Do we still need breast cancer screening in the era of targeted therapies and precision medicine?

Mammography screening was introduced in 1956 for the “detection of early cancer of the breast” [11]. In the 1970s, several randomised controlled trials confirmed a beneficial improvement of mammography screening on breast cancer mortality [12]. Such benefit pertains to women aged 50 to 74 years and is maximal, about 30%, in those aged 60 to 69 years [13‐16]. Mammography screening programs have the general aim of reducing breast cancer-specific mortality both by minimising the risk of diagnosing breast cancer in the advanced stage and by maximising the efficacy and safety of anticancer treatments and their impact on prognosis.

The first x-ray units dedicated to mammography used x-ray film and paired fluorescent screens to capture the image, in the so-called screen-film mammography, which has been gradually replaced since the early 2000s by digital mammography units. In 2005, a landmark study by Pisano et al. [17] compared the diagnostic accuracy of digital and screen-film mammography in 42,760 asymptomatic women. At receiver operating characteristic analysis over the entire population, the areas under the curve (AUCs) of the two methods differed by a non-significant 0.03. However, the digital technique was significantly superior among women over 50 (0.15 AUC difference), among women having heterogeneously or extremely dense breasts (0.11 AUC difference), and among women being in pre- or perimenopause (0.15 AUC difference). Digital mammography allowed also to deliver a lower radiation dose and resulted in easier image storage and subsequent access. Its initially higher costs dwindled quite quickly, its uptake steadily increased [18, 19]. By 2015, 96% of all mammography units in the United States had gone digital [20].

The incidence of advanced-stage tumours in the target population is a specific proxy indicator of screening efficacy, given the absence of confounding effects of treatments at the time of diagnosis. However, studies on this topic have adopted different thresholds in defining the severity of the diseases and are characterised by poor to fair overall quality. Nevertheless, a meta-analysis of trials combined their results – including the most severe disease categories available – and showed a significant reduction in the risk of advanced-stage disease for women aged 50 years or older (relative risk 0.62, 95% confidence interval [CI] 0.46–0.83) randomly assigned to undergo screening: this benefit was however not seen in women aged 39 to 49 years [14].

More recently, a prospective Italian cohort study was conducted on 413,447 women undergoing screening in the 1990s and followed up for 13 years as part of the IMPACT project [21]. Screening attendance was positively associated with: (a) a 39% reduction of the incidence of pT2–T4 lesions (66.3‰ versus 108.6‰, incidence rate ratio 0.61, 95% CI 0.57–0.66), including a 28% reduction of pT2 lesions and a 68% reduction of pT3–T4 lesions; (b) a 28% reduction in the incidence of stage II–IV disease (130.1‰ versus 180.6‰, incidence rate ratio 0.72, 95% CI 0.68–0.76), including a 35% reduction of stage IIB, a 43% reduction of stage III, and a 73% reduction of stage IV. Notably, patients undergoing screening also benefited from a 17% reduction in the incidence of poorly differentiated carcinomas and from a 50% increase of breast-conserving surgery rates. These results were also confirmed following adjustments to exclude self-selection bias, a significant reduction in the use of mastectomy being also observed [22].

Another interesting large population study – including 549,091 women across nine Swedish counties [23] – was conducted to evaluate the impact of screening mammography on breast cancer mortality. The study used an analytic strategy focusing on the incidence of fatal breast cancers within 10 years from the date of diagnosis, whereas other studies had been retroactively considering 10 years from the date of death. Women who attended mammography screening had a significant 41% reduction in their risk of dying of breast cancer within 10 years (relative risk 0.59, 95% CI 0.51–0.68) and a 25% reduction in the rate of advanced breast cancer (relative risk 0.75, 95% CI 0.66–0.84), regardless of the recent improvements in systemic treatments.

In 2019, the American Cancer Society and the National Cancer Institute published cancer treatment and survivorship statistics in the United States. Most patients (44%) were diagnosed with stage I disease, 30% with stage II, 9% with stage III and 5% with stage IV. The 5-year breast cancer relative survival ranged from approximately 100% for stage I disease to 26% for stage IV breast cancer [9]. Interestingly, in 2018 Mariotto et al. [24] provided the first population-based summaries of the risk of breast cancer recurrence in United States women, using cancer registry disease-specific survival: stage remained the strongest predictor of the risk of recurrence, along with age (60–74 years) and HR-negative status. Thus, we can still affirm that tumour stage has a substantial impact on prognosis and on the risk of disease recurrence.

Chemotherapy has been used in the adjuvant setting since the early 1970s when landmark studies in the United States and in Italy documented a benefit of regimens such as L-phenylalanine mustard or the combination of cyclophosphamide, methotrexate and 5-fluorouracil (CMF) in patients with node-positive breast cancer [25, 26]. In the 1980s and 1990s, anthracyclines and taxanes proved to be more effective than CMF [27]; in the meantime, tamoxifen was found to substantially improve the survival of women with HR-positive tumours [28], and later on, aromatase inhibitors (AIs) proved to further improve outcomes in postmenopausal patients. During the 2000s, anti-HER2 therapies were developed as one of the first targeted systemic treatment options, changing dramatically the management and prognosis of HER2-positive breast cancer; in the same decade, genome expression profiling was deployed in routine practice, further improving the personalisation of breast cancer treatments [29]. Nonetheless, patient selection remains crucial to maximise efficacy and safety of systemic treatments. The current standard of care for different disease subtypes is summarised below.

Defining the relative contribution of screening mammography and systemic treatments to improve breast cancer outcomes is a challenging task. The increasing use of screening and the gradual implementation of more effective therapeutic approaches occurred over nearly the same period since the 1970s. In the meantime, breast surgery has also evolved, along with the introduction of sentinel lymph node biopsy and radiation therapy. Overall, each of all these developments substantially contributed to improve patient outcome.

The Cancer Intervention and Surveillance Modelling Network used modelling techniques to provide estimates of the contributions of screening mammography and adjuvant treatment to the reduction of breast cancer mortality in the United States from 1975 to 2000 [68]. Seven independent statistical models were developed and yielded similar qualitative conclusions, namely that “the decline in mortality rate can be explained by a combination of screening and therapy and not by either one alone”. The proportion of the total reduction in the rate of death from breast cancer attributed to screening varied in the seven models from 28% to 65% (median 46%). On the other hand, the contribution of systemic therapy including chemotherapy and tamoxifen varied from 35% to 72% (median 54%): differences in these estimates reflect the mutual interaction between the two interventions [68].

Saadatmand et al. [69] investigated – in a large population-based cohort study – whether tumour stage at diagnosis still influences survival in the context of the current therapeutic approaches. This prospective nationwide population-based study was conducted in the Netherlands and included 173,797 women diagnosed with breast cancer, with two cohorts being identified according to the year of breast cancer diagnosis. A total of 80,228 patients were diagnosed with breast cancer from 1999 to 2005, while 93,569 from 2006 to 2012: in this second period – following national guidelines – systemic therapy was more widely used. At univariate and multivariate analyses, tumour stage and nodal status significantly influenced overall and relative survival in both cohorts. Relative survival rates ranged from almost 100% in both cohorts for in situ tumours to 57% and 59% for T4 tumours in the older and more recent cohort, respectively. At multivariate analysis, breast-conserving treatment (more frequently pursued in the 2006–2012 cohort) resulted in a significant survival benefit compared with mastectomy, whereas lymph node dissection (less frequently pursued) was associated with a significantly worse overall survival. The wider use of chemotherapy in the 2006–2012 cohort conferred a hazard ratio for death of 0.86 (95% CI 0.80–0.92). These large-scale results clearly demonstrate that while the use of chemotherapy may impact on survival, tumour size at diagnosis still matters. Authors concluded that “in the current era of effective systemic therapy, diagnosis of breast cancer at an early stage remains vital” [69]. Moreover, considering United States women diagnosed with breast cancer at age 60–74, summaries of the risk of breast cancer recurrence showed a 5-year recurrence rate of 2.5%, 9.6% and 34.5% for stages I, II and III HR-positive breast cancers, and a 5-year recurrence rate of 6.5%, 20.2% and 48.5% for stages I, II and III HR-negative breast cancers [24].

Another study from the United States [6] also assessed to what extent digital mammography screening and novel systemic therapies contributed to the improvement in breast cancer mortality in different disease subtypes from 2000 to 2012: its results are summarised in Table 1. Complex simulation models from the Cancer Intervention and Surveillance Network projecting breast cancer mortality trends for women aged 30 to 79 years estimated a 12% difference (model range, 10–16%) in the overall disease-specific mortality reduction between 2000 (37%, model range, 27–42%) and 2012 (49%, model range, 39–58%). The relative contribution to the decrease in overall breast cancer mortality in 2012 was 37% (model range, 26–51%) for screening and 63% (model range, 49–74%) for treatment. Of the 37% mortality reduction associated with screening in 2012, 33% (model range, 29–48%) was associated with screening advances before 2000 and 4% (model range, 1–8%) after 2000 (the shift from screen-film to digital mammography was relatively less relevant than previous improvements when considering the conspicuous advancements in therapeutic options). Similarly, of the 63% mortality reduction associated with treatments in 2012, 32% was associated with chemotherapy, 27% with hormone therapy and 4% with trastuzumab. Of the 31% mortality reduction (model range, 23–37%) associated with chemotherapy, 9% (model range, 7–14%) was associated with chemotherapy advances after 2000 (largely taxanes). Of the 27% mortality reduction (model range, 18–36%) associated with hormone therapy, 7% (model range, 2–12%) was associated with advances in hormone therapy after 2000 (largely from AIs) [6].

Subtype analyses also demonstrated significant variations in the relative contribution of screening and treatment to the mortality reduction in different tumour molecular subtypes. The largest benefit of screening (48%, model range, 38–57%) was found in the triple-negative tumour breast cancer cohort, while the largest benefit of treatment (69%, model range, 59–77%) was documented in luminal tumours. According to this model-based analysis, both screening and treatment contributed and still contribute to the improvement in breast cancer mortality, with progressively greater contributions of therapeutic advances in the last decades [6].

The relative contribution associated with mammography screening and advances in systemic therapies to the reduction in the recurrence rate is simulated in Figs. 2 and 3. Here, four scenarios combine variably the contribution of screening mammography and chemotherapy or novel systemic treatments (i.e. AIs and targeted agents). In this simulation, stage distribution was derived from Puliti et al. [21], considering attenders and non-attenders to mammography screening and adjusted for self-selection. A 5% overdiagnosis attributed to population screening has been considered when it was included in the scenario. Ten-year disease-specific survival for HR-positive and HR-negative cancers was retrieved from Mariotto et al. [24], based on women aged 60–74 years treated in the 1992–1999 period or in the 2000–2013 period, when more advanced therapies were available. This simulation shows how combining screening mammography and novel systemic agents represent the most favourable scenario with the lowest number of recurrences, also detailing how the recurrence rate at 10 years would be 30% for chemotherapy without screening and 19% for chemotherapy with screening, 25% for novel treatments without screening and 15% for novel treatments with screening.

Two important aspects should however be considered in the analysis of the evolution of screening and systemic therapies. Despite the widespread use of screening mammography, the introduction of digital techniques represents the most important technical advancement: of note, tomosynthesis has become the technique of choice for symptomatic women and those who are recalled for further assessment after a screening mammogram [13]. Conversely, the evidence in favour of its use as a first-level screening tool is still not sufficient in terms of reduction of interval cancer rate (possibly due to an underlying rate of overdiagnosis [70‐73]) and European guidelines [13] have advised against its use for population-based screening programs. On the other hand, increasingly effective novel systemic treatment options are being introduced as a standard of care and gene expression profiling is progressively enabling better decision-making.

This is once again a turning point. In case of further outcome improvements, most of the contribution would then be attributable to better systemic treatments. Nevertheless, MRI [74] and other novel imaging modalities, such as contrast-enhanced mammography [75‐77] could come to be considered screening tools in selected populations, their contribution being also promising. Furthermore, a recent survey among members of the European Society of Radiology [78] showed that there are high expectations on the use of artificial intelligence [79, 80]. Dedicated algorithms will focus on personalised risk prediction and prognosis [81] and machine/deep learning software has already shown high performance in interpreting screening mammography. Finally, the availability of molecular analyses on liquid biopsy could represent another promising option [82, 83].

However, further widespread benefits in breast cancer outcomes are expected to become ever slimmer in the context of the efficacy of the current standard treatments: research efforts will therefore need to focus on the role of precision medicine. Once outcomes are maximal in the overall breast cancer patient population, improvements might indeed be sought in selected populations and especially in those with poor prognosis.