Routine mammography is associated with earlier stage disease and greater eligibility for breast conservation in breast carcinoma patients age 40 years and older†
The following are the Clinical Staff of the Breast Evaluation Center at the Fox Chase Cancer Center. Department of Medical Oncology: Lori J. Goldstein, M.D. (Director), Michael M. Millenson, M.D., Mary E. Cianfrocca, D.O., and Margaret von Mehren, M.D.; Department of Radiation Oncology: Gary M. Freedman, M.D. (Associate Director), Penny R. Anderson, M.D., and Nicos Nicolaou, M.D.; Department of Surgical Oncology: Marcia C. Boraas, M.D., F.A.C.S., Michael H. Torosian, M.D., Elin R. Sigurdson, M.D., Ph.D., and John P. Hoffman, M.D., F.A.C.S.; Department of Diagnostic Imaging: Kathryn Evers, M.D. and Arthur D. Magilner, M.D.; and Department of Pathology: Arthur S. Patchefsky, M.D., Harry S. Cooper, M.D., Tahseen I. Al-Saleem, M.D., Hormoz Ehya, M.D., Nancy A. Young, M.D., and Hong Wu, M.D., Ph.D.
Reduction in breast carcinoma mortality is a major benefit of screening mammography and has been demonstrated in multiple randomized clinical trials and service screening programs. Another benefit from screening is that it allows the patient a wider choice of treatment options, particularly the possibility of conservation surgery. The current study analyzed the impact of mammography in the staging and treatment of breast carcinoma.
A total of 1591 women aged ≥ 40 years were treated for breast carcinoma between July 1995 and October 2001. Three subgroups were defined and compared. Group 1 had 192 patients with no previous mammography, Group 2 was comprised of 695 patients who underwent mammography on average less often than once yearly, and Group 3 was comprised of 704 patients who on average underwent mammography once yearly or more often.
The difference in tumor stage was found to be statistically significant between the groups (P < 0.0001). In Group 1, 15% of the patients had ductal carcinoma in situ (DCIS) compared with 21% of patients in Group 2 and 26% of patients in Group 3. In addition, 32% of patients in Group 1 had T1 tumors, whereas 50% of patients in Group 2 and 56% of patients in Group 3 had T1 tumors. The tumor size was ≤ 1 cm in 8% of the patients in Group 1 compared with 20–23% of patients in Groups 2 and 3 (P = 0.0092). Breast conservation was an option for 41% of the patients in Group 1 but mastectomy was recommended in another 41% of patients. However, in Groups 2 and 3, 61% of patients were offered breast conservation and mastectomy was recommended to 28% (P < 0.0001).
In the current study, women age ≥ 40 years with breast carcinoma who underwent mammography at least once yearly were diagnosed with DCIS more often compared with patients who underwent mammography less frequently or those who had no prior mammography. Women who underwent mammo-graphic screening were found to have smaller tumors, which resulted in a majority of these patients being able to consider breast conservation as an alternative to mastectomy. Cancer 2003;98:918–25. © 2003 American Cancer Society.
Routine mammography is widely accepted in the U.S. for breast carcinoma screening. Mammography is recommended for women ages ≥ 40 years by governmental organizations,1, 2 medical societies,3–5 and patient advocacy groups.6, 7 These recommendations are based in large part on prospective randomized trials that tested the screening of asymptomatic women with mammography.8–12 The five Swedish trials, the Edinburgh Scotland, and HIP trials8–10 have proven a benefit from screening mammography. Although to our knowledge the two Canadian trials11, 12 have not yet demonstrated benefit, their negative findings can be explained on the basis of poor technical quality and problems with the study design.13–16 The reported magnitude of reduction in breast carcinoma mortality in the positive screening trials is 20–30%. Considerable publicity and controversy has been created by recent reanalysis of the randomized trials that cast doubt on the strength of evidence for reducing breast carcinoma mortality by essentially disqualifying much of the data based on perceived flaws in the quality of the trials.17, 18 However, the methodologies of the Olsen and Gøtzsche critique have themselves been criticized.19, 20 Further evidence of a reduction in mortality with mammographic screening has been found in meta-analyses of the randomized trials, multiple organized screening programs, and comprehensive reviews of the literature.21–26
The purpose of the current study was to evaluate the benefits associated with mammography screening in a large population of women with newly diagnosed breast carcinoma. The frequency of mammography screening prior to a diagnosis of breast carcinoma was analyzed for an association with the method of cancer detection, tumor stage and tumor size at the time of presentation, clinical lymph node status, and physician recommendation for local treatment of the breast by conservation therapy or mastectomy. The question of whether there is a difference between annual and less frequent screening was examined within these endpoints and, in addition, subgroup analysis was performed for women ages 40–49 years.
MATERIALS AND METHODS
The study population was comprised of 1591 women age ≥ 40 years with newly diagnosed breast carcinoma who were seen in consultation in the multidisciplinary Breast Evaluation Center (BEC) at the Fox Chase Cancer Center between July 1995 and October 2001. The major eligibility criterion for the study population was information from the patient by self-report using questionnaire or physician history regarding the number and/or frequency of mammograms prior to the diagnosis of breast carcinoma. Patients seen in the BEC during this period for a diagnosis of benign breast conditions, lobular carcinoma in situ, nonepithelial breast malignancies, or ipsilateral breast tumor recurrence after previous breast carcinoma treatment were excluded. The BEC offers a multidisciplinary consultation to patients with biopsy proven breast carcinoma. Each patient is evaluated separately with a detailed history and physical examination by a radiation oncologist, medical oncologist, and breast surgeon. A radiologist with expertise in mammography reviews the mammograms and other pertinent diagnostic imaging studies, and the slides from biopsies and/or definitive surgeries from each patient also are reviewed with a pathologist. At the conclusion of the 4-hour consultation, the patient has detailed information regarding her disease and a coherent multidisciplinary plan for additional testing and treatment for her breast carcinoma. Patient information from the BEC is collected prospectively by a single independent data manager from questionnaires completed by the patients, outside records, reports from the clinical staff consultations, and reviews of the outside imaging studies and pathology. Data are coded and analyzed by study number and individual patients remain anonymous.
For each patient with a history of prior mammography screening, a mammogram frequency index (MFI) was assigned. This index was created to measure the approximate frequency of screening for those patients with a history of mammography prior to diagnosis. The MFI was determined by a formula (patient age–40)/number of prior mammograms. The patients were divided into 3 separate groups based on their MFI. Group 1 was comprised of 192 patients with no mammography prior to the time of diagnosis so that a MFI would be incalculable (i.e., division by 0). Group 2 was comprised of 695 patients with an MFI > 1. This subgroup of patients represented a partially screened population who underwent mammography on average less often than once yearly prior to their diagnosis of breast carcinoma. Group 3 was comprised of 704 patients with an MFI ≤ 1. This subgroup represented more highly screened patients with an average frequency of mammography of once yearly or more often. Patients who stated their frequency of prior mammography as yearly without giving a specific number of prior mammograms also were included in Group 3.
Differences in percentages were compared for the groups of patients using the chi-square statistic. Distributions were compared for method of detection, tumor stage, tumor size, lymph node status, and the recommendation of the multidisciplinary team for treatment of the involved breast. For nearly all patients, the recommendation for breast-conserving surgery included a recommendation for postoperative radiation.
The characteristics of the 1591 women who met the eligibility requirements for the study population are shown in Table 1. The median age of the patients was 54 years. The method of breast carcinoma detection was physical examination alone (patient and/or physician) in 9% of patients and mammography in 91%. The presentation was mammography alone in 51% of patients and both mammography and physical examination in 40% of patients. For these 40% of patients presenting with both findings, a physical finding first led to the mammography in approximately 80% of cases. Therefore, in approximately 59% of women, the mammography was the first method of detection. The vast majority of these patients presented with early-stage disease. The AJCC tumor stage was classified as Tis (ductal carcinoma in situ [DCIS]) in 22% of patients, T1 in 50% of patients, and T2 in 19% of patients. Tumor size was ≤ 1 cm in 20% of patients, ≤ 2 cm in 50% of patients, and ≤ 5 cm in 73% of patients. The clinical lymph node status was negative in 92% of patients and positive in 8% of patients. A consensus for recommending treatment by mastectomy with or without radiation was made in 30% of patients. Breast-conserving surgery, usually with postoperative radiation, was offered to 58% of patients. The remaining 12% of patients received recommendations to begin with nonsurgical treatment, or to undergo additional evaluation without a final recommendation given on that day.
Table 1. Characteristics of Study and Nonstudy Patients
|No. of patients||1591||409|| |
| Age range (yrs)||40–92||40–86|| |
| Median age (yrs)||54||56|| |
|Method of detection (%)|| || ||< 0.0001|
| Physical examination||9.0||12.2|| |
| Mammography only||51.2||33.5|| |
| Both||39.7||54.0|| |
|Tumor classification (%)|| || ||< 0.0001|
| Tis||22.2||11.5|| |
| T1||50.3||50.1|| |
| T2||18.7||23.7|| |
| T3||4.0||8.6|| |
| T4||4.3||5.9|| |
|Size in cm (%)|| || ||0.9147|
| 0 ≤ 1||20.1||20.3|| |
| > 1 ≤ 2||29.7||28.1|| |
| > 2 ≤ 5||22.9||24.2|| |
| > 5||27.3||27.4|| |
|Clinical lymph node status (%)|| || ||0.0002|
| N0||92.1||85.6|| |
| N1||7.3||13.4|| |
| N2||0.6||1.0|| |
|Treatment recommended (%)|| || ||0.1421|
| Breast conservation||58.1||53.5|| |
| Mastectomy||30.2||35.2|| |
| Other||11.7||11.2|| |
Also shown in Table 1 are the characteristics of 409 women seen in the BEC during the same study period who were excluded from the study population only because of missing data regarding their mammography history. Fewer of these nonstudy patients had mammography only as the method of detection compared with study patients (34% vs. 51%; P < 0.0001). In addition, fewer presented with noninvasive stage disease compared with study patients (12% vs. 22%; P < 0.0001). In addition, more nonstudy patients presented with clinically positive axillary lymph nodes (13% vs. 7%; P = 0.0002). These numbers for the nonstudy patients are most comparable to the unscreened patients in Group 1 of the study population, which suggests that the nonstudy patients were relatively unscreened or less screened than the overall study population. The two groups were similar with regard to the distribution of other examined factors in Table 1.
Of the study population, 192 patients had no prior history of screening mammography before diagnosis; these patients comprise Group 1. Of the 1399 women with a history of prior mammography, 431 stated a frequency of mammography of once yearly and therefore were placed in Group 3. Of the remaining 968 women, the number of reported prior mammographies ranged from 2–54 with a median of 5. Based on their MFI, 695 of these patients (MFI > 1) comprised Group 2 and 273 of these patients (MFI ≤ 1) were placed into Group 3. The three groups of patients defined by mammography screening history and their characteristics are shown in Table 2.
Table 2. Characteristics of All Patients by Mammogram Screening History
|No. of patients||192||695||704|| |
| Age range (yrs)||40–92||43–90||40–90|| |
| Median age (yrs)||57||56||53|| |
|Method of detection (%)|| || || ||< 0.0001|
| Physical examination||11.5||6.8||10.5|| |
| Mammography only||32.8||48.5||58.8|| |
| Both||55.7||44.5||30.7|| |
|Tumor classification (%)|| || || ||< 0.0001|
| Tis||15.1||20.6||25.7|| |
| T1||31.8||50.1||55.7|| |
| T2||29.7||20.9||13.6|| |
| T3||8.9||4.0||2.6|| |
| T4||14.6||3.6||2.3|| |
|Size (cm) (%)|| || || ||< 0.0001|
| 0 ≤ 1||8.3||20.3||23.2|| |
| >1 ≤ 2||21.9||30.4||31.1|| |
| > 2 ≤ 5||37.5||24.5||17.3|| |
| > 5||32.3||24.9||28.4|| |
|Clinical lymph node status (%)|| || || ||< 0.0001|
| N0||82.8||92.5||94.3|| |
| N1||15.1||7.1||5.4|| |
| N2||2.1||0.4||0.3|| |
|Treatment recommended (%)|| || || ||< 0.0001|
| Breast conservation||40.6||60.1||60.9|| |
| Mastectomy||41.2||29.9||27.4|| |
| Other||18.2||9.9||11.7|| |
Table 2 shows that there were statistically significant differences between Groups 1, 2, and 3 for all study endpoints. The method of tumor detection was by mammogram only in 33% of Group 1 patients, 49% of Group 2 patients, and 59% of Group 3 patients (P < 0.0001). Approximately 15% of the patients in Group 1 presented with DCIS (Stage 0), compared with 21% of the patients in Group 2 and 26% of the patients in Group 3. Patients with early-stage T1 disease comprised 32% of Group 1 patients compared with 50% of patients in Group 2 and 56% of patients in Group 3. There was a decrease in T2 tumors (30% in Group 1, 21% in Group 2, and 14% in Group 3) and a decrease in locally advanced T3/T4 tumors (24%% in Group 1, 8% in Group 2, and 5% in Group 3) in those groups with a greater frequency of mammography screening. The stage distribution between Groups 1, 2, and 3 was highly statistically significant (P < 0.0001). The tumor size was ≤ 1 cm in 8% of Group 1 patients compared with 22% of patients in Groups 2 and 3. There was an increase in the frequency of tumor size > 1 cm and ≤ 2 cm from 22% in Group 1 to approximately 30% in Groups 2 and 3, and corresponding decreases in the frequency of tumor sizes > 2 cm was noted in the screened groups as well. The distribution of tumor sizes was found to be highly statistically significant between all three groups (P < 0.0001).
The clinical lymph node status was negative in 83% of Group 1 patients (positive in 17%) compared with 93% lymph node-negative patients in both Groups 2 and 3 (7% lymph node-positive patients) (P < 0.0001). For Group 1 patients, the multidisciplinary team offered the option of breast-conserving surgery and radiation to 41%, and recommended mastectomy to 41% of patients. For patients in Groups 2 and 3, 61% were eligible for breast conservation and only 27–29% were recommended to undergo a mastectomy. The difference between Group 1 and Groups 2 and 3 was highly statistically significant (P < 0.0001).
There were statistically significant differences between the less-screened Group 2 patients and the more-screened Group 3 patients with regard to method of detection (P < 0.0001), tumor stage (P = 0.0002), and tumor size (P = 0.0092). However, the characteristics of clinical lymph node status (P = 0.7379) and treatment recommendation (P = 0.4137) were not found to be statistically significantly different between patients in Groups 2 and 3.
The characteristics and recommendations for those patients ages 40–49 years in the 3 groups based on mammography screening are shown in Table 3. For this subgroup of patients, statistically significant differences remained with regard to the distribution of tumor stage (P = 0.0193) and tumor size (P = 0.0333). In the most-screened patients (Group 3), 80% had early-stage Tis or T1 disease (30% with DCIS and 50% with T1 disease) compared with 68% of patients in Group 2 and 60% of patients in Group 1. Group 3 patients also were found to have the highest percentage of tumors measuring ≤ 1 cm. The distributions between Group 1, Group 2, and Group 3 patients were not found to be significantly different for method of detection, lymph node status, and treatment recommendation. Similarly, the distributions of Group 2 compared with Group 3 were not found to be statistically significantly different for any of the study endpoints in this subgroup.
Table 3. Characteristics of Patients Ages 40–49 Years by Mammogram Screening History
|No. of patients||72||161||274|| |
|Method of detection (%)|| || || ||0.2572|
| Physical examination||11.1||9.9||12.0|| |
| Mammography only||38.9||39.1||47.5|| |
| Both||50.0||50.9||40.5|| |
|Tumor classification (%)|| || || ||0.0193|
| Tis||26.4||23.6||30.0|| |
| T1||33.3||44.7||50.0|| |
| T2||27.8||23.0||15.0|| |
| T3||4.2||5.0||2.2|| |
| T4||8.3||3.7||2.9|| |
|Size (cm) (%)|| || || ||0.0333|
| 0 ≤ 1||6.9||14.9||18.3|| |
| > 1 ≤ 2||23.6||30.4||30.3|| |
| > 2 ≤ 5||31.9||24.2||16.8|| |
| > 5||37.5||30.4||34.7|| |
|Clinical lymph node status (%)|| || || ||0.7007|
| N0||87.5||90.1||92.0|| |
| N1||12.5||9.3||7.7|| |
| N2||0||0.6||0.4|| |
|Treatment recommended (%)|| || || ||0.4205|
| Breast conservation||45.8||53.4||58.0|| |
| Mastectomy||43.1||36.0||31.8|| |
| Other||11.1||10.6||10.2|| |
The study population was comprised of 1591 women with newly diagnosed breast carcinoma with prospective collection of their data during the study period. The patients were divided into 3 study cohorts based on their frequency of screening: no prior mammography in Group 1, less frequent (> 1 mammogram per year) mammography in Group 2, and more frequent (< 1 mammogram per year) mammography in Group 3. Although the patients were not prospectively randomized, during their lifetime patients were essentially dividing themselves up into one of our three cohorts based on factors other than any potential bias of the researchers. During the study time period, it should be assumed that most or all of these women had the opportunity for screening. The current study differs from the majority of the randomized trials that analyzed patients based on those offered mammography versus control groups, but could not correct for compliance issues or cross-contamination of the control group with screening. However, the major weakness of the current analysis is reliance on self-reporting of mammogram frequency. There was the possibility of study bias in excluding some patients during the study period from the study population, but we believe this did not negatively affect the study or its conclusions. The analysis of those patients who were excluded because of a lack of screening history suggests that this group was screened either relatively less or not at all because of similarities to Group 1 with regard to studied characteristics (Tables 1 and 2). This may have been because of many patients leaving the space for information concerning prior mammography blank, causing an absence of data, when actually they intended an answer of no prior mammograms. Their inclusion would likely have strengthened, not weakened, the study conclusions.
The results of the current study confirm a significant downstaging of tumor stage and tumor size associated with mammography screening that has been shown in other recent studies as well.27, 28 There was a highly statistically significant increase in the frequency of DCIS among women who underwent mammography screening with an average frequency of at least yearly. There also was a significant increase in the percentage of T1 tumors and tumors measuring ≤ 1 cm associated with frequent mammography screening. This tumor stage and size downstaging is likely responsible for the increased recommendation for breast conservation reported in women with a history of mammography screening. This also is consistent with the observed trend for the increasing use of breast conservation in the U.S., particularly for women with AJCC Stage I breast carcinoma.29 A tumor size ≤ 1 cm also may be a significant breakpoint for the need for adjuvant systemic therapy as part of the initial therapy for breast carcinoma for many patients.3, 30 These potential differences in treatment when associated with screening mammography could result in additional psychologic benefits to the patients.31, 32 Improvement in the extent or cost of initial treatment and types of initial treatment, as well as the cost-effectiveness of screening mammography, are other potential benefits that need to be confirmed in future research.
The screening interval used in the majority of randomized prospective trials of mammography screening was 12–33 months. Most randomized trials were generally not designed to provide direct evidence regarding the effect of screening frequency on mortality decrease, but several modeling analyses based on trial data as well as tumor growth rate studies have suggested that more frequent screening should be more effective.33–37 One randomized trial directly randomized patients to a mammography screening interval of 1 year versus an interval of 3 years after an initial prevalence screen.38 Women with shorter screening intervals were found to have a significant trend toward small tumor size, but the decrease in mortality from breast carcinoma that was observed with greater screening (5–11% depending on the predictive model) did not reach statistical significance. The interval between screening mammograms recommended for asymptomatic women is 1–2 years by some authors,1, 2 but annually by the majority of other organizations.3–7 In the current study, for the endpoints of method of detection, tumor size, and tumor stage, there was a significant advantage to annual or more frequent screening mammography compared with less frequent screening.
Outcomes data were not available from the patients in the current study population at the time of last follow-up. However, based on their differences in tumor characteristics at the time of presentation, it would be reasonable to expect that long-term differences in disease-free and overall survival would become apparent between the cohorts. There was a greater percentage of patients with noninvasive breast carcinoma in the cohorts of patients with more mammography screening. The cause-specific survival after treatment of DCIS was > 95% long-term and was considerably better than that noted with invasive disease.39–44 Moreover, there is evidence that many or most cases of DCIS have the potential to develop into invasive carcinoma.44, 45 There also were higher percentages of patients with T1 tumors, tumors measuring ≤ 1 cm, and clinically negative axillary lymph nodes in those patients who underwent mammography screening, which also is associated with a better long-term prognosis than more advanced invasive disease.30, 46 There has been an observed decrease in the mortality rates associated with breast carcinoma in the U.S. when the periods 1974–1976 and 1992–1997 are compared.47 To our knowledge, it is unknown to what degree this is because of a shift toward the disease being diagnosed at an earlier stage as a result of screening versus changes in outcome because of improved treatment.
The use of screening mammography is most controversial for women ages 40–49 years. A benefit of mammography screening in women ages 40–49 years has been shown in the trials from Malmö, Sweden,8 Gothenburg, Sweden,8 and the HIP trials.10 A recent meta-analyses of trial data22 and long-term results from the Swedish two-county service screening program48 also demonstrated a significant benefit for these younger women. To our knowledge, only the Canadian NBSS-1 study did not demonstrate a benefit for screening mammography in these women.12 An important mortality benefit from screening may exist in this subgroup of patients, but could be of lesser magnitude than in older women. Although screening mammography may be less sensitive in younger women compared with older women, the sensitivity for detecting breast carcinoma in women ages 40–49 years has been shown to be in the range of 67–88%.28 The relative incidence of breast carcinoma is lower for women age < 50 years, so that larger numbers of asymptomatic women with long-term follow-up would be needed to detect any differences in cancer detection and outcome. In the current study, approximately 39–48% of women ages 40–49 years had mammography as the sole method of detection of their tumor, and in approximately 90% of patients the tumor was detectable on mammography even if preceded by other symptoms. In this younger age subgroup, similar to the overall study population, patients in combined Groups 2 and 3 (those with any history of prior mammography) had statistically significant differences with regard to the distribution of tumor stage and tumor size compared with Group 1 patients. We were unable to discern a benefit for more frequent versus less frequent screening in this subgroup. In addition, there were similar trends but no statistically significant differences observed for clinical lymph node status and recommendations for local treatment. This is either because of a lack of real differences between the groups or a lack of sufficient numbers of patients in this subgroup. The study authors and the majority of organizations in the U.S. continue to recommend screening mammography for this subgroup of patients ages 40–49 years.1–7
In the current study, women ages ≥ 40 years who were diagnosed with breast carcinoma and who had a history of mammogram frequency of at least yearly were diagnosed with DCIS more often than women who underwent less frequent screening or no prior mammography. Women who underwent mammography screening prior to the time of diagnosis also presented with earlier stages of invasive tumors and smaller tumor sizes. There was a less frequent recommendation for mastectomy and an increased recommendation for breast conservation in women with prior mammogram screening as well. In the subgroup of women ages 40–49 years, there were similar trends observed for earlier tumor stages and smaller tumor size that reached statistical significance. The results of the current study support many of the benefits associated with mammographic screening for women age ≥ 40 years, and support an at least yearly interval versus more than yearly for the optimal detection of the most curable forms of breast carcinoma.
The authors thank Cindy Rosser for the data management required for the study population and Katherine Farlow for her assistance in the preparation of the article.