SEARCH

SEARCH BY CITATION

Keywords:

  • mammography;
  • breast carcinoma;
  • axillary lymph nodes;
  • metastasis;
  • Vermont

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

BACKGROUND

A trend toward earlier breast carcinoma detection in the United States has been attributed to screening mammography, although direct evidence linking this trend to the increased use of mammography in a general population is lacking. This study examined the effects of mammography on tumor size and axillary lymph node metastasis in Vermont over 25 years.

METHODS

Pathology and mammography data from 3499 Vermont women who were diagnosed with invasive breast carcinoma during 1975–1984, 1989–1990, and 1995–1999 were compared. Logistic regression analysis was used to estimate the effects of age, mammography use, and period on the odds of a tumor ≤ 2 cm and the odds of negative lymph nodes.

RESULTS

The proportion of breast tumors that were detected by screening mammography increased from 2% during 1974–1984 to 36% during 1995–1999 (P < 0.001), and these tumors were more likely to measure ≤ 2 cm than tumors that were detected by other methods. Among women age > 50 years, the odds ratio (OR) was 4.5, with a 95% confidence interval (95% CI) of 3.5–6.4. The effect was smaller in younger women (OR, 1.8; 95% CI, 1.1–3.0). Mammographic detection increased the odds of negative lymph nodes by a similar amount in both age groups, although women age > 50 years were more likely to have negative lymph nodes than younger women (OR, 1.3; 95% CI, 1.1–1.6). Tumor size and lymph node metastasis also were related to the number of mammograms and to the mammographic interval.

CONCLUSIONS

Most of the trend toward earlier detection in Vermont was due to mammography. Mammography had a lesser effect on tumor size among younger women, which may be related to less frequent screening, although its effect on lymph node metastasis was not age dependent. Women age < 50 years were more likely to have positive lymph nodes, independent of the method of detection or the frequency of mammography. Cancer 2002;94:2160–8. © 2002 American Cancer Society.

DOI 10.1002/cncr.10459

Despite increases in the incidence of invasive breast carcinoma from 1973 to 1987, mortality rates have decreased steadily since 1989.1–3 This decline has been attributed to earlier detection due to the use of screening mammography as well as advances in treatment. Since 1980, the incidence of localized breast tumors has increased, whereas the incidence of regional disease has decreased.3 This would be expected to start reducing mortality in the following 5–10 years because of the lower mortality associated with localized disease.4, 5 The assertion that increased use of screening mammography in the general population has been responsible for this shift toward earlier detection of invasive breast carcinoma is supported by a marked increase over the same period in the incidence of in situ tumors, which almost always are detected by mammography and, thus, are indicative of mammography use.3

More direct evidence that the use of screening mammography has led to earlier breast carcinoma diagnosis has been provided by studies of the pathologic characteristics of breast tumors detected by different methods. Several hospital-based and practice-based studies have found that invasive tumors detected by mammography tend to be smaller and are less likely to be accompanied by positive axillary lymph nodes than tumors that are found by clinical examination or self-detection.6–9 Similar results relating disease stage to the regular use of mammography were obtained in a small study of a general population10 and in a larger population-based study of women age ≥ 65 years.11 However, both of those studies grouped in situ carcinoma with localized invasive carcinoma.

Breast cancer mortality rates in Vermont have paralleled the national trend, showing a steady decline from 1989 to 1996.12 A study of breast carcinoma detected in Vermont during 1975–1984 and 1989–1990 found a shift toward earlier stages at detection in the latter period.13 Information about the mode of detection enabled the investigators to determine that this was due to mammographically detected tumors; for nonmammographically detected breast carcinoma, there was no difference in maximum tumor size or axillary lymph node metastasis between the two study periods. In the current investigation, pathology and mammography data on invasive breast carcinoma detected in Vermont during 1995–1999 were combined with the earlier data to document the changes in tumor size and axillary lymph node metastasis that have occurred over the past 25 years and to determine the extent to which the use of mammography has contributed to these changes.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

The study population included all women resident in Vermont who had a pathologic diagnosis of invasive breast carcinoma at Vermont hospitals during three periods: July 1, 1975 through December 31, 1984; January 1, 1989 through December 31, 1990; and January 1, 1995 through December 31, 1999. Thus, these periods covered 16.5 years of 25.0 years from 1975 to 1999. In the earlier studies, patients with breast carcinoma who were diagnosed in Vermont during 1975–1984 and during 1989–1990 were identified by searching all pathology logs as well as any tumor registries in all nonfederal general hospitals in the state of Vermont.13, 14 Clinical and pathologic data were then abstracted from patient charts, operative notes, pathology records, and radiology reports. Women who were diagnosed during 1995–1999 were identified using data routinely collected in the Vermont Breast Cancer Surveillance System (VBCSS).15 All pathology facilities in Vermont send breast and associated axillary lymph node pathology reports to the VBCSS. The VBCSS also collects information on every woman who receives a mammogram in Vermont. Women undergoing mammography are asked to complete a questionnaire requesting health history information relevant to breast cancer, including previous mammograms. The radiology staff provides information about the mammogram, and the completed form is sent to the VBCSS. Ninety-five women (5%) who objected to the use of their data for research by checking a box on the questionnaire were excluded from the study. This study was approved by the Institutional Review Board at the University of Vermont with an alteration of informed consent according to guidelines set forth in the federal regulations governing research on human subjects.

Data from all three periods included maximum tumor dimension and information about axillary lymph node metastasis. For women who were diagnosed during 1975–1984, clinical and radiology records were used to determine the method by which the breast tumor was detected, but no additional information about mammography was recorded. More information about mammography was available on women who were diagnosed during the latter two periods and had a mammogram prior to diagnosis, including the date, results, and reason for the mammogram. These women were considered to have had their breast tumors detected by screening mammography if their most recent mammogram was less than 12 months prior to diagnosis; was a routine screening mammogram; and was positive, suspicious, or indeterminant with follow-up recommended. If the reason for the mammogram was unknown, then the method of detection also was considered unknown unless the mammogram had a negative or benign result and, thus, did not detect the tumor. The VBCSS did not collect information routinely about the reason for the mammogram until 1996, and the method of detection could not be determined for 20% of women in the 1995–1999 period, compared with 5% of women in the 1989–1990 period.

If a woman had a mammogram prior to diagnosis, then there was additional information about whether or not there had been a previous mammogram and when it had been done. For women who were diagnosed during 1989–1990, data about previous mammography were abstracted from clinical and radiologic records; whereas, for women who were diagnosed during 1995–1999, data were obtained from the questionnaire that was completed at the time of the last mammogram prior to diagnosis and, thus, were self-reported. This information was used to classify women according to the number of mammograms they had and to determine the interval between the last two mammograms for those women who had multiple mammograms. Women who had a mammogram prior to diagnosis and had no evidence of a previous mammogram were classified with one mammogram, and all women with information about a previous mammogram were classified with at least two mammograms.

Statistical Analysis

An analysis of variance was used to test for the effects of period and mammography use, as well as their interaction, on tumor size at diagnosis. Multiple pair-wise comparisons were tested at P ≤ 0.05 using the Student–Newman–Keuls procedure. Trends over the three periods were tested by linear contrasts. The proportions of women with tumors in specific size categories and the proportions of women with axillary lymph node metastasis were analyzed by chi-square tests using the Cochran–Mantel–Haenzsel test for linear association to assess trends over time and differing categories of mammography use. Logistic regression was used to estimate the independent effects of age, mammography, and period on the odds of having a tumor measuring ≤ 2 cm and the odds of not having axillary lymph node metastasis. Age was categorized as < 50 years or ≥ 50 years, because screening recommendations have differed for these age groups. Models that included interaction terms were fitted to test whether the effect of mammography use differed by age group or period. All tests were two sided and were considered statistically significant at the P ≤ 0.05 level.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

The numbers of women diagnosed with in situ and invasive breast tumors in Vermont during each of the data collection periods are shown in Table 1. Patient ascertainment appears to have been incomplete for the 9.5 years comprising the period 1975–1984, because there was an average of only 157 patients per year compared with 276 and 298 patients per year during 1989–1990 and 1995–1999, respectively. The proportion of in situ tumors increased from 2.3% in the earliest period to 8.0% and 19.5% in the latter two periods, respectively (P < 0.001). In situ tumors were not included in the current study, but they serve as an indicator of the increasing use of mammography.

Table 1. In Situ and Invasive Carcinomas Diagnosed in Vermont 1975–1999
Carcinoma1975–19841989–19901995–1999
No.%No.%No.%
In situ352.3488.036119.5
Invasive145997.755192.0148980.5

Data on invasive breast tumors are summarized in Table 2. During 1975–1984, only 27 women (2%) had invasive tumors that were detected by screening mammography. This increased significantly to 19% during 1989–1990 and to 36% during 1995–1999 (P < 0.001). Details of mammography use were not available in the 1975–1984 data, but usage increased significantly between the latter two periods (P < 0.001). During 1989–1990, only one-third of women had two or more mammograms prior to diagnosis, whereas 78% of women who were diagnosed during 1995–1999 had two or more mammograms. Among women with multiple mammograms, the majority had their two most recent mammograms within 2 years.

Table 2. Summary Data on Women with Invasive Breast Carcinoma
Characteristic1975–19841989–19901995–1999
No.%No.%No.%
  1. SD: standard deviation.

Age (yrs)
 < 5030020.712622.939226.3
 50–6964944.821739.562542.0
 ≥ 7050034.520737.647231.7
Mammography detected
 No143298.142881.476664.0
 Yes271.99818.643036.0
No. of mammograms
 07113.01047.4
 129153.220614.7
 ≥ 218533.8109477.9
Time between mammograms (yrs)
 ≤ 212470.167664.4
 2–54424.927726.4
 > 595.1969.2
Tumor size (cm)
 ≤ 258447.330661.988365.5
 2–558447.316834.039729.4
 > 5665.3204.0695.1
 Mean (SD)2.6 (1.6)2.1 (1.5)2.0 (1.5)
 Median2.31.81.6
Axillary lymph node resection
 No28219.59417.335323.9
 Yes116180.544982.7112676.1
Axillary lymph node metastasis
 No62453.728764.174666.3
 Yes53746.316135.938033.7

Tumor size at diagnosis decreased significantly over the three periods (P < 0.001), with the largest decrease occurring between 1975–1984 and 1989–1990. The mean tumor size during 1975–1984 (2.6 cm) was significantly larger (P < 0.05) than during either 1989–1990 (2.1 cm) or 1995–1999 (2.0 cm). The latter two mean sizes did not differ significantly. A similar, significant trend was seen for the proportion of women who had axillary lymph node metastasis (P < 0.001). Of the women who underwent lymph node resection, 46% of women had positive lymph nodes during 1975–1984, compared with 36% of women during 1989–1999 and 34% of women during 1995–1999. Fewer women underwent lymph node resection during 1995–1999 (76%) than during 1989–1990 (83%) and 1975–1984 (81%). This was due to a sharp decrease in lymph node resection among women age > 70 years, from 72% in 1989–1990 to 51% in 1995–1999. Among women age < 70 years, the proportions of women who underwent lymph node resection were similar in all three periods (data not shown).

In Table 3, tumor size and axillary lymph node metastasis for patients with invasive breast tumors that were detected by screening mammography are compared with tumors that were not detected by mammography. Within each period, the tumors that were detected by screening mammography were significantly smaller (P < 0.001) and were less likely to involve axillary lymph node metastasis (P = 0.002) compared with the tumors that were detected by some other method. A two-way analysis of variance of tumor size indicated a significant interaction between period and detection method (P = 0.032), reflecting the smaller difference in size between tumors detected and not detected by screening mammography during 1995–1999 than during the earlier periods.

Table 3. Pathologic Findings by Method of Detection
FindingsDetected by screening mammography
1975–19841989–19901995–1999
NoYesNoYesNoYes
  1. SD: standard deviation.

Tumor size (no.)12112338190688407
 ≤ 2 cm46.4%95.7%54.6%88.9%57.7%81.1%
 2–5 cm48.1%4.3%40.7%10.0%34.9%16.2%
 > 5 cm5.5%0.0%4.7%1.1%7.4%2.7%
 Mean (SD)2.7 (1.6)1.2 (0.6)2.3 (1.5)1.2 (0.8)2.3 (1.7)1.6 (1.2)
 Median2.31.12.01.01.81.3
Axillary lymph node resection (no.)11382333991579328
Lymph node metastasis (%)46.913.041.616.539.923.8

To determine whether the increased use of diagnostic imaging also may be contributing to earlier detection during the two later periods, women with clinically detected or self-detected tumors were grouped according to whether or not they had nonscreening mammography or sonography prior to diagnosis. There were no significant differences in tumor size or axillary lymph node metastasis between women who had diagnostic imaging and those who had no imaging prior to diagnosis during either 1989–1990 or 1995–1999 (data not shown). This information was not available for the period 1975–1984.

The proportion of tumors that measured ≤ 2 cm and the proportion of tumors with no axillary lymph node metastasis were analyzed by logistic regression to adjust for the effects of age. The odds ratios (OR) and 95% confidence intervals (95% CI) shown in Table 4 are from a multivariate model that included patient age (< 50 years vs. ≥ 50 years), method of detection, period, and the interaction between age and detection method. There were no significant interactions between period and detection method after adjustment for age. The odds of having a small (≤ 2 cm) tumor did not differ by age group in women with tumors that were not detected by mammography. Detection by screening mammography significantly increased the odds of a small tumor, but the effect was much greater in women age ≥ 50 years (OR, 4.47; 95% CI, 3.53–6.41; relative to women in the same age group who did not have tumors that were detected with a mammogram) compared with younger women (OR, 1.79; 95% CI, 1.08–2.96). In addition, the odds of a small tumor were significantly greater during the later two periods compared with 1975–1984, independent of age and detection method.

Table 4. Multivariate Odds Ratios for Early Breast Carcinoma Detection
ParameterTumor ≤ 2 cmNegative lymph nodes
OR95% CIOR95% CI
  1. OR: odds ratio; 95% CI: 95% confidence interval.

Age in yrs (screen detected)
 < 50 (no)1.00Referent1.00Referent
 < 50 (yes)1.791.08–2.962.081.25–3.47
 ≥ 50 (no)0.940.78–1.141.321.09–1.61
 ≥ 50 (yes)4.473.24–6.163.352.45–4.58
Period
 1975–19841.00Referent1.00Referent
 1989–19901.451.16–1.811.311.03–1.65
 1995–19991.481.23–1.781.311.07–1.59

The odds of having no axillary lymph node metastasis also were significantly greater when tumors were detected by screening mammography, but there was no significant interaction between detection method and age. The OR among women age < 50 years (2.08; 95% CI, 1.25–3.47) was similar to that for older women (OR, 2.53; 95% CI, 1.88–3.40) relative to women in the same age group who did not have tumors that were screen detected. Also, older women had significantly higher odds of negative axillary lymph nodes compared with younger women, regardless of the method of detection. The overall OR for women age ≥ 50 years, independent of the method of detection, was 1.31 (95% CI, 1.09–1.57) from the regression model without an interaction term. Like tumor size, after adjustment for age and detection method, the odds of having negative lymph nodes differed significantly between periods, with similar ORs for 1989–1990 and 1995–1999 relative to 1975–1984.

Associations of mammography use with tumor size and metastasis are shown in Tables 5–7. In Table 5, women with breast carcinoma who were diagnosed during 1989–1990 and during 1995–1999 were grouped according to the number of mammograms they had: 0, 1, or ≥ 2. The mean tumor size decreased significantly over these three categories during both periods (P < 0.001). The proportion of women with axillary lymph node metastasis also decreased as the number of mammograms increased, although this trend was statistically significant only during the 1995–1999 period (P = 0.016). Logistic regression analysis indicated that the number of mammograms had a greater effect on the odds of a small tumor among women age ≥ 50 years compared with younger women (Table 6). For this analysis, the women who had multiple mammograms (two or more) were compared with women who had no or only one mammogram. Among women who had one or no mammogram, the odds of having a small tumor did not differ between the two age groups, but there was a significant interaction between the effects of having multiple mammograms and age group. For women age < 50 years, the odds of having a tumor measuring ≤ 2 cm was 1.62 times higher (95% CI, 1.10–2.40) in women who had multiple mammograms compared with women who had no or one mammogram; whereas, among older women, this OR was 3.30 (95% CI, 2.52–4.32). Women who were diagnosed during 1995–1999 had a significantly lower odds of having a small tumor compared with women who were diagnosed during 1989–1990, after adjustment for the effects of age and number of mammograms.

Table 5. Pathologic Findings by Number of Mammograms
FindingsNo. of mammograms (1989–1990)No. of mammograms (1995–1999)
01≥ 201≥ 2
  1. SD: standard deviation.

Tumor size (no.)53270168831891001
 ≤ 2 cm47.2%58.1%73.2%42.2%45.0%70.5%
 2–5 cm43.4%37.8%24.4%45.8%48.1%25.1%
 > 5 cm9.4%4.1%2.4%12.0%6.9%4.4%
 Mean (SD)2.6 (1.7)2.2 (1.5)1.9 (1.3)2.8 (1.7)2.5 (1.5)1.9 (1.5)
 Median2.32.01.72.52.31.5
Axillary lymph node resection (no.)4124316441150873
Lymph node metastasis (%)41.538.730.548.838.032.4
Table 6. Associations between Number of Mammograms and Early Detection
VariableTumor ≤ 2 cmNegative lymph nodes
OR95% CIOR95% CI
  1. OR: odds ratio; 95% CI: 95% confidence interval.

Age in yrs (no. of mammograms)
 < 50 (0 or 1)1.00Referent1.00Referent
 < 50 (≥ 2)1.621.10–2.401.180.79–1.75
 ≥ 50 (0 or 1)0.830.58–1.181.360.92–2.01
 ≥ 50 (≥ 2)2.731.94–3.841.991.40–2.83
Period
 1989–19901.00Referent1.00Referent
 1995–19990.700.55–0.900.970.75–1.27
Table 7. Pathologic Findings by Time Between Mammograms: 1995–1999 Data
FindingsYears between mammograms
≤ 22–5> 5
  1. SD: standard deviation.

Tumor size (no.)61525590
 ≤ 2 cm74.5%65.5%58.9%
 2–5 cm22.1%29.4%31.1%
 > 5 cm3.4%5.1%10.0%
 Mean (SD)1.8 (1.4)2.1 (1.7)2.3 (1.6)
 Median1.41.61.8
Axillary lymph node resection (no.)53522874
Lymph node metastasis (%)28.637.339.2

The odds of not having axillary lymph node metastasis were significantly higher for older women and for women who had multiple mammograms, but there was no significant interaction between these effects. For women who had two or more mammograms, the OR of having negative lymph nodes was 1.18 (95% CI, 0.79–1.75) among younger women and 1.46 (95% CI, 1.07–2.00) among older women, relative to women in the same age group who had fewer mammograms. The overall ORs were 1.35 (95% CI, 1.05–1.74) for women who had at least two mammograms and 1.57 (95% CI, 1.24–1.97) for women age ≥ 50 years from the model without an interaction term (data not shown). After adjustment for age and number of mammograms, there was no significant effect of period on the odds of negative lymph nodes.

Women who were diagnosed during 1995–1999 and who had at least two mammograms were grouped according to the interval of time between their last two mammograms prior to diagnosis: ≤ 2 years, 2–5 years, and > 5 years. Women who were diagnosed during 1989–1990 were not included, because only one-third had multiple mammograms, and most of these women had their last two mammograms 2 years apart or less. Tumor size and metastasis data for each of the three categories are shown in Table 7. The mean tumor size increased significantly as the time between mammograms increased (P < 0.001), and a similar trend was seen for the proportion of women with axillary lymph node metastasis (P = 0.009). Logistic regression analysis indicated that the odds of having a small tumor (≤ 2 cm) were higher for women whose last two mammograms were within 2 years of each other compared with women who had a longer interval between mammograms. This effect did not differ significantly between age groups, with an overall OR of 1.59 (95% CI, 1.19–2.12) independent of age, but older women had higher odds of a small tumor (OR, 1.65; 95% CI, 1.20–2.26) independent of the interval between their mammograms. Very similar results were seen for the odds of not having regional metastasis. The OR associated with having the last two mammograms within 2 years of each other was 1.44 (95% CI, 1.07–1.95), and the OR associated with age ≥ 50 years was 1.63 (95% CI, 1.18–2.25).

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

The results of this study clearly show that invasive breast tumors that were detected by screening mammography were smaller in size and were less likely to be accompanied by axillary lymph node metastasis compared with tumors that were not detected by mammography. However, increases in the proportion of tumors detected by screening mammography did not appear to account fully for the time trend observed for tumor size and axillary lymph node metastasis. Tumors that were not detected by mammography tended to be larger and were more likely to have axillary lymph node metastasis if they were diagnosed during 1975–1984 rather than during 1989–1990 or 1995–1999. This is consistent with a continuing trend toward earlier detection between 1939 and 1988 reported by Cady et al.16 It may reflect a heightened awareness of breast carcinoma and the benefits of earlier detection by women and clinicians with a consequent increase in clinical breast carcinoma detection.

There are potential biases in the data that may account for at least some of the differences between periods. Incomplete ascertainment in 1975–1984 may have skewed the data if advanced tumors were more likely to be included than early-stage tumors. However, comparison with Surveillance, Epidemiology, and End Results (SEER) data from the same period does not confirm this.17 The proportion of women who had local disease (no axillary lymph node metastasis) was significantly higher (P < 0.001) in the Vermont data (53.7%) compared with the SEER data (50.0%). A smaller, nonsignificant difference (P = 0.28) was seen for later data, with 65.6% of women in the 1989–1999 Vermont data having local disease compared with 64.3% of women in the 1989–1996 SEER data.1 The 1975–1984 Vermont data are very similar to data from 719 tumors that were diagnosed at New England Deaconess Hospital during 1974–1983.16 The mean tumor size for those tumors was 2.6 cm, the same as the Vermont data; however, 52.5% of tumors from the New England Deaconess Hospital study had negative axillary lymph nodes compared with 53.7% of the Vermont tumors. These comparisons suggest that any ascertainment bias in the Vermont data during 1974–1984 is likely to be small.

Another possible source of bias is the information used to classify tumors that were detected by screening mammography. For women who were diagnosed during 1995–1999, the reason for mammography given by the radiology facility was used to determine whether the mammogram done prior to diagnosis was for routine screening. For women who were diagnosed during 1989–1990, both clinical and radiology records were used to determine the reason, so the data from the two periods are not completely comparable. Also, women sometimes report a breast problem when they come in for a routine screening mammogram, and the reason for the visit may be recorded as evaluation of a breast problem. This type of misclassification, which may have occurred during both periods, would lead to an underestimation of tumor size and metastasis for nonscreen-detected tumors, and the effect of screening mammography, thus, would appear to be lower than it actually was. During 1975–1984, the method of detection was determined by review of clinical and radiology records, but the reason for the mammogram and other details were not recorded. Even if the same criteria were used to classify a tumor as detected by mammography, far fewer women received mammography during 1975–1984 and, hence, there would be less opportunity for misclassification.

After controlling for age and method of detection, there was no significant difference in either tumor size or axillary lymph node metastasis between the 1989–1990 and 1995–1999 periods. However, after adjustment for age and number of mammograms, women who were diagnosed during 1995–1999 were less likely to have small tumors compared with women who were diagnosed during 1989–1990 (Table 5). The reason for this is unclear, but the self-reported data about previous mammography for women who were diagnosed during 1995–1999 may not be as accurate as the information for women who were diagnosed during 1989–1990, which was obtained from radiology and clinical records. The VBCSS began collecting mammography data on January 1, 1994, and it may have been used to obtain more accurate information about previous mammography for women who were diagnosed during the latter part of 1995–1999, but this information would be incomplete for women who were diagnosed earlier and for women who recently moved to Vermont. It has been shown that self-report of whether or not a woman had a mammogram has good accuracy;18, 19 thus, to avoid inconsistencies, self-report was used to determine whether a women had more than one mammogram. However, women tend to recall the date of their last mammogram less accurately,20 which may introduce error when computing the time between their last two mammograms.

This study provided evidence that women who received mammograms more frequently had smaller tumors and less axillary lymph node metastasis. The number of mammograms (0, 1, or ≥ 2) and the time between the two most recent mammograms (≤ 2 years, 2–5 years, or > 5 years) were used as indicators of the frequency of mammography, but they are not very precise markers of screening frequency. Although recalls for additional views were not considered as separate mammograms, women may have had multiple diagnostic mammograms prior to diagnosis or may have received their second mammogram because short-term follow-up was recommended after the earlier mammogram. To determine the number of interval tumors in women undergoing screening and to evaluate the effect of screening interval on tumor size and regional metastasis, both the reason for the previous mammogram and the reason for the mammogram prior to diagnosis need to be taken into account. This information is available in the VBCSS for women who had both their mammograms in Vermont since 1996, and, as data collection continues, it will be possible to define screening intervals more accurately and to assess the efficacy of screening in future studies.

The results shown in Tables 4 and 6 indicate that mammography was more effective in detecting small tumors among women age ≥ 50 years compared with younger women. This finding is consistent with studies showing that mammography had lower sensitivity among women age 40–49 years.21 However, it also may reflect less frequent use of mammography among women age < 50 years, resulting in a higher proportion of prevalent tumors and incident tumors with longer screening intervals for those women who have their tumors detected by screening mammography. This is supported by the analysis of data from women who had at least two mammograms: the increased odds of having a small tumor when a woman's last two mammograms were no more than 2 years apart did not differ significantly between the two age groups.

The effectiveness of mammography in reducing the odds of regional metastasis was similar among women age < 50 years and women age ≥ 50 years. The risk of developing regional metastasis increases with tumor size;5, 22, 23 thus, if screening mammography is more effective in detecting small tumors among older women, then it also may be expected to be more effective in reducing the occurrence of regional metastasis in older women. However, women age ≥ 50 years were less likely to have positive lymph nodes than younger women, attenuating the impact of mammography on lymph node metastasis among women in this age group.

Higher rates of regional metastasis among younger women have been observed in other studies23 and did not appear to be due to tumor size, histopathologic classification, estrogen receptor status, or DNA content.24 The current study not only confirmed earlier findings of increased lymph node metastasis in younger women but indicated that this age effect is independent of the method of detection and frequency of mammography. The analyses included only women who underwent lymph node resection. Fewer older women underwent lymph node resection, particularly during 1995–1999, and they are less likely to have had regional metastasis. Lymph node metastasis in women age ≥ 50 years, therefore, may have been overestimated, and the actual effect of age on the odds of having negative lymph nodes may be greater than reported in this study.

This study focused on invasive tumors, but our data also showed that in situ tumors increased from 2% during 1975–1984 to 20% during 1995–1999. Because in situ tumors almost always are detected mammographically, this provides further evidence of the effect of mammography on earlier breast cancer detection. The impact that in situ detection will have on reducing invasive breast carcinoma is unclear, because the rate of progression from in situ disease to invasive disease and the effectiveness of treatment for in situ tumors are not fully understood. The incidence of invasive breast carcinoma increased over the 25-year period examined by this study, so any benefits associated with detection of in situ breast tumors were not yet apparent.

Despite the difficulties inherent in comparing data collected over 25 years by differing methods, this population-based study showed a clear trend toward earlier detection of breast carcinoma in Vermont between 1975 and 1999. It also showed that most, if not all, of this trend was due to increases in the use of mammography and the consequent increase in the proportion of tumors that were detected by screening mammography. It will be possible to determine more accurately the use of mammography as the VBCSS continues to accrue statewide mammography and pathology data and to characterized more precisely the correlation between screening frequency, detection rates, and tumor characteristics in different age groups.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  • 1
    ReisLAG, EisnerMP, KosaryCL, et al., editors. SEER cancer statistics review: 1973–1997. Bethesda: National Cancer Institute, 2000.
  • 2
    Chevarley F, White E. Recent trends in breast cancer mortality among white and black U.S. women. Am J Public Health. 1997;87: 775781.
  • 3
    Chu KC, Tarone RE, Kessler LG, et al. Recent trends in U.S. breast cancer incidence, survival and mortality rates. J Natl Cancer Inst. 1996;88: 15711579.
  • 4
    Day NE, Williams DR, Khaw KT. Breast cancer screening programmes: the development of a monitoring and evaluation system. Br J Cancer. 1989; 59: 954958.
  • 5
    Carter CL, Allen C, Henson DE. Relation of tumor size, lymph node status, and survival in 24,740 breast cancer cases. Cancer. 1989; 63: 181187.
  • 6
    Maibenco D, Daoud Y, Phillips E, Saxe A. Relationship between method of detection of breast cancer and stage of disease, method of treatment and survival in women aged 40 to 49 years. Am Surg. 1999; 65: 10611066.
  • 7
    Nakhleh RE, Zarbo RJ. Surgical pathology-based outcomes assessment of breast cancer early diagnosis. Arch Pathol Lab Med. 2001; 125: 325331.
  • 8
    Dee KE, Sickles EA. Medical audit of diagnostic mammography examinations: comparison with screening outcomes obtained concurrently. Am J Roentgenol. 2001; 176: 729733.
  • 9
    Hunt KA, Rosen EL, Sickles EA. Outcome analysis for women undergoing annual versus biennial screening mammography: a review of 24,211 examinations. Am J Roentgenol. 1999; 173: 285289.
  • 10
    Wu Y, Weissfeld JL, Weinberg GB, Kuller LH. Screening mammography and late-stage breast cancer: a population-based study. Prev Med. 1999; 28: 572578.
  • 11
    McCarthy EP, Burns RB, Freund KM, et al. Mammography use, breast cancer stage at diagnosis, and survival among older women. J Am Geriatr Soc. 2000; 48: 12261233.
  • 12
    Vermont Department of Health. Cancer in Vermont: a report of 1994–1996 cancer incidence data from the Vermont Cancer Registry. Burlington: Vermont Department of Health, 2000.
  • 13
    Farwell MF, Foster RS, Costanza MC. Breast cancer and earlier detection efforts: realized and unrealized impact on stage. Arch Surg. 1993; 128: 510513.
  • 14
    Foster RS, Farwell ME, Costanza MC. Breast-conserving surgery for breast cancer: patterns of care in a geographic region and estimation of potential applicability. Ann Surg Oncol. 1995; 2: 275280.
  • 15
    Geller BM, Worden JK, Ashley JA, Oppenheimer RG, Weaver DL. Multipurpose statewide breast cancer surveillance system: the Vermont experience. J Registry Manage. 1996; 23: 168174.
  • 16
    Cady B, Stone MD, Wayne J. New therapeutic possibilities in primary invasive breast cancer. Ann Surg. 1993; 218: 338349.
  • 17
    Division of Cancer Prevention and Control, National Cancer Institute. 1987 Annual cancer statistics review including cancer trends: 1950–1985. DHHS publication no. 88-2789. Bethesda: National Institutes of Health, 1988.
  • 18
    Degnan D, Harris R, Ranney J, Quade D, Earp JA, Gonzalez J. Measuring the use of mammography: two methods compared. Am J Public Health. 1992; 82: 13861388.
  • 19
    King ES, Rimer BK, Trock B, Balshem A, Engstrom P. How valid are mammography self-reports? Am J Public Health. 1990; 80: 13861388.
  • 20
    Fulton-Kehoe D, Burg MA, Lane DS. Are self-reported dates of mammograms accurate? Public Health Rev. 1992; 20: 233240.
  • 21
    Fletcher SW, Black W, Harris R, Rimer BK, Shapiro S. Report of the international workshop on screening for breast cancer. J Natl Cancer Inst. 1993; 85: 16441656.
  • 22
    Feig SA. Benefits and risks of mammography. Recent Results Cancer Res. 1984; 90: 1127.
  • 23
    Foster RS. The biologic and clinical significance of lymphatic metastases in breast cancer. Surg Oncol Clin North Am. 1996; 5: 79104.
  • 24
    Holmberg L, Lindgren A, Norden T, Adami HO, Bergstrom R. Age as a determinant of axillary node involvement in invasive breast cancer. Acta Oncol. 1992; 31: 533538.