Breast Cancer Screening
Breast Cancer Screening
Current breast cancer mammography screening has clear demonstrated benefits but is also associated with some side effects. The risk-benefit balance of these effects may differ according to each individual’s risk of developing breast cancer. This is why we have created MyPeBS, which aims at comparing the benefits and risks of the two strategies: the current ongoing strategy (mammogram for all women after a starting age) versus a risk-based strategy (mammograms at different time intervals according to individual risk, + eventually MRI if necessary).
Breast cancer: a major global health problem
Breast cancer is the most common cancer and the second leading cause of cancer death among women in Western countries. Breast cancer (BC) is a dramatic worldwide issue with almost 1.7 million new BC diagnoses and 521,900 BC deaths estimated worldwide in 2012.
Breast cancer remains a potentially lethal disease. Indeed, 20 to 25% of women developing breast cancer will eventually die, due to the development of metastases. Risk of metastases and global prognosis are linked to both tumor biology and burden at diagnosis.
Although very long survivals are sometimes possible, metastatic breast cancer remains an incurable disease.
Breast cancer remains a potentially lethal disease. Indeed, 20 to 25% of women developing breast cancer will eventually die, due to the development of metastases. Risk of metastases and global prognosis are linked to both tumor biology and burden at diagnosis.
Although very long survivals are sometimes possible, metastatic breast cancer remains an incurable disease.
Localized breast cancer currently most of the time still requires aggressive and prolonged treatments associated with long-term consequences. Treatment intensity and heaviness is clearly deeply linked to the cancer’s biology, but also to tumor burden at diagnosis. Tumor burden is the major determinant of the extent of the local therapies, including surgery (partial versus complete mastectomy, axillary surgery) and radiation therapy. Adjuvant medical treatments for breast cancer, including chemotherapy, endocrine and targeted therapies remain difficult for women; they are associated with long-term sequelae, and represent high management costs.
There is therefore a major need for prevention, including earlier diagnosis (associated with a better prognosis, less treatments needs, less morbidity from the therapies, and lower costs) through secondary prevention, but also, of course, primary prevention.
There is therefore a major need for prevention, including earlier diagnosis (associated with a better prognosis, less treatments needs, less morbidity from the therapies, and lower costs) through secondary prevention, but also, of course, primary prevention.
66 million
inhabitants
inhabitants
50,000
new cases per year
new cases per year
11,500
deaths per year
deaths per year
Current breast cancer screening policies: known benefits
In Western countries, breast cancer screening is part of national organized screening systems with monitoring of screening quality and with double reading (except in Israel) of the mammograms. Certified radiologists and radiographers are responsible for quality of the diagnostic performance.
Apart from very rare patients at very high-risk, age is currently the only criterion for starting screening. Depending on the country, mammograms are offered every 1 to 3 years, starting from the age of 40-50 years up to 69-74 years.
These screening recommendations are based on large-scale randomized studies (New York, Malmo 1 and 2, Edinburgh, Swedish 2 county, Canada trials 1 and 2, Stockholm, Goteborg, UK age trial) that have globally shown that screening reduced breast cancer specific mortality by about 20% in the intent-to-treat populations (invited women), or 30-40% in the per-protocol populations (participating women).
Apart from very rare patients at very high-risk, age is currently the only criterion for starting screening. Depending on the country, mammograms are offered every 1 to 3 years, starting from the age of 40-50 years up to 69-74 years.
These screening recommendations are based on large-scale randomized studies (New York, Malmo 1 and 2, Edinburgh, Swedish 2 county, Canada trials 1 and 2, Stockholm, Goteborg, UK age trial) that have globally shown that screening reduced breast cancer specific mortality by about 20% in the intent-to-treat populations (invited women), or 30-40% in the per-protocol populations (participating women).
Several reappraisals of these mammographic screening-associated benefits in the randomized trials have been published in the past 10 years with variable interpretation of data. Indeed, trials’ methodologies are somehow heterogeneous most trials dated at times incidence and therapies were quite different. The UK independent panel in 2011 estimated the benefit of mammographic screening starting at age 50 to be in the range of one breast cancer death prevented for 250 women invited.
The benefit and risk-benefit ratio of mammographic screening between the age of 40 and 50 is controversial and each country currently has its own policy. Mammographic screening has also been demonstrated to reduce the number of stage 2 and higher cancers at diagnosis in women older than 50.
The benefit and risk-benefit ratio of mammographic screening between the age of 40 and 50 is controversial and each country currently has its own policy. Mammographic screening has also been demonstrated to reduce the number of stage 2 and higher cancers at diagnosis in women older than 50.
Current breast cancer screening by mammography: harms and weaknesses
Current screening by mammography is associated with a number of harms or weaknesses that have been largely debated in the medical literature in the past 10 years:
- The sensitivity of 2-yearly (and even more so if 3-yearly e.g. in UK) mammogram is not optimal: 1-2 (or more for UK) breast cancers every 1,000 examined women are interval cancers. This turns to 16 to 35% of cancers being interval cancers according to the screening interval. Furthermore, about one fourth of the cancers occurring in regularly screened women are still diagnosed at stage 2 or more.
- A small percentage of screening mammograms lead to additional check-ups or biopsies for an image that turns to finally be benign: these “false positive” results, according to the way they are estimated concern 3-14% of all screening mammograms, causing useless patient’s anxiety.
- Another criticism is overdiagnosis (screen detection of a cancer that would not have become clinically apparent without screening) which is estimated in average as 10% of all screen-detected cancers (estimates are highly variable; they range from 1% to 30%, depending on the population and estimation methods), leading to an inherent overtreatment.
- Mammographic screening is associated with a risk of radio-induced breast cancer. This risk appears extremely low (about 1 in 1,000 women screened during 30 years) compared to the benefits of early diagnosis and radiation doses delivered are now very closely monitored.
Changing the schedule can be beneficial
Annual screening may be more effective than biennial screening for women at high risk due to dense breasts and other risk factors. On the opposite, several studies have suggested that women who would undergo less screening in a low risk situation would derive a good benefit to harm ratio.
For all these reasons, risk-based screening is expected to be non-inferior, and potentially superior to standard age-based screening since:
- In high risk individuals, although screening harms will not decrease and may even increase due to a higher screening frequency, such screening has large chances to be more efficient, as demonstrated in many publications;
- In low risk individuals, benefit to harm ratio should be driven by much less harms in terms of false positive findings, overdiagnoses, radio-induced cancers, whereas efficacy should not be decreased if a lower frequency screening is used. However, low risk does not mean no risk and appropriate surveillance measures (self-palpation, etc.) would have to be reinforced.
“4P medicine” and application to breast cancer screening
Since individual factors, except for family history, have a limited impact when used alone, several multivariable mathematical models to estimate breast cancer risk in the general population have been developed over the past 25 years. All of these models use clinical variables based on family history, history of benign breast disease, as well as variables that summarize a certain amount of endogenous and exogenous hormone exposure derived from epidemiological studies.
These breast cancer risk models can be separated into those that utilize mainly hormonal and environmental factors and those that focus more on hereditary risk. Indeed, specific models have been developed in high familial risk populations that are able to predict for the probability of a germline mutation as well as for a woman’s individual breast cancer risk in this setting: they include the extended Claus and more recently, BRCAPRO and Bodicea models. These models are, however, not suitable for the general population, and have been developed to predict for BRCA1/2 mutations but may be less relevant for other germline alterations.
These models are not suitable for the general population, in which the most accurate models are the three renewed Breast Cancer Risk Assessment Tool (BCRAT/Gail), Tyrer-Cuzick (IBIS) and Breast Cancer Screening Consortium (BCSC) models.
These breast cancer risk models can be separated into those that utilize mainly hormonal and environmental factors and those that focus more on hereditary risk. Indeed, specific models have been developed in high familial risk populations that are able to predict for the probability of a germline mutation as well as for a woman’s individual breast cancer risk in this setting: they include the extended Claus and more recently, BRCAPRO and Bodicea models. These models are, however, not suitable for the general population, and have been developed to predict for BRCA1/2 mutations but may be less relevant for other germline alterations.
These models are not suitable for the general population, in which the most accurate models are the three renewed Breast Cancer Risk Assessment Tool (BCRAT/Gail), Tyrer-Cuzick (IBIS) and Breast Cancer Screening Consortium (BCSC) models.
A family history of breast cancer suggests the presence of an inherited genetic variant such as those in the BRCA1 and BRCA2 genes which confer a high susceptibility (Couch et al. 2014). Recently, additional genetic risk factors have been identified, including rare variants in genes such as PALB2, CHEK2, associated with moderate to high risk and common low-risk variants.
Non-genetic breast cancer risk factors include hormonal factors (e.g. use of hormone replacement therapy, oral contraception), reproductive factors (e.g. age of first pregnancy, breastfeeding, age at menarche, age at menopause) and lifestyle factors (e.g. obesity, physical activity, alcohol consumption).
Overall, except for true genetic predisposition, each of these factors alone has a limited impact, with relative risks between 1.1 (reproductive factors) and 3.
Beside this, and over the past 20 years, breast density has been explored and validated in many studies as an important breast cancer risk factor, together but independently of its other effect (masking effect): density is indeed currently regarded as an indicator that summarizes / can be used a surrogate marker of both a woman’s genetic background and exogenous exposures to hormones or other risk modifiers.
Non-genetic breast cancer risk factors include hormonal factors (e.g. use of hormone replacement therapy, oral contraception), reproductive factors (e.g. age of first pregnancy, breastfeeding, age at menarche, age at menopause) and lifestyle factors (e.g. obesity, physical activity, alcohol consumption).
Overall, except for true genetic predisposition, each of these factors alone has a limited impact, with relative risks between 1.1 (reproductive factors) and 3.
Beside this, and over the past 20 years, breast density has been explored and validated in many studies as an important breast cancer risk factor, together but independently of its other effect (masking effect): density is indeed currently regarded as an indicator that summarizes / can be used a surrogate marker of both a woman’s genetic background and exogenous exposures to hormones or other risk modifiers.
Breast cancer risk estimation
Although deaths from breast cancer have been decreasing in many Western countries, the incidence of breast cancer is continuing to increase. In particular, in countries with historically low incidence, breast cancer rates are rising rapidly making it now the world’s most prevalent cancer. The increase in incidence is almost certainly related to changes in dietary and reproductive patterns associated with Western lifestyles. There is evidence from genetic studies in the USA, Iceland and the UK of a 3-fold increase in incidence in the last 80 years, not only in the general population, but also in those with BRCA1 and BRCA2 mutations.
Exploration and description of breast cancer-associated risks through large retrospective and prospective cohorts have allowed a very high amount of data regarding potential individual risk factors of breast cancer.
A number of breast cancer risk factors have been identified, including family history, hormonal exposure, reproductive history and lifestyle.
Exploration and description of breast cancer-associated risks through large retrospective and prospective cohorts have allowed a very high amount of data regarding potential individual risk factors of breast cancer.
A number of breast cancer risk factors have been identified, including family history, hormonal exposure, reproductive history and lifestyle.
A family history of breast cancer suggests the presence of an inherited genetic variant such as those in the BRCA1 and BRCA2 genes which confer a high susceptibility (Couch et al. 2014). Recently, additional genetic risk factors have been identified, including rare variants in genes such as PALB2, CHEK2, associated with moderate to high risk and common low-risk variants.
Non-genetic breast cancer risk factors include hormonal factors (e.g. use of hormone replacement therapy, oral contraception), reproductive factors (e.g. age of first pregnancy, breastfeeding, age at menarche, age at menopause) and lifestyle factors (e.g. obesity, physical activity, alcohol consumption).
Overall, except for true genetic predisposition, each of these factors alone has a limited impact, with relative risks between 1.1 (reproductive factors) and 3.
Beside this, and over the past 20 years, breast density has been explored and validated as an important breast cancer risk factor, together but independently of its other effect (masking effect). Many studies have now acknowledged this breast cancer risk effect, which may be seen as a surrogate of genetic background and lifetime hormonal/other exposures: density is indeed currently regarded as an indicator that summarizes both a woman’s genetic background and exogenous exposures to hormones or other risk modifiers.
Non-genetic breast cancer risk factors include hormonal factors (e.g. use of hormone replacement therapy, oral contraception), reproductive factors (e.g. age of first pregnancy, breastfeeding, age at menarche, age at menopause) and lifestyle factors (e.g. obesity, physical activity, alcohol consumption).
Overall, except for true genetic predisposition, each of these factors alone has a limited impact, with relative risks between 1.1 (reproductive factors) and 3.
Beside this, and over the past 20 years, breast density has been explored and validated as an important breast cancer risk factor, together but independently of its other effect (masking effect). Many studies have now acknowledged this breast cancer risk effect, which may be seen as a surrogate of genetic background and lifetime hormonal/other exposures: density is indeed currently regarded as an indicator that summarizes both a woman’s genetic background and exogenous exposures to hormones or other risk modifiers.
Breast cancer risk assessment models
Since individual factors, except for family history, have a limited impact when used alone, several multivariable mathematical models to estimate breast cancer risk in the general population have been developed over the past 25 years. All of these models use clinical variables based on family history, history of benign breast disease, as well as variables that summarize a certain amount of endogenous and exogenous hormone exposure derived from epidemiological studies.
These breast cancer risk models can be separated into those that utilize mainly hormonal and environmental factors and those that focus more on hereditary risk. Indeed, specific models have been developed in high familial risk populations that are able to predict for the probability of a germline mutation as well as for a woman’s individual breast cancer risk in this setting: they include the extended Claus and more recently, BRCAPRO and Bodicea models. These models are, however, not suitable for the general population, and have been developed to predict for BRCA1/2 mutations but may be less relevant for other germline alterations.
These breast cancer risk models can be separated into those that utilize mainly hormonal and environmental factors and those that focus more on hereditary risk. Indeed, specific models have been developed in high familial risk populations that are able to predict for the probability of a germline mutation as well as for a woman’s individual breast cancer risk in this setting: they include the extended Claus and more recently, BRCAPRO and Bodicea models. These models are, however, not suitable for the general population, and have been developed to predict for BRCA1/2 mutations but may be less relevant for other germline alterations.
Recent breast cancer risk models are based on screening cohorts and integrate mammographic breast density as a factor. This addition has slightly increased their accuracy in discriminating women who do and do not get breast cancer, with concordance statistics (c-statistics) of about 0.65 compared to 0.58 for models that do not include density.
A crucial point is to use models that are internationally validated, and all 3 of these scores/models are currently being externally validated. The teams involved in MyPeBS have experience with two major, recently updated, and well renewed, breast cancer risk assessment models:
- The American BCSC model has been validated in the Mayo clinic cohort and, more recently, in French general breast screening populations (after adjustment on national incidence, c-statistic 0.61, E/0 1.005) and can be used as such.
- The Tyrer-Cuzick model has been largely described in general populations as well as high-risk family clinics or clinical trial populations (IBIS1). It has particular relevance for women with a family history: its accuracy is average in the general population (c-statistics between 0.57-0.60), while it is very high in family-risk populations (c-statistics up to 0.70).
Genotyping allows breast cancer risk identification
A single-nucleotide polymorphism (SNP), is a variation of the DNA in a single nucleotide that occurs at a specific position in the genome, where each variation is present to some appreciable degree within a population (e.g. > 1%).
Beside the previous clinical risk factors and their aggregation in scores, huge international efforts (Europe and North America), through advances in genome technology, have led to the identification of over a hundred and fifty common, validated single nucleotide polymorphisms (SNPs) associated with breast cancer risk
These SNPS predict either for invasive breast cancer in general for most of them and/or for risk of hormone-receptor negative, or risk of death from breast cancer. Most SNPS have a low effect, those described initially having the highest impact (OR 1.01-1.30 overall).
Beside the previous clinical risk factors and their aggregation in scores, huge international efforts (Europe and North America), through advances in genome technology, have led to the identification of over a hundred and fifty common, validated single nucleotide polymorphisms (SNPs) associated with breast cancer risk
These SNPS predict either for invasive breast cancer in general for most of them and/or for risk of hormone-receptor negative, or risk of death from breast cancer. Most SNPS have a low effect, those described initially having the highest impact (OR 1.01-1.30 overall).
The latest publications identified More than 300 new independent loci that are associated with overall breast cancer risk at P < 5 × 10−8. The majority of credible risk single-nucleotide polymorphisms in these loci fall in distal regulatory elements, and by integrating in silico data to predict target genes in breast cells at each locus, they demonstrated a strong overlap between candidate target genes and somatic driver genes in breast tumors.
The complementarity of SNPs to predict cancer risk, with respect to other breast cancer risk factors, namely mammographic density, reproductive history, and lifestyle factors, is now demonstrated. The combinations of risk models including classical variables + mammographic density + SNP score (namely a polymorphism risk score = PRS) are used to estimate more accurately population risk, refining the high- and low-risk groups, such as in MyPeBS.
The complementarity of SNPs to predict cancer risk, with respect to other breast cancer risk factors, namely mammographic density, reproductive history, and lifestyle factors, is now demonstrated. The combinations of risk models including classical variables + mammographic density + SNP score (namely a polymorphism risk score = PRS) are used to estimate more accurately population risk, refining the high- and low-risk groups, such as in MyPeBS.