Impact of Breast Cancer Screening on 10-Year Net Survival in Canadian Women Age 40-49 Years

Abstract

PURPOSE

In Canada, some provincial/territorial mammography screening programs include women age 40-49 years, whereas others do not. This study examines the impact of this dichotomy on the 10-year breast cancer (BC) net survival (NS) among women age 40-49 years and 50-59 years at diagnosis.

METHODS

Using the Canadian Cancer Registry data record linked to death information, we evaluated the cohort of Canadian women age 40-49 years and 50-59 years diagnosed with BC from 2002 to 2007. We compared 10-year NS estimates in the jurisdictions with organized screening programs that included women age 40-49 years, designated as screeners (Northwest Territories, British Columbia, Alberta, Nova Scotia, and Prince Edward Island), with comparator programs that did not (Yukon, Manitoba, Saskatchewan, Ontario, Quebec, New Brunswick, and Newfoundland and Labrador).

RESULTS

BC was the primary cause of 10-year mortality in women age 40-49 years diagnosed with BC (90.7% of deaths). Among these women, the 10-year NS for screeners (84.8%; 95% CI, 83.8 to 85.8) was 1.9 percentage points (pp) higher than that for comparators (82.9%; 95% CI, 82.3 to 83.5; P = .001). The difference in favor of screeners was significant among women age 45-49 years (2.6 pp; P = .001) but not among women age 40-44 years (0.9 pp; P = .328). Similarly, the incidence-based BC mortality rate was significantly lower in screener jurisdictions among women age 40-49 years and 45-49 years, but not for 40-44 years. Provincial/territorial NS increased significantly with higher mammography screening participation (P = .003). The BC incidence rate was virtually identical in screener and comparator jurisdictions among women age 40-49 years (P = .976) but was significantly higher for comparators among women age 50-59 years (P < .001).

CONCLUSION

Screening programs that included women in their 40s were associated with a significantly higher BC 10-year NS in women age 40-49 years, but not an increased rate of BC diagnosis. These results may inform screening guidelines for women age 40-49 years.

INTRODUCTION

Breast cancer (BC) is the most common cancer diagnosed in Canadian women in their 40s.^1,2 Despite a relatively favorable prognosis, BC is a leading cause of death among women in their forties and fifties in Canada. In 2020, it accounted for 10.5% of all deaths occurring in women in their 40s, second only to all accidental deaths (11.5%), and was the leading cause of death in women in their 50s (10.4%), greater than heart disease (9.3%).³

CONTEXT

• Key Objective

• This study leveraged the variable implementation of organized mammography screening programs for women age 40-49 years in jurisdictions across Canada to evaluate the impact of screening on 10-year net survival (NS) and incidence-based mortality rate ratios among women age 40-49 years.

• Knowledge Generated

• Breast cancer (BC) was the primary cause of 10-year mortality among women age 40-49 years diagnosed with BC (90.7% of deaths). The incidence of BC was not increased among women age 40-49 years living in screening jurisdictions but was increased among women age 50-59 years in comparator jurisdictions. The 10-year NS was significantly higher among women age 40-49 years in screener than comparator jurisdictions, particularly among women age 45-49 years (by 2.6 percentage points). These survival results were corroborated by an incidence-based BC mortality analysis. NS increased significantly with higher mammography screening participation. A significant survival advantage was observed among women age 40-49 years, and especially age 45-49 years, when included in organized mammography screening programs.

• Relevance (I. Cheng)

• With the recent release of draft recommendations by the US Preventive Services Task Force of biennial mammography screening for women starting at the age 40, this study provides timely findings of a higher 10-year NS for women participating in screening programs of the ages of 40-49.*

  *Relevance section written by JCO Associate Editor Iona Cheng, PhD, MPH.

The value of BC screening in women age 50-74 years is widely accepted,⁴ and organized BC screening programs for these women are well established. However, BC screening in women in their 40s continues to be debated, with concerns that despite the recognized benefits, the risks, including false positives, benign biopsies, and overdiagnosis, may overwhelm the benefits in these women.⁵ Overdiagnosis has been more accurately described as overdetection because screen detection of nonprogressive tumors does not necessarily imply inappropriate treatment.⁶ Critics of screening mammography tend to favor the use of randomized controlled trials (RCTs), which have many limitations including age of study, small numbers, intention-to-treat analyses, older technology, and long screening intervals. The Canadian Task Force on Preventive Health Care was informed by RCTs and recommended against routine screening in women in their 40s in 2011 and 2018.^7,8 However, evidence from community-based exposure to screening using modern protocols in large observational population-based studies has shown mortality benefits of 29%,⁹ 41%,¹⁰ and 44%¹¹ in these women. A systematic review of global guidelines for BC screening recommends initiation of screening at age 40 years,¹² as do guidelines from the National Comprehensive Cancer Network and the most recent US Preventive Services Task Force.^13,14

Screening with mammography decreases BC mortality by diagnosing BC at an earlier, more treatable stage.¹⁵ A comparison of provinces and territories in Canada with organized breast screening programs for women age 40-49 years with those jurisdictions without such programs found evidence of downstaging with screening. Jurisdictions with no organized screening programs for women age 40-49 years had a significantly lower proportion of BCs diagnosed at stage I and significantly higher proportions of stage II, III, and IV BCs at diagnosis compared with those jurisdictions that had organized screening.¹⁶ Treatments for early-stage BC (stage 0 and I)¹⁷ involve less invasive surgical techniques and less need for chemotherapy.¹⁸ Beyond the decreased morbidity of treatments in early-stage BC, there are also clear survival benefits. The 5-year net survival (NS) for stage I BC is 99.8%, compared with 91.9%, 74.0%, and 23.2% for stage II, III, and IV BC.¹⁹

NS estimates survival in the hypothetical situation where the disease under study would be the only possible cause of death.²⁰ In the context of cancer, NS measures the effect on survival of a cancer diagnosis after removing the effects of competing causes of death.²¹ NS is crucial for comparing cancer survival in population-based cancer studies because it adjusts for the fact that different populations might have different levels of background risk of death.²⁰ Although survival could appear increased by screening because of the lead time created by detecting cancer before it presents symptomatically, in the setting of a mortality benefit, the relevance of lead time bias is diminished.

This study assesses the impact of breast screening programs for women in their 40s on survival. For women age 40 to 49 years, ten-year NS is compared between those diagnosed with BC in jurisdictions with organized screening programs and those diagnosed in jurisdictions with no organized screening. To mitigate the potential for lead time bias, incidence-based mortality rates were also compared between the two groups.

METHODS

Survival analyses were undertaken using a pre-existing analytic file created by linking Canadian Cancer Registry²² cases to mortality information complete through December 31, 2017. Additional details regarding this file can be found elsewhere.²³ Age-specific BC incidence rates were derived using the Canadian Cancer Registry²⁴ and population data from Statistics Canada.²⁵ On the basis of the International Classification of Diseases for Oncology, Third Edition,²⁶ only primary invasive BC cases, defined as site code C50,^27,28 were included. Ductal carcinoma in situ (DCIS) cases were not included. This study was a secondary analysis of nationally deidentified data collected by Statistics Canada, and as such, ethics approval was not required.

BC NS estimates were derived using an algorithm²⁹ that has been augmented by Ron Dewar of the Nova Scotia Health Cancer Care Program to include the Pohar Perme estimator of NS²⁰ using the hazard transformation approach (R. Dewar, 2020, email communication, June 22). The Pohar Perme estimator is an unbiased and consistent estimator of NS.²⁰ NS controls for background differences in mortality and accounts for both the direct and indirect (eg, suicide) increased mortalities because of a BC diagnosis. It does not require information on the cause of death, which is susceptible to error. Expected survival probabilities, necessary for the calculation of NS in a relative survival framework, were mostly obtained from sex-specific complete annual provincial population life tables.³⁰ Further details on the calculation of expected survival are provided elsewhere.²³ Cases with an undefined survival time (eg, cases for which the diagnosis had been established through autopsy or death certificate only) were necessarily excluded. Only the first BC case diagnosed per person on the file was considered.^31,32

To alleviate concerns regarding lead time bias, a form of incidence-based mortality analysis was also conducted.³³ The rate of BC deaths for the 2002-2017 period for BC cases diagnosed from 2002 to 2007 in screener jurisdictions was compared with the corresponding rate in comparator jurisdictions. The denominator for each jurisdictional group was the female population in that group from 2002 to 2007. Comparisons were made for women age 40-44 years, 45-49 years, 50-54 years, and 40-49 years, with a null hypothesis of no difference in rates between screener and comparator jurisdictions (ie, the ratio of rates equals one).

Screening programs in Canada were established between 1989 and 1992 and vary according to each provincial or territorial health authority.¹⁶ Ten-year NS estimates were calculated for the cohort of eligible cases diagnosed from 2002 to 2007. Ten-year NS was selected as the longest period possible to observe the recognized benefits of BC mortality reduction within the context of available data and established screening programs in Canada. It was also selected to minimize the impact of the lead time, which, for BC in premenopausal women, is 1-2 years. Canadian provinces and territories were designated as screeners or comparators on the basis of screening practices for women age 40-49 years in this period.^34,35 Five jurisdictions with organized screening programs, including self-referral and annual recall, were designated as screeners: British Columbia, Alberta, Nova Scotia, Prince Edward Island, and the Northwest Territories. Except for the territory of Nunavut, which was not included because of insufficient data with which to classify its screening status, the remaining jurisdictions: Ontario, Quebec, Manitoba, Saskatchewan, New Brunswick, Newfoundland and Labrador, and the Yukon Territory, formed the comparator group.

The percentage screening participation for each province and territory was established through the Canadian Community Health Survey (CCHS). The CCHS is a national cross-sectional survey that allows determination of the percentage of women age 40-49 years who reported having a screening mammogram within the previous 2 years.^36,37 The 2-year period for mammograms was used such that screening percentages would apply to the period 2002-2007.

The Z-test was used to determine P values for percentage point (pp) differences in 10-year NS between the screener and comparator jurisdictional groups. P values correspond to two-sided tests of the null hypothesis that the difference in NS is zero, with a significance level of .05. For interpretational context, the analysis was also performed for women diagnosed with BC in their 50s using the same jurisdictional groupings. A weighted linear regression analysis was used to assess the relationship between provincial-specific 10-year NS among women diagnosed in their 40s and the average of self-reported jurisdictional screening percentages in 2003 and 2008. Finally, overall crude 10-year mortality was indirectly partitioned into that attributable to a BC diagnosis and that because of other causes of death, for women diagnosed in their 40s and for other age groups for which screening programs exist across most jurisdictions.^38,39

RESULTS

This study included 21,103 cases of BC in women age 40-49 years diagnosed between 2002 and 2007. Of these, 5,680 cases of BC were diagnosed in jurisdictions, which offered screening in organized population-based screening programs, and 15,408 cases were in jurisdictions with no organized screening programs and only limited opportunistic screening (Table 1). The remaining cases were diagnosed in Nunavut for which the screening status was unclassified.

TABLE 1.

Comparison of 10-Year NS of Female With BC by Jurisdictional BC Screening Status Among Women Age 40-49 Years, by Age Group (40-59 years), Canada, 2002-2007 Diagnosis Period

On the basis of women diagnosed with BC in their 40s between 2002 and 2007, the 10-year NS was observed to be 1.9 pp higher (P < .001) among those cases originating from screening jurisdictions (84.8%; 95% CI, 83.8 to 85.8) than among those cases from comparators (82.9%; 95% CI, 82.3 to 83.5; Table 1). The absolute difference in 10-year NS between screener and comparator jurisdictions was greater among women age 45-49 years (2.6 pp; P = .001) than among women age 40-44 years (0.9 pp; P = .328). The incidence-based BC mortality rate was significantly lower in screener than comparator jurisdictions among women age 40-49 years at diagnosis (incidence-based BC mortality rate ratio, 0.92; 95% CI, 0.85 to 0.99; data not shown) and among women age 45-49 years (rate ratio, 0.89; 95% CI, 0.81 to 0.98) but not among women age 40-44 years (rate ratio, 0.95; 95% CI, 0.85 to 1.07; Fig 1).

FIG 1.

Screener-to-comparator jurisdictions’ incidence-based BC mortality rate ratios, by 5-year age group at diagnosis, Canada, diagnosis years 2002-2007. Screening jurisdictions were defined as those with screening programs that included women in their 40s for annual recall during the diagnosis period. Screeners: Prince Edward Island, Nova Scotia, Alberta, British Columbia, and Northwest Territories; Comparators: Newfoundland and Labrador, New Brunswick, Quebec, Ontario, Manitoba, Saskatchewan, and Yukon Territory. The territory of Nunavut is excluded from both screeners and comparators because there were insufficient data to classify their screening status. Follow-up of cases for BC mortality is performed until the end of 2017. Adapted from Statistics Canada.²² BC, breast cancer.

Significant differences in NS were observed at all single year survival durations, up to 10 years, among women diagnosed in their forties (Fig 2A), but not in any such duration among women diagnosed in their fifties (Fig 2B). No difference in age-specific BC incidence rates between screeners and comparators was noted among women age 40-49 years (133.1 cases per 100,000 women in both cases; P = .976), whereas corresponding rates among women age 50-59 years were significantly higher in comparator jurisdictions (238.4 v 217.6 cases per 100,000 women; P < .001; Appendix Table A1 and Fig A1, online only).

FIG 2.

Ten-year cumulative female BC NS by jurisdictional screening status, Canada, 2002-2007 diagnosis period. Screeners (age 40-49 years): Prince Edward Island, Nova Scotia, Alberta, British Columbia, and Northwest Territories; Comparators: Newfoundland and Labrador, New Brunswick, Quebec, Ontario, Manitoba, Saskatchewan, and Yukon Territory. The territory of Nunavut is excluded because there were insufficient data to classify their screening status. The 95% CIs are denoted by vertical bars overlaid on the trend lines. Follow-up of cases is performed until the end of 2017. Adapted from Statistics Canada.²² BC, breast cancer; NS, net survival.

BC was the primary cause of 10-year mortality in women age 40-49 years diagnosed with BC, implicated in 90.7% of their deaths (Table 2). This figure decreased to 80.9% among women diagnosed with BC in their 50s, to 61.5% for the 60s, and finally to 48.3% among those diagnosed from age 70 to 74 years as the influence of competing causes of death increased with age.

TABLE 2.

Ten-Year Observed Survival of Female With BC, and Partitioning of Overall 10-Year Mortality Into That Because of BC and That Because of Other Causes of Death, Age 40-74 Years, Canada, 2002-2007 Diagnosis Period

The provincial CCHS screening participation for women age 40-49 years showed some screening activity in all provinces/territories, ranging from 21.8% in Manitoba to 52.9% in Alberta (Table 3). Provincial/territorial NS increased significantly with higher screening participation percentage (P = .003; Table 3; Fig 3). For every increase of 10 pp in screening, there was a 1.1 pp increase in 10-year NS. The NS ranged from a low of 81.8% (95% CI, 78.3 to 84.7) in Manitoba, a province with no organized screening, to a high of 85.8% (95% CI, 84.3 to 87.1) in British Columbia where women were able to access mammogram screening if they wished.

TABLE 3.

CCHS Screening Participation Rates for all Women Age 40-49 Years and 10-Year NS for Women Age 40-49 Years Diagnosed With BC by Province/Territory, 2002-2007 Diagnosis Period

FIG 3.

Weighted linear regression of 10-year female BC NS by provincial screening rate, age 40-49 years, Canada excluding the territories, 2002-2007 diagnosis period. The provincial screening rate is the average of the 2003 and 2008 rates obtained from the Canadian Community Health Survey. Provincial importance weights were calculated on the basis of the relative incidence of BC over the 2002-2007 period and represented by the relative size of the respective provincial bubbles. P value corresponds to a two-sided test of the null hypothesis that the slope of the line is zero, with a significance level of .05. Adapted from Statistics Canada,²² Government of Canada Statistics Canada.³⁶ BC, breast cancer; NS, net survival.

INTERPRETATION

The 10-year NS among women diagnosed with BC in their 40s from 2002 to 2007 was 1.9 pp higher in jurisdictions that had organized screening programs for these women compared with those that did not. Among women age 45-49 years, the NS was 2.6 pp higher, and the incidence-based BC mortality rate was significantly lower (rate ratio, 0.89), in screening jurisdictions. For context, the absolute mortality benefit seen with tamoxifen, one of the most accepted and widely used treatments for BC, is 3.3 pp after 5 years of treatment.⁴⁰ NS was significantly correlated with the screening participation rates in women age 40-49 years, with jurisdictions with higher rates of mammography in women age 40-49 years displaying higher NS. No significant difference in 10-year BC NS between screener and comparator jurisdictions was observed among women in their 50s for whom organized screening programs exist in all jurisdictions.

While stage data are not available at a national level before 2010, a recent study, that examined BC stage at diagnosis by jurisdictional screening status from 2010 to 2017, found a significant stage shift in those jurisdictions which allowed self-referral and had annual recall in women age 40-49 years.¹⁶ Given the inherent survival advantage for lower-stage BCs,¹⁹ the higher NS observed in screener jurisdictions may be the direct result of early detection efforts. The lower incidence of BC observed in women in their 50s in screener jurisdictions may also signal the benefit of early detection in women age 40-49 years because of lead time and the treatment of DCIS, which may be associated with a reduction in the incidence of invasive cancer in the next decade of life.⁴¹ A recent modeling study found that the majority of DCIS lesions progress to invasive BC within 2.6 years and that the overdiagnosis rate is only 1.3%-2.4%.⁴² Rates of obligate overdiagnosis are age-dependent and are lowest in women in their 40s, where screening contributes to the overdiagnosis of DCIS by only 0.15% and to the overdiagnosis of invasive BC by 0.1%.⁴³

The higher NS observed among women age 40-49 years in screening jurisdictions in the 2002-2007 period is potentially a conservative estimate of the magnitude of this difference because of recent advancements in the treatment of earlier-stage disease, which reduces mortality in these cancers. In the United States, the mortality reduction in early-stage BC increased from 35.8% in 2000 to 48.2% in 2017, and for HER2 NEU–positive (HER2+) disease, from 19.1% to 42.1%.⁴⁴ Trastuzumab, which reduces 10-year mortality in HER2+ BC by 6.4%, became available for adjuvant treatment in 2005.^45,46 Aromatase inhibitors were approved by Health Canada in 2004⁴⁷ and subsequently became widely used in the treatment of postmenopausal estrogen receptor–/progesterone receptor–positive cancers.

It is also important to consider that screening jurisdictions only offered screening, and participation rates in these jurisdictions ranged from 31.2% to 52.9%, at no time approaching the target screening participation of 70%.³⁴ At the same time, there was significant opportunistic (nonorganized) screening activity ongoing in the comparator group, with screening rates greater in some comparator jurisdictions (ie, New Brunswick) than in screener provinces (ie, Prince Edward Island). As such, the observed difference between screeners and comparators may underestimate the benefits, which could be seen with complete screening participation within screener jurisdictions and no screening activity within their comparator counterparts.

Other factors beyond screening, such as access to care, available treatments, and ethnicity of populations, may have also affected survival differences. This might have been the case with Newfoundland and Labrador, which is noted to have a lower NS despite higher screening activity, perhaps because of known founder mutations.⁴⁸ The relationship between screening and NS might have been underestimated as the CCHS percentages were for biennial screening. Optimal BC screening in women age 40-49 years should be annual given the more rapid growth of BCs in women with higher levels of hormones.^49–52 Annual screening should be based on menopausal status not age, and although women in their 40s are more likely premenopausal, about 10% may be postmenopausal and similarly women in their fifties may be premenopausal.^53,54 Finally, there was no ability to determine which BCs were screen-detected and which were symptomatically detected, so a direct correlation of NS with screening activity could not be obtained.

A population-based approach allows for a robust analysis of thousands of BC cases. Although the comparison of survival between screener and comparator jurisdictions removes the issue of volunteer bias, such a comparison is subject to both lead and length time bias.^55,56 However, with BC, these biases are thought to be in the order of a few years, and long-term follow-up reduces their impact.⁵⁵ The 10-year follow-up period of this study would reduce the impact of length time bias, given that 80% of all BC cases that are ultimately fatal die within 10 years, whereas >95% of these deaths occur within 20 years.⁵⁷ Incidence-based BC mortality supported the survival findings for women diagnosed with BC in their 40s, indicating that lead time bias was unlikely to be the reason behind the higher NS observed in screener jurisdictions.

Overdetection is often cited as a reason not to screen women age 40-49 years as between 1% and 32% of BCs are thought to be indolent and therefore not affect mortality.^58,59 Although it is unknown how many BCs were treated which would not have affected mortality, the exclusion of DCIS cases likely mitigated a component of potential overdetection.^6,60 The similar incidence of BC in women age 40-49 years in both screener and comparator jurisdictions suggests that screening did not lead to an increased rate of a BC diagnosis in this age group. While we cannot account for the specific risk factors in all jurisdictions, the increased incidence of BC in women age 50-59 years in comparator jurisdictions may be the result of effective treatment of screen-detected noninvasive neoplasms, such as DCIS in women in their 40s, which did not subsequently progress to invasive cancers in the 50s.

BC was implicated in over 90% of all deaths in women age 40-49 years diagnosed with this cancer. This figure declined with age to just under 50% among corresponding women age 70-74 years at diagnosis. Women in their 40s have few competing causes of mortality, and so BC has a much larger contribution to their mortality than in older women. These results support the findings of a recent modeling study that reported that overdetection was a much larger problem in older women.⁶¹ Although this study is unable to assess overdetection on an individual level, the observed significant relationship between screening and NS suggests that, on a population basis, screening is finding potentially fatal BCs.

Decisions around BC screening in women in their 40s must weigh the potential harms and benefits of screening; however, this is not possible without accurate information concerning the impact of BC on mortality in women age 40-49 years. This study elucidates the survival benefits which screening may provide for these women and highlights the importance of BC as a cause of death in women age 40-49 years. Screening detects most BCs when they are early stage, resulting in less morbidity from treatment. Women with advanced BC at diagnosis will not only face more intensive treatments with resultant lost productivity, but also have a higher risk of premature death. Screening younger women is associated with a marked increase in life-years gained.⁶²

Despite suboptimal screening participation and a cohort diagnosed before two major treatment advances, this study suggests that BC screening programs for women age 40-49 years may translate into a significant survival benefit. Concerns regarding overdetection of BC among women in their 40s may be attenuated not only by the higher BC survival noted in screener jurisdictions but also by the similarity of BC incidence rates between screener and comparator jurisdictions among women in their 40s and the finding that BC has a relatively large impact on overall mortality in women in this age group with a BC diagnosis. The findings of this study can be used to inform decision making regarding screening guidelines for women in their 40s.

APPENDIX

TABLE A1.

BC Incidence Rate Per 100,000 Females by Jurisdictional Status and Age Group, Canada, 2002-2007

FIG A1.

Breast cancer incidence rate per 100,000 females by jurisdictional screening status and age group, Canada, 2002-2007. Screeners (age 40-49 years): Prince Edward Island, Nova Scotia, Alberta, British Columbia, and Northwest Territories; Comparators: Newfoundland and Labrador, New Brunswick, Quebec, Ontario, Manitoba, Saskatchewan, and Yukon Territory. The territory of Nunavut is excluded because there were insufficient data to classify their screening status. The 95% CIs are denoted by vertical bars overlaid on the trend lines. Adapted from Statistics Canada.²⁴ Version released on August 23, 2022.

AUTHOR CONTRIBUTIONS

Conception and design: All authors

Collection and assembly of data: Larry F. Ellison, Jean-Michel Billette

Data analysis and interpretation: All authors

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Impact of Breast Cancer Screening on 10-Year Net Survival in Canadian Women Age 40-49 Years

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).