer registry data were used to supplement the pathology data if necessary. We did not attempt to standardize the reporting of pathology because the lack of standardization reflects true underlying community heterogeneity in the classification of pathologic samples, and the recorded diagnosis would have been used to determine treatment. It also is possible that not all benign diagnoses were reported when that was the only pathologic outcome. Variation of reporting across sites is discussed below.
SEER*Stat (Surveillance Research Program, NCI SEER*Stat software [version 6.1.4], Bethesda, MD; available at URL: www.seer.cancer.gov/seerstat [accessed March 15, 2005]) allows a registered user to download deidentified individual tumor data from the SEER data base. By using SEER data from the same calendar period, we computed percentages from the cross-classification of lymph node status, tumor size, and age. We also compared lymph node-negative rates within traditional categoric tumor sizes for invasive carcinomas that were detected prior to the widespread use of mammography with the BCSC rates determined from the current analysis. Two comparison series cited by the American Joint Committee on Cancer staging manual were selected.15 The first was a series of 24,749 patients who had invasive breast carcinoma with known tumor size and lymph node status: That comparison series was extracted from the population-based NCI SEER Program based on data collected from 1977–1982 (data from Table 2 in Carter et al., 198916). The second series was collected by the American College of Surgeons through a voluntary survey of 498 hospitals and included 12,981 women with known tumor size and lymph node status who were diagnosed prior to 1972 (data from Table 7 in Nemoto et al., 198017).
We identified 1,664,032 screening mammograms from 786,846 women ages 40–89 years. A single mammographic examination was most common (43% of the women), but some woman had as many as 7 screening examinations during the 6-year calendar period covered by this study. The percentages of mammograms from the 5 sites were 33.0%, 22.1%, 18.3%, 16.4%, and 10.3%, respectively; therefore, no single site dominated the analysis. The majority of the mammograms (92.7%) were from women who had undergone previous mammography.
Of the 1,664,032 screening mammograms, 26,748 women (1.61%) underwent a known breast biopsy, and 8815 women (0.53%) had a diagnosis of breast carcinoma within 1 year of the screening examination. The percentage of screening mammograms with a subsequent biopsy varied from 1.29% to 1.94% over the 5 sites, and the percentage that yielded a diagnosis of carcinoma ranged from 0.50% to 0.72%. The overall rate of breast carcinoma was 5.3 per 1000 screening mammograms. Of the 8815 breast carcinomas, 7143 (81.0%) were invasive carcinomas and the remainder were DCIS.
Not all screening mammograms that resulted in a recommendation for a subsequent surgical consultation or biopsy had an associated pathology outcome in the data base: Only 68% (range across sites, 63–78%) actually had a pathology result. The percentages of screening mammograms that resulted in a recommendation for biopsy and the percentages that resulted in women undergoing a biopsy were similar across all age groups (Table 1). Most biopsies occurred after a positive mammogram (Category 0, 4, or 5 assessment), but 34% occurred after a Category 1, 2, or 3 assessment (Table 2).
|Variable||Screening examinations||Recommendation for biopsya||Biopsy performed within 1 yrb|
|No.||Column %||No.||Column %||Percent of screenings||No.||Column %||Percent of screenings|
|Variable||Total no. of screens||Final biopsy resulta|
|Invasive||In situ (DCIS)||Benign||Total no.|
Of the 26,748 biopsies, only 194 (0.73%) could not be classified as invasive carcinoma, DCIS, or benign based on available data, leaving 26,554 biopsies for final analysis. The percentages of biopsies that were classified as carcinoma (invasive or DCIS) increased as age increased (P < 0.0001) (Table 2). For example, 18.8% of biopsies performed on women ages 40–49 years had a carcinoma outcome compared with 62.9% for women ages 80–89 years. The increase with age was much more striking for invasive carcinoma (14% for women ages 40–49 yrs compared with 55% for women ages 80–89 yrs) than for DCIS (4.6% for women ages 40–49 yrs compared with 8.1% for women ages 80–89 yrs).
The probability that a biopsy would result in a malignant outcome also was related strongly to the final assessment (Tables 2, 3). If a woman received a Category 5 assessment, then the likelihood that her biopsy would be classified as carcinoma was high, and it increased with age (Table 3). Carcinoma rates increased with age within each level of assessment; however, within age categories, the probability of a diagnosis of carcinoma after BI-RADS Category 1–4 assessments was similar.
|Final BI-RADS assessment by age groupb||Final biopsy result||Total no.|
|Invasive||In situ (DCIS)||Benigna|
|Negative (Category 1)|
|Benign finding (Category 2)|
|Probably benign finding (Category 3)|
|Suspicious abnormality (Category 4)|
|Highly suggestive of malignancy (Category 5)|
|Needs additional imaging (Category 0)|
Table 4 characterizes the benign and LCIS pathology findings by age. Fibroadenoma and ductal hyperplasia without atypia comprised 18.5% and 19.6%, respectively, of the benign findings overall. Much less common were LCIS (0.8%) and atypical hyperplasia (5.0%). The proportion of biopsies that exhibited fibroadenoma decreased with age. Fibroadenoma was more common in women ages 40–49 years compared with women age 50 years or older, whereas atypical hyperplasia was slightly less common in the youngest age group. There were no pronounced trends noted for other benign findings by age.
|Variable||Final biopsy result|
|LCIS||AH||DH||Fibroadenoma||Other benign||Total no.|
The distribution according to tumor size and the number of positive lymph nodes by age for women with invasive breast carcinoma is shown in Figure 1. Most invasive tumors measured between 0 mm and 10 mm (35%) or between 11 mm and 20 mm (36%), and 92% and 78% were lymph node negative, respectively. Of the 7143 invasive breast carcinomas, 683 (9.6%) did not have tumor size information. These included 2 patients with Paget disease, 35 patients with a mammographic diagnosis only, and 646 patients with pathologically invasive carcinoma, all without a stated tumor size in the data base. Tumors generally were smaller in women age 60 years or older.
Lymph node status by age also was examined (Fig. 2). Among the women who had invasive breast carcinoma, 639 patients (9.0%) did not have lymph nodes examined, and 339 patients (4.8%) had missing information regarding lymph node status, although it was noted in the record that their lymph nodes had been examined: This included 10 patients who had unknown numbers of positive lymph nodes. The proportion of women who had unexamined lymph nodes, unknown lymph node status, or unknown numbers of positive lymph nodes, despite the record that indicated they had been examined, increased dramatically with age; small incremental increases to approximately 18% were observed to up ages 70–79 years, after which time, a large increase to 43% was observed among women ages 80–89 years. The proportion of lymph node-negative breast carcinomas decreased after age 60 years. However, the low percentage of lymph node-negative breast carcinomas among women ages 80–89 years for the most part was because of the high percentage of women who had unexamined lymph nodes or who had lymph nodes examined but still had unknown numbers of positive lymph nodes or unknown lymph node status. No pathologic lymph node examination was recorded in 5.8%, 6.0%, 6.7%, 12.3%, and 24.1%, respectively, of women ages 40–49 years, 50–59 years, 60–69 years, 70–79 years, and 80–89 years. Among the women who had their lymph nodes examined and had known lymph node status, the overall proportion of lymph node-negative tumors was 78.2%.
The number of positive lymph nodes by tumor size was tabulated for women with invasive breast carcinoma (Table 5). The 339 women who had their lymph nodes examined but who had an unknown number of positive lymph nodes or had unknown lymph node status were not included. Large tumor size was associated with an increasing number of positive lymph nodes, as expected. No pathologic lymph node examination was recorded in 13.3%, 7.0%, 4.2%, and 6.3%, respectively, of tumors that measured 0–10 mm, 11–20 mm, 21–50 mm, and > 50 mm.
|Lymph node status|
|No. of positive lymph nodes|
|Lymph nodes not examined||Negative||1–3||4–9||≥ 10|
|Variable||No.||Row %||No.||Row %||No.||Row %||No.||Row %||No.||Row %||Total no.|
|> 50 mm||13||6.3||79||38.2||48||23.2||36||17.4||31||15.0||207|
The rates of lymph node-negative, invasive breast carcinoma were lower within the national SEER registries population (66.1%) compared with the BCSC sample of screened women (78.2%) for the same period (1996–2001). When it was examined by tumor size, this pattern was observed for all tumor size categories, with the greatest difference observed for tumors that measured > 50 mm. Figure 3 provides a comparison of BCSC rates versus SEER rates from 1996–2001 and provides a comparison of rates from two population-based cancer registry sources based on cancers that were diagnosed in the 1970s and early 1980s, before mammography screening was in wide use in the U.S. The proportions of lymph node-negative, invasive carcinomas in the historic reference populations were lower than those calculated from the SEER data from 1996–2001, when screening mammography use was more widespread.
This report provides an analysis of pathology outcomes from breast biopsies after screening mammography for women participating in the NCI BCSC, a population-based group of mammography, pathology, and cancer registries encompassing a diverse demography representative of the general U.S. population.18 The data are recent, and most were collected after implementation of the Mammography Quality Standards Act (MQSA) final regulations; therefore, the findings represent outcomes after dissemination of improved imaging and training standards within the community. The MQSA generates additional focus on mammography and mandates minimal equipment standards, strict processing standards, follow-up of mammographic outcomes, and continuing education for both radiologists and technologists.
Women ages 40–49 years and ages 50–59 years are represented approximately equally in the BCSC and, together, comprise 62% of the screened population. Because no age group exhibited a disproportionate biopsy rate relative to screening examinations performed, this suggests that mammographic features primarily are influencing biopsy recommendations and that patient age does not have a strong influence on the decision to undergo a biopsy. Overall, 32% of women who received a recommendation for biopsy had no pathology record of undergoing a core needle biopsy or an open surgical biopsy. These cases may have been resolved with surgical consultation or fine-needle aspirate biopsy. It is more likely that the biopsy was performed, and either the result was not captured by our system or it was benign and was not reported. If a woman underwent biopsy outside the BCSC registry catchments and the result was benign, then it would not be recorded. This would be much less likely with a diagnosis of breast carcinoma, because carcinoma diagnoses are captured through SEER and/or state cancer registries. Therefore, we may have underestimated the biopsy rate and overestimated the carcinoma yield per biopsy by not capturing biopsies performed outside the catchments.
Patient age was expected to be a major predictor of biopsy outcome and, in contrast to the observed biopsy rates, carcinoma outcomes (either invasive or DCIS) were more likely with increasing age. Less predictable was the carcinoma outcome, or malignant yield, stratified by mammography interpretation as measured by the ACR BI-RADS assessment categories. Age-associated carcinoma prevalence again was observed across all BI-RADS assessment categories.19 Category 5 (highly suggestive of malignancy) and Category 4 (suspicious abnormality) routinely lead to biopsy and imply a higher expectation for a carcinoma outcome. The highest prevalence of breast carcinoma was associated with Category 5 mammogram assessments, as expected. Although Category 4 assessments (suspicious for malignancy) may be expected to have a higher prevalence of breast carcinoma compared with Category 1 (normal), Category 2 (benign finding), or Category 3 (probably benign) assessments, this was not observed among women who did undergo biopsy. Biopsy after a Category 3 assessment (probably benign finding) generally would be expected to confirm a benign lesion, such as fibroadenoma. Category 2 (benign finding) and Category 1 (negative) assessments would not be expected to generate biopsies. Women who receive a Category 4 assessment usually would undergo an immediate biopsy, whereas women who receive Category 1, 2, or 3 assessments would not undergo biopsy unless they were symptomatic at the time of the screening examination, or if symptoms developed that led to biopsy during the year after mammography, or if a change in lesion size was observed on short-interval follow-up mammography. This can alter the general expectation that the percentage of biopsies that are malignant will increase with the BI-RADS assessment category. This explanation also may account for women with Category 0 (needs additional assessment) mammogram interpretations. The majority of Category 0 mammographic abnormalities are resolved and reclassified as normal or benign after additional imaging. However, symptomatic women may undergo biopsy prior to additional imaging. In women age > 60 years, the prevalence of cancer after a Category 0 mammogram assessment is exceeded only by the prevalence after a Category 5 mammogram assessment.
When a diagnosis of breast carcinoma was established by biopsy, invasive disease was much more common than DCIS, and this ratio increased with age and varied by mammogram assessment category. The greatest proportion of invasive carcinoma was observed in biopsies after Category 5 assessments, and the greatest proportion of DCIS was observed after Category 4 assessments. Biopsies after Category 1 or 2 assessments occurred late in the follow-up period (data not presented), suggesting that symptoms prompted the biopsy. The proportional distribution of invasive carcinoma and DCIS after a Category 1 or 2 assessment was found to be similar to that after a Category 5 assessment.
The proportion of invasive tumors for which information regarding size was missing was approximately 10% and demonstrated little variation by age, suggesting incomplete reporting to cancer registries or incomplete size notations on pathology reports. Further research on the accuracy of data reported to cancer registries may be needed. The proportion of tumors that measured 11–20 mm also did not appear to vary by age. Women age > 60 years were more likely to have tumors that measured 0–10 mm and were less likely to have tumors that measured > 20 mm. In part, this may be because of less dense breasts and more easily detected tumors, but it also may be because of biologically less aggressive, slower growing tumors in older women.4, 20, 21
Previously published reports using data from surgical and SEER populations established prevalence rates for lymph node metastases for different sizes of breast carcinoma in the premammography screening era.16, 17 SEER data and other cancer registry data can be used to estimate the proportion of lymph node-negative, invasive breast carcinoma by tumor size but cannot differentiate rates in screened versus unscreened populations. Therefore, to our knowledge, the prevalence of lymph node-negative disease and lymph node-positive disease stratified by tumor size has not been established among women who participate in screening mammography in the U.S. The current analysis from the BCSC has established population-based prevalence rates of lymph node metastases stratified by primary tumor size and age. For women who participated in mammography screening, the overall lymph node-negative rate, regardless of tumor size, was 78%. For the same period, 1996–2001, the SEER lymph node-negative rate was 66%. The rate of lymph node-negative, invasive disease was greater in a screened population, as anticipated.22 The SEER Program does not collect information regarding mammography, a factor that contributed to the formation of the NCI BCSC. The SEER data include women who participate and do not participate in breast carcinoma screening; for women who do participate in screening mammography, SEER includes tumors that are detected after screening and diagnostic breast imaging. This is in contrast to analyses from much earlier periods that demonstrated lower lymph node-negative rates. A SEER series of 24,749 women collected from 1977 to 1982, a time when screening mammography was very uncommon, demonstrated a 54.4% lymph node-negative rate.16 Another premammography series collected by the American College of Surgeons demonstrated a 57.1% lymph node-negative rate, which is similar to the older SEER rate.17 Both the BCSC data and the current SEER analysis data, which were collected from 1996 to 2001, represent a more recent, population-based sample and reflect the influence of screening mammography.
In the current analysis, we also documented the prevalence of omitting a pathologic lymph node examination, a more common practice in the era of mammography screening.22 There was a consistent, increasing trend by age toward omitting lymph node evaluation for invasive tumors in older women; 76% of women ages 80–89 years and 88% of women ages 70–79 years underwent pathologic lymph node examination compared with 93–94% of women younger than age 70 years. There also was a trend toward omitting the examination of lymph nodes when tumors were small. Overall, pathologic lymph node examination was common but was performed least frequently in tumors that measured ≤ 10 mm, in which the rate was 87%, compared with the greatest frequency of 96% for tumors that measured 21–50 mm. In the prescreening era, tumor size was correlated highly with the presence of axillary metastases,16 and this pattern remains today in the analyses we performed with the 1996–2001 SEER data. We suspect that community practitioners are not offering or women are declining surgical excision and pathologic examination of lymph nodes, particularly when the likelihood of metastases is low, as it is in small tumors, or when the documentation of pathologic lymph node disease may not influence therapeutic decisions, as in many older women. The advent of axillary sampling techniques with reputedly lower morbidity, such as sentinel lymph node biopsy, that began to enter clinical practice after 1993, may affect future and recent past trends.
The categoric distribution of benign findings in women who did not have invasive carcinoma or DCIS was remarkably similar for all age ranges, with exception to the greater prevalence of fibroadenoma in women ages 40–49 years. This is a perplexing finding. It may be hypothesized that, as women age and the incidence of breast carcinoma increases, there would be a greater proportion of pathologic diagnoses that have been associated with the future risk of cancer. Specifically, these would include LCIS, atypical ductal hyperplasia, and ductal epithelial hyperplasia (hyperplasia of usual type). However, circulating levels of estradiol are higher in younger women and diminish with age. Therefore, conversely, it may be hypothesized that as women age and estradiol levels decrease and breast epithelium atrophies, there would be a lower proportion of pathologic diagnoses that have been associated with the future risk of carcinoma. However, the distribution of lesions associated with the risk of future breast carcinoma was nearly identical in all age groups. Several hypothetical explanations exist. If these lesions are precursors, then their rates of appearance and progression to malignancy may be equal, which may mask detectable age associations. If these lesions are only indicators of risk, then once they are present, they may be stable, persist indefinitely, and not undergo atrophy such as normal epithelium in postmenopausal women, or they may appear spontaneously and regress in all women at similar rates. Finally, other circulating factors or factors within the microenvironment of the breast may initiate or promote epithelial proliferation independent of circulating estradiol, confounding any age association. It should be noted that the distribution of risk-associated lesions within the current series is biased, because the women underwent both screening and biopsy, and the distribution may not be the same in women who present with a breast lump and undergo biopsy without prior screening. It has been documented that women who participate in recommended screening report greater compliance with recommended health behaviors, but they also are more likely to have a family history of breast carcinoma.23–25 Therefore, it is not clear whether screened women would be more or less likely to be at risk for breast carcinoma compared with other women in the same age group.
Issues that were not considered for this report include differences in outcomes between women who underwent previous mammography and women who did not. In the group with prior mammography, it has been reported that tumor size depends on time since the last examination.7 We also did not consider racial differences in cancer outcomes. Finally, we did not consider differences in tumors detected by screening mammography versus interval tumors that occur in the first year of follow-up. All these issues are important but were outside the scope of the current study, which highlighted the pathology outcomes of screening mammography by age and BI-RADS assessment.
The current analysis of population-based pathology outcomes in women who underwent biopsy within 1 year of screening mammography confirmed the general association of age with cancer prevalence and the high prevalence of cancer after Category 5 mammograms. In general, the probability of invasive carcinoma was greater than the probability of DCIS, with the highest prevalence of DCIS after Category 4 mammograms and the highest prevalence of invasive carcinoma after Category 5 mammograms. Pathologically negative lymph nodes were more prevalent in this screened population across all size categories of invasive tumors than in previous population-based series, in which screening was uncommon, and in the general population of screened and unscreened women, as assessed in our analysis of SEER data from the same period. The biopsy prevalence and rates for invasive carcinoma, DCIS, and benign findings presented here establish a range of expected biopsy outcomes after screening mammography in the general U.S. population.
- 2American Cancer Society. Cancer facts and figures 2004 [monograph online]. Available at URL: http://www.cancer.org/downloads/STT/CAFF_finalPWSecured.pdf [accessed April 28, 2005].
- 10Illustrated breast imaging reporting and data system: illustrated BI-RADS™, 3rd ed. Reston, VA: American College of Radiology, 1998., , , et al.
- 13International classification of diseases for oncology, 2nd ed. Geneva: World Health Organization, 1990., , .
- 14The SEER Program code manual, 3rd ed. Bethesda, MD: Surveillance Program, Division of Cancer Control and Population Sciences, National Cancer Institute, National Institutes of Health, 1998., .
- 15Breast. In: GreeneFL, PageDL, FlemingID, et al., editors. AJCC cancer staging manual, 6th ed. New York: Springer-Verlag, 2002: 223–240., , , et al.