**4. Discussion**

This study leveraged differences in provincial and territorial screening practices and used robust Canadian BC data to investigate the impact of mammography screening programs on women aged 40–49 and 50–59. Even with suboptimal screening participation, a stage shift was noted in jurisdictions without organised screening programs. A significant relationship between the degree of jurisdictional screening participation and the incidence rate of stage I BC at diagnosis was observed. On a proportionate basis, women aged 40–49 in comparator provinces were significantly more likely to be diagnosed with stages II, III, or IV BC than their screened peers. For the first time, it was observed that Canadian screening programs that included women in their 40s were associated with earlier stage migration in women aged 50–59. In jurisdictions where women in their 40s were not included in the screening programs, there were significantly higher rates of stages II and III BC in women aged 50–59, and a significant increase in the incidence of metastatic BC

over time. The stage profile of BC was observed to have changed since 2011 when the CTFPHC recommended against screening women aged 40–49. Changes appeared more evident among women aged 40–49 than among women aged 50–59 for whom screening continued to be recommended. The ongoing screening activity, albeit lessened, for women aged 40–49 may have mitigated the extent of this shift.

These findings have implications for outcomes for these women, as stage III and IV BC have an overall five-year net survival of only 74.0% and 23.2% compared to 91.9% and 99.8% for stage II and I, respectively [54]. The profile of BC in women in their 40s is often of later stage (II, III, and IV) disease at diagnosis, compared to their screened peers in their fifties. This late-stage disease, along with the increased proportion of stage III BC in women aged 50–59, results in higher mortality and life years lost, given the young age at diagnosis and lower stage-specific survival [38]. In addition to the inherently increased mortality risk with later-stage disease, these women must also undergo treatments that are far more extensive than those for early BC. A diagnosis with late-stage BC often means more invasive surgeries such as mastectomy and axillary dissection, and more intensive and longer duration of therapy with chemotherapy and radiation, causing morbidity with potential long-term toxicities such as lymphedema, secondary malignancies, and cardiotoxicity [55].

The cost of annual mammography screening for women aged 40 to 49 has been estimated at CAD 2355 per woman [56]. However, these new costs would only be applicable to those women not already undergoing screening. They could also be partially offset by the reduction in the incremental costs of otherwise having to treat more stage II, III, and IV BC with increasingly complex and expensive therapies. In Ontario in 2014, the two-year treatment costs for stage I through IV BC were CAD 29,938; CAD 46,893; CAD 65,369, and CAD 66,627, respectively [57]. The treatment of hormone receptor-positive (HR+)/human epidermal growth factor 2 negative (HER2-) metastatic disease in Ontario from 2012 to 2017 cost more than CAD 1.2 billion [58]. These costs likely underestimate the current financial impacts of treating advanced BC. A two-year time frame does not capture all costs of stage III treatment, including recurrence and ongoing survivorship care, nor the full duration of treatment for metastatic disease. As well, these costs predate the introduction of cyclin-dependent kinase (CDK) inhibitors in metastatic BC, drugs which cost more than CAD 5000 per month [59] and can be continued for more than five years [60]. These costs also do not account for the loss of productivity from time away from the workforce, due to the lengthier treatment of more advanced cancers. The high costs of implementing a population-based screening program for all Canadian women 40–49 might be offset by a risk-stratified approach; research underway may provide greater insights into the best screening approach [61].

Several reasons for not screening women in their 40s have been cited: overdiagnosis, lead time bias, and the presence of more aggressive molecular subtypes which present with more advanced disease and have a higher risk of recurrence and death [62,63]. A recent analysis, after adjusting for tumour subtype and detection method (screening or diagnostic) showed no difference in survival for women aged 40–50 compared to women aged 51–60 [64]. This finding supports diagnosing breast cancers at an early stage in women aged 40–49, perhaps even more so if the cancers are aggressive. Overdiagnosis is more likely to be a concern in older populations who have higher rates of co-morbidities and is less likely to impact younger women with few underlying medical issues [38]. Overdiagnosis has been estimated to be <0.1% in women aged 40–49 [4]. Our study supports that cancers diagnosed in women in their 40s are not overdiagnosed, as higher proportions of stage III and metastatic disease in unscreened women sugges<sup>t</sup> progression of the earlier stage BC diagnosed in screened women. An additional downside of breast cancer screening is that many women who are screened are never found to have breast cancer. Because 80% of women who develop breast cancer have no identifiable risk factors, a risk-based approach to limiting screening to only those at high risk has not been beneficial. Active research on strategies to evaluate a personalised approach to screening may include more intensive supplemental screening for some and reduced screening intensity for others [61].

Just as the absence of ethnicity in Canadian cancer data does not allow us to analyse BC outcomes related to race, current screening guidelines may not account for our racially diverse society. Screening guidelines are largely based on eight randomised controlled trials performed 30–60 years ago in Sweden, Scotland, the USA, and Canada [19]. Ethnicity was not recorded but given each country's population, it is likely that white women were mainly studied. The biology of BC differs based on ethnicity. The incidence of BC in White women peaks in their 60s, while the highest incidence for Black, Hispanic, and Asian women is in their 40s [65,66]. Higher mortality rates in Black women are independent of socioeconomic factors and are driven by higher rates of triple-negative BC, an aggressive cancer that is associated with later stage presentation and resultant lower survival [67–69]. Current screening guidelines amplify disparities, as the requirement for self-referral, or need for primary care referral for screening, decreases screening uptake, and skews participation to those patients who are health-literate and of a higher socioeconomic background [36]. Initiating breast screening at the age of 40 has been shown to reduce the mortality disparities between Black and White women and create a more equitable screening strategy [70].

This study has several limitations. It is a retrospective analysis of the impact of jurisdictional screening program policies. Quebec, the second largest jurisdiction in Canada, had to be excluded as their BC cases had not been submitted to the CCR for diagnosis years 2011 onward. None of the screener jurisdictions had optimal participation of >70%, and many of the comparator provinces had some screening activity, including greater than 50% in Newfoundland and Labrador. As such, our analysis likely underestimates the impact of organised screening programs. Among women aged 40–49, the five-year net survival estimates for unstaged and unknown stage BC cases are intermediate to that of stage II and III cases, and a few percentage points below stage I to IV combined [54]. Definitively staging these cases likely would have led to slightly higher proportions of late-stage disease in both screener and comparator groups, but more so in the comparator group given the higher rate of unstaged cases in this group. Although the CCHS survey questions attempt to measure screening mammograms only, we cannot definitively determine if the mammograms reported were diagnostic or for screening purposes. The inclusion of diagnostic mammograms in this study could overestimate late-stage or symptomatic, palpable cancers in the screener jurisdictions, thereby further attenuating the impacts of screening. Although the aim of screening is to detect cancers at an early stage, we did not have information about how the BC was diagnosed, and some cancers may have been detected symptomatically or as interval cancers. The power to detect trends in women in their 40s and 50s, especially with sub-analyses by screening status, was limited by the fact that we only have seven years of stage data. Because this study only evaluated invasive carcinomas, the impact on in situ disease was not assessed. Risk factors could not be assessed in this study and it is possible that 40–49-year-old women had more risk factors in the comparator than in the screening groups. However, given that we measured the stages of breast cancer at diagnosis this was unlikely to have significantly impacted the outcomes. In addition, the Canadian database did not record race or ethnicity, or breast tissue density, and we could not assess the impact of screening guidelines on the groups who would most likely be affected, such as Black and Asian women whose incidence of breast cancer peaks in the 40s, or those at higher risk. While the use of national registry data stands out as one of the main strengths of this study, the absence of statistical information on cancer recurrence and longer-term outcomes after an initial BC diagnosis narrows the scope of our conclusions to primary tumour cases. We were unable to evaluate breast cancer mortality outcomes using 10-year follow-ups for cases diagnosed from 2010–2017 as follow-up was restricted to the end of 2017.

This study presents important new information regarding the impact on Canadian women in their 40s and 50s of current BC mammography screening guidelines. The results can be used to reassess the optimum lower age for BC screening in Canada. Although women aged 40–49 have a lower incidence of breast cancer than those 50–59, women in both these age groups may have been negatively impacted by the exclusion of women younger than 50 from organised breast screening programs. This study supports two large observational trials [20,21], showing the benefit of breast cancer mortality reduction among women 40 years and older. Future research into the best method of including women 40–49 years old in population-based screening programs is needed. The thousands of Canadian women in their 40s who are diagnosed with BC each year have proportionally more stage III and metastatic cancers than women involved in organised screening, and there may be downstream stage migration with its associated decreased stage-specific survival in women aged 50–59. Identifying these cancers through screening at an early stage, where associated treatment costs are reduced relative to late-stage disease, could result in substantial savings for the Canadian health care system. The current screening guidelines are inequitable because they necessitate self-referral, selecting for health-literate women of higher socioeconomic status, and creating barriers for women who are required to have a primary care provider who will agree to the referral. These same guidelines preferentially benefit White women while disadvantaging ethnic groups with differing BC biology with an earlier peak incidence. Our guidelines may lead to increased treatment morbidity, and most importantly increased mortality. It is time to focus not on the harms of screening, but on the harms of not screening women aged 40–49.
