Breast Cancer Research and Treatment

, Volume 135, Issue 3, pp 831–837

Improving size, lymph node metastatic rate, breast conservation, and mortality of invasive breast cancer in Rhode Island women, a well-screened population

Authors

  • Natalie G. Coburn
    • Division of Surgical OncologySunnybrook Health Sciences Centre
    • Department of SurgeryCambridge Hospital
  • John P. Fulton
    • Rhode Island Department of Health
  • Calvin Law
    • Division of Surgical OncologySunnybrook Health Sciences Centre
  • Maureen A. Chung
    • John Wayne Cancer Center
Clinical Trial

DOI: 10.1007/s10549-012-2215-1

Cite this article as:
Coburn, N.G., Cady, B., Fulton, J.P. et al. Breast Cancer Res Treat (2012) 135: 831. doi:10.1007/s10549-012-2215-1

Abstract

The beneficial impact of screening mammography on breast cancer outcome continues to be debated as demonstrated by guidelines published by the United States Preventive Services Task Force. A previous report from Rhode Island, which has a very high rate of mammographic screening, demonstrated significant improvements in invasive breast cancer presentation and mortality through 2001. This report updates data through 2008 to determine whether previous favorable trends continued. Rhode Island Cancer Registry data regarding invasive breast cancer presentation and mortality in 17,522 female residents diagnosed between 1987 and 2008, inclusive, were analyzed for demographic and pathological factors. Data were analyzed by four time periods: 1987–1992, 1993–1998, 1999–2003, and 2004–2008 and overall. Statistically significant improvements occurred over the four successive time periods, in mean cancer size (23.7, 20.9, 19.6, and 19.3 mm, p < 0.0001), pathologic grade (Grade I: 12, 15, 19, and 17 %; Grade III 57, 41, 36, and 35 %, p < 0.0001), breast conserving surgery (38, 56, 67, and 71 %, p < 0.0001) and mortality (37.3, 31.4, 25.1, and 22.6 per 100,000/year, p < 0.0001). The results showed that high screening rates favorably impacted presentation of and mortality from invasive breast cancer in Rhode Island. From 1987 to 2008, there has been a 39 % decline in breast cancer mortality considering 5 year periods (37.3 vs. 22.6 deaths per 100,000) and 41 % comparing the period from 1990 to 2008, which may exceed the goal of 50 % mortality reduction by 2015 established by the American Cancer Society.

Keywords

BreastCancerScreeningMammogramsMortality

Introduction

Multiple randomized clinical trials have shown decreased mortality associated with the offering of screening mammography [1]. The applicability of these mammography trials to a general population has been questioned [2, 3] but generally accepted [4]. Others have questioned whether a plateau in improved size and stage presentation will be reached in an extensively screened population (Michaelson JS: Breast cancer detection, screening, and survival over the past 30 years. Manuscript in preparation, 2012). The Behavioral Risk Factor Surveillance System (BRFSS) (http://www.cdc.gov/brfss/) reports that Rhode Island (RI) consistently has one of the highest mammography rates in the United States (US). We previously reported decreased tumor size at presentation, greater breast conservation, and decreased breast cancer mortality through 2001 in RI women [5, 6], and reported that 72 % of breast cancer deaths occurred in the 15 % of RI women who did not have screening mammograms [7]. We have examined clinical presentation, surgical treatment, and mortality from breast cancer in this well-screened population over a longer time period to determine if favorable trends continued.

Methods

Mammographic screening rates

In 1987 and 1989, the RI Department of Health conducted telephone surveys of over 850 female RI residents to ascertain whether they had received mammography biannually if they were between 40 and 49, or annually if they were 50 and older. Since 1990, BRFSS has conducted telephone surveys including 1,600–3,800 RI women, to determine mammography use in the previous 2 years (www.cdc.gov). BRFSS was conducted annually from 1990 to 2000 and biannually thereafter. Since different survey techniques were utilized in 1987 and 1989 data, these numbers are reported, but not included in Fig. 1.
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-012-2215-1/MediaObjects/10549_2012_2215_Fig1_HTML.gif
Fig. 1

Rate of mammographic breast cancer screening in women over 40 years of age. An increase in mammographic screening has been observed across the United States. Compared to the U.S. population, RI women were more likely to report having a mammogram within the previous one (age >50) or 2 years (age group 40–49)

RI cancer registry (RICR)

Since October 1, 1986, state law mandated that the RICR should maintain a computerized database of all cancers detected or treated in RI. Breast cancer data are collected at diagnosis from hospitals, physician offices, and radiation centers, and includes age, size, grade, lymph node involvement, stage, and surgical treatment. Data are validated by 100 % review of case abstracts, computerized edit checks, and 10 % re-abstraction audits. RICR data are certified annually by the North American Association of Central Cancer Registries. IRB approval from the Department of Public Health was obtained.

From 1987 to 2008, all incident cases of invasive breast cancer in female RI residents were identified utilizing comprehensive breast cancer codes. Incidence and mortality rates of invasive breast were age-adjusted (direct age standardization) using the 2000 US Standard Population.

Data were analyzed by year and also four time periods: 1987–1992, 1993–1998, 1999–2003, and 2004–2008. Size, grade, stage, number of positive lymph nodes, and rate of breast conservation were recorded. Student’s t test and a non-parametric test were used to analyze changes in the mean and median cancer diameter, respectively; Chi square test determined changes in grade, American Joint Commission on Cancer (AJCC) stage, and proportion presenting with nodal disease. The first and the last time periods were compared to examine changes across the entire time period, and between the third and the last time periods to examine recent changes.

Death from breast cancer was determined from death certificates. Age standardized female breast cancer mortality rates for RI were constructed using SEER*Stat software, (http://seer.cancer.gov/seerstat/) and based on death records from the National Center for Health Statistics. Analogous rates were constructed for the United States using the same data.

Statistical analysis systems (SAS) 9.1 SP2 (Cary, NC) was used to analyze all data. All tests were two-sided and a p value of <0.00101 was considered statistically significant, reflecting a Bonferroni correction for multiple comparisons.

Results

The mammography rate in RI has risen significantly since the mid 1980’s, with rates higher than the overall US population (Fig. 1), and consistently ranking among the top three states. In 2008, 82 % of RI women of 40 and over reported a mammogram within 2 years, compared to 76 % for the US. There has been no change in invasive breast cancer incidence in RI (Table 1).
Table 1

Change in invasive breast cancer presentation, treatment, and outcome in Rhode Island women from 1987 to 2008

Variable

Time periods

p value

A 1987–1992

B 1993–1998

C 1999–2003

D 2004–2008

A versus D

C versus D

Number of cases

4,503

4,772

4,074

4,173

  

Incidence (per 100,000 women)

131.3

133.3

131.9

132.3

0.624

0.849

Mean age at diagnosis (years)

64.4

64.8

63.4

62.8

<0.0001

0.083

Mean tumor size (mm)*

23.7

20.9

19.6

19.3

<0.0001

0.496

Median tumor size (mm)*

20.0

15.0

15.0

15.0

<0.0001

0.155

Grade**

 One

12.1 %

15.1 %

19.4 %

16.7 %

<0.0001

0.002

 Two

31.0 %

43.9 %

44.2 %

48.6 %

<0.0001

0.000

 Three

56.9 %

41.0 %

36.3 %

34.7 %

<0.0001

0.156

Number of LN positive

 0

62.5 %

66.1 %

68.4 %

69.4 %

<0.0001

0.408

 1

12.0 %

12.2 %

13.9 %

13.6 %

0.067

0.761

 2–3

10.6 %

9.0 %

8.7 %

9.0 %

0.031

0.677

 4+

14.8 %

12.8 %

9.1 %

8.1 %

<0.0001

0.156

Stage***

 I

46.0 %

52.0 %

53.3 %

53.1 %

<0.0001

0.822

 II

38.3 %

35.3 %

35.7 %

32.8 %

<0.0001

0.007

 III

9.6 %

7.6 %

6.9 %

9.3 %

0.646

<0.001

 IV

6.1 %

5.2 %

4.1 %

4.9 %

0.023

0.102

Breast conservation (%)

37.5 %

55.2 %

67.0 %

70.7 %

<0.0001

<0.001

Mortality (deaths/100,000 women)

37.3

31.4

25.1

22.6

<0.0001

<0.0124

Data were collated in four time periods from 1987 to 2008 and invasive breast cancer presentation (tumor size, tumor grade, lymph node involvement, AJCC stage), treatment, and breast cancer mortality were determined. Results were compared for the earliest (1987–1992) and the latest (2004–2008) time periods, and the two latest time periods (1999–2003 and 2004–2008, respectively). Data were not available for tumor size (21 %), tumor grade (A—66 %; B—40 %; C—31 %; D—20 %), nodal status (A—36 %; B—32 %; C—26 %; D—20 %), and AJCC stage (A—30 %; B—11 %; C—8 %; D—5 %). Nodal status includes only surgically staged patients

Mean and median breast cancer diameter decreased from 1987 to 2008 (Fig. 2); mean size between 1987 and 1992 was 23.7 mm, which significantly decreased to 19.3 mm in 2004–2008 (p < 0.001) (Table 1), with a significant decrease in median size from 20 mm in 1987–1992 to 15 mm in 2004–2008.
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-012-2215-1/MediaObjects/10549_2012_2215_Fig2_HTML.gif
Fig. 2

Change in size of invasive breast cancers from 1987 to 2008. The high rate of mammographic screening in RI women has translated into smaller invasive breast cancers. Overall, the average size of invasive breast cancer size has decreased from 23.7 to 19.3 mm from 1987–1992 to 2004–2008 (p < 0.001). Median tumor size has stabilized at 15.0 mm from 1993 to 2008 after an initial decrease from 20.0 mm in 1987–1992

More women were diagnosed with favorable cancers over our study time period; 12 % with Grade 1 cancers from 1987 to 1992 increased to 17 % in 2004–2008 (p < 0.001), and Grade 3 cancers decreased from 57 % in the period 1987–1992 to 35 % in 2004–2008 (p < 0.001) (Table 1).

Metastatic nodal involvement decreased over the four time periods from 37.5 to 30.6 %. The extent of node involvement significantly decreased from 1987–1992 to 2004–2008 with fewer patients with four or more positive lymph nodes (Table 1). The decrease in size and nodal involvement from 1987 to 2008 resulted in more Stage I breast cancers. There has been a marked increase in the percentage of women with breast conservation surgery from 38 % in 1987–1992 to 71 % in 2004–2008 (Table 1).

Smaller size, lower grade, and decreased incidence and frequency of lymph node metastasis should result in reduced breast cancer mortality, which declined significantly from 37.3/100,000 in 1987–1992 to 22.6/100,000 in 2004–2008 (Table 1). Between the 1987–1992 and 2003–2008 periods, there was a 39 % mortality rate decrease from breast cancer (p < 0.001) with a 41 % decrease between 1990 and 2008 (Fig. 3).
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-012-2215-1/MediaObjects/10549_2012_2215_Fig3_HTML.gif
Fig. 3

Change in breast cancer mortality from 1987 to 2008. Breast cancer mortality has decreased from 1987 to 2008 for both the United States and Rhode Island. The rate of breast cancer mortality has decreased quicker in Rhode Island than the United States. Initially, Rhode Island had a higher rate of breast cancer mortality than the United States in 1987. However, by 2008, breast cancer mortality was lower in Rhode Island than that observed in the United States

Discussion

RI has experienced high rates of mammography screening for many years and has consistently ranked among the top three states on BRFSS results. This high breast cancer screening rate has resulted in statistically significant decreases in cancer size, grade, and nodal involvement since 1987 without an increase in breast cancer incidence. Our current report through 2008, demonstrates the improving presentation of invasive breast cancer noted in our report from R.I. between 1987 and 2001 [6]. Early detection of breast cancer by mammographic screening is postulated to interrupt the usual biologically progressive course of invasive breast cancers as proposed by Hellman in his “spectrum” biological model [8]. Our findings support this model; since, over time, there has been a smaller cancer size, decreasing rates and number of nodal metastases, an increase in proportion of Grade I and decrease in proportion of Grade III cancers, and a greatly reduced mortality rate in a state with high screening rates. Sentinel lymphadenectomy was introduced to RI in 1998 [9], and this new technology resulted in increased detection of nodal metastases because of more detailed pathological examination. A previous report summarizing earlier trends in breast cancer presentation from 1930 to 1994 emphasized marked changes in presentation over 65 years [10]. The sharp decrease in size and node metastatic rate beginning in 1969 at introduction of mammographic screening noted in that report [10] is substantiated in a contemporary screened cohort for RI over 22 years.

The widespread use of mammography in RI women has an historical context since RI was one of the initial five sites for the Breast Cancer Detection Demonstration Project (BCDDP) funded by congress in 1972, with the RI site opening in 1974 [11]. Wide publicity, public interest, and women’s participation in mammographic screening was generated by detection of breast cancers in Betty Ford and “Happy” Rockefeller (wives of the president and vice-president) in the fall of 1974. This publicity produced an upsurge in mammography and an apparent increase in incidence of breast cancer in 1973 and 1974 in the US [12]. This was temporary, as concerns developed regarding radiation dose, equipment standards, and other possible harmful effects [13]. By 1978, breast cancer incidence in RI and the US returned to baseline [12]. However, mammography screening rates steadily increased through the 1980’s resulting in increase in breast cancer incidence [14] because of appreciation of potential benefits of screening following early results from the BCDDP [15] and the Hospital Insurance Program (HIP) [16] mammography trial. In retrospect, the increasing breast cancer incidence in the 1980’s, followed by a plateau of incidence between 1987 and 1993, was attributed to earlier detection of non-palpable breast cancers that otherwise would appear clinically later (lead time bias).

The mortality rate from breast cancer in RI decreased 39 % between the 1987–1992 and 2004–2008 time periods and, unlike survival rates, is not subject to lead time or length bias, and thus useful for comparisons of breast cancer mortality. Currently, RI mortality rate is lower than the US average in a state which in the 1980’s had one of the highest reported breast cancer mortality rates. Other groups have found an association between lower mortality rates in areas with higher rates of screening mammography using SEER data [17]; several other population-based studies have also reported decreases in mortality [1820]. The proportional reduction in breast cancer mortality for RI is similar to Iceland and England, which reported 44.5 and 34.9 % reductions in mortality in a recent report from Europe [20]. Statistically significant decreases in mortality rates have continued over the last two time periods. As a result, a 50 % decreased mortality rate, the 2015 challenge goal of the American Cancer Society, may be surpassed in RI.

While randomized population trials demonstrate reduced mortality rates accompanying the offering of screening mammography [15, 21], there is little detailed data regarding the assumed mechanism of this accomplishment—the earlier presentation of invasive cancer in a progressive disease that evolves from small, lower grade, negative node cancers to larger, higher grade, node positive clinically detected breast cancers postulated by the spectrum model [8]. The American Cancer Society speculated that a 50 % reduction in breast cancer mortality could be achieved between 1990 and 2015 if 90 % of American women obtained yearly mammography screening, accompanied by known systemic treatment guidelines, and further research (http:www.cancer.org). Although many current breast screening guidelines suggest annual mammograms for women greater than 40 years, the BRFSS scores women as complying with mammographic screening if they report having a mammogram within 2 years. Therefore, although 85 % of RI women report having mammograms, it is likely that not all women are receiving mammography at an optimal interval. Comparing 1 year versus 2 year intervals for routine screening, it has been documented that the shorter interval markedly improves breast cancer presentation, increases the proportion of all cancers detected by mammography, and decreases the proportion of interval cancers [22]. Considering the size and node metastatic rates in RI women compared to yearly or biennial screening, we estimate that actual screening interval in RI is approximately 18 months. Data from Massachusetts implies that the usual “yearly” frequency may be about 16 months [23], and previous RI data [7] also estimates an approximate 16 month average screening interval. Further reductions in tumor size, node metastatic rate, proportion of Grade 3 cancers, and an increased proportion of Grade I cancers could be achieved if women actually underwent mammography at 12 month intervals.

We were unable to define the proportion of the declining breast cancer mortality due to earlier detection compared to usage of systemic chemotherapy and anti-hormone treatments because of incomplete data on systemic therapy in RICR. Berry et al. [24] proposed in 2006 that mammography screening contributed a median of 46 % (27–68 %) of the mortality decline based on seven models of the relationship between systemic therapy and screening. A study from the Netherlands reflects actual population data comparing mammography screening introduction over time in separate Dutch municipalities [25, 26]. All these Dutch communities had established systemic therapy programs, and comparisons could be made between onset of screening and onset of decreasing mortality in each municipality, and concluded that between 75 and 80 % of the overall 35 % mortality decline could be attributed to screening, and the remainder to systemic therapy [25, 26]. A recent report from Holland reveals a 65 % mortality reduction in the 16 years from 1992 to 2008 comparing screened to unscreened women [27]. Whatever is the outcome of debates regarding the relative impact of screening versus systemic therapy, it is clear that population screening mammography programs are a critical component of decreasing breast cancer mortality. Decreasing cancer size, grade, nodal metastasis, and stage, which resulted in better survival for breast cancers detected by mammographic screening, was well documented in summary data from the 29 BCDDP sites [15]. In every traditional assumption about invasive breast cancer, smaller, lower grade, and node negative cancers will display decreased mortality, so the RI data of improved breast cancer presentation are consistent with reduced mortality and suggest that screening played a dominant role. While some breast cancers may disseminate cancer cells systemically even at the T1a size [28], and others may never be accompanied by clinical metastatic disease even when very large [29], assumptions about the impact on mortality of size, node metastases, and grade on improved outcome are well documented [30, 31]. Although no screening program offered to women will enroll all women, 85–90 % participation may well be achieved, as in RI.

There are several limitations to our study. The rate of mammography by BRFSS is self-reported and may reflect a favorable bias in women’s recollection of prior mammography. Furthermore, phone surveys detect fewer women of low socioeconomic status who may have limited access to a listed telephone number, and immigrants who do not speak English, groups that have lower rates of screening than the general population [32]. Cases of breast cancer in RI women that were treated wholly within another state may have been missed by the RICR but is not felt to be significant since receipt of any breast cancer treatment or death within RI would result in registration in the RICR. Information regarding grade and stage was often incomplete in the early years of the RICR. Mortality data are based on cause of death listed on death certificates. Patients who died of treatment complications may be misclassified. Women, who died of other primary causes, but with active breast cancer, also may be misclassified.

Conclusion

This report describes the improved presentation of invasive breast cancer through 2008 in terms of cancer size, node metastases, grade, and breast conservation in an entire state with a high rate of mammographic screening. These significant improvements in clinical presentation of invasive breast cancer were associated with decreased mortality. The RI experience should support and encourage population programs of mammographic screening. The ultimate mortality reduction from breast cancer is not quantifiable, but the 50 % mortality reduction by 2015 proposed by the American Cancer Society is clearly achievable in RI if women over 40 obtain yearly mammography screening. A still greater reduction in mortality might well occur if size, grade, and lymph node metastatic rates continue to improve.

Acknowledgments

The authors acknowledge the National Program of Cancer Registries (NPCR) of the Centers for Disease Control and Prevention (CDC) for their support of the Rhode Island Cancer Registry (Cooperative Agreement NPCR RI DP000830), from which most of the data for this study were obtained. Drs. Calvin Law and Natalie Coburn are supported by Ministry of Health and Long Term Care Career Scientist Awards.

Conflict of interest

The authors declare that they have no conflict of interest.

Copyright information

© Springer Science+Business Media, LLC. 2012