Introduction
Breast cancer is one of the most diagnosed cancers worldwide, representing around 12% of all new cancer cases and approximately 25% of the total number of new cancer cases in women in 2020
[1 ]. Estimates for Europe show a similar trend, where breast cancer represented 13% of all new cancer cases in 2020 and is estimated to account for 29% of all
new cancers in women [2 ], [3 ].
Early breast cancer, defined as cancer confined to the breast or that has only spread to the axillary lymph nodes, accounts for more than 90% of breast cancer diagnoses [4 ]. Earlier detection and improved treatment options in recent years have contributed to increased survival for patients with breast cancer. However, this improvement
in survival, coupled with the rising incidence, also means that more patients are at risk of recurrence (ROR) than ever before [3 ].
Between 8 and 10% of women diagnosed with breast cancer will present with locoregional recurrences, and 15 – 30% will develop distant metastases [5 ], [6 ], [7 ]. Mortality after local recurrence varies by disease stage, with 15-year mortality rates of 16, 32, and 59% for stage 0, I, and
II, respectively [8 ].
Recurrence in breast cancer can appear early (usually defined as within 5 years of diagnosis) or late (beyond 5 years). Patients with hormone receptor-positive (HR+) tumours (which include
tumours with oestrogen receptor positivity [ER+] with or without progesterone receptor positivity [PR+]) have a lower ROR soon after diagnosis, but their risk persists longer than for patients
with HR− tumours. Cases that are also human epidermal growth factor receptor 2 (HER2)-negative (HER2−) clearly have a higher risk of late recurrence than treated HER2-positive (HER2+)
individuals [9, 10] as a consequence of effective anti-HER2 treatment for HER2+ patients. HR+, HER2− breast cancer constitutes around 70% of all breast cancers [4 ], [11 ]. More than half of women (55%) have been reported to remain disease free in the first 10 years after diagnosis, and more than 10% of these
individuals will go on to develop recurrences after 10 years [12 ]. In women with breast cancer aged 60 – 74 years, the estimated recurrence rates within 5 years
were 2.5, 9.6, and 34.5% for stage I, II, and III HR+ tumours, and 6.5, 20.2, and 48.5% for stage I, II, and III HR− tumours, respectively [10 ]. Estimated
recurrence rates for stage I, II, and III HR+ tumours were 5.2, 16.6, and 45.5% within 10 years and 7.8, 21.4, and 50.7% within 15 years, respectively. For stage I, II, and III HR− tumours,
estimated recurrence rates were 9.7, 23.5, and 51.9% within 10 years and 11.2, 24.6, and 53.1% within 15 years, respectively [10 ]. A meta-analysis comparing 5
yearsʼ treatment with aromatase inhibitors versus tamoxifen in early breast cancer found that both endocrine therapies provided significant benefits in reducing the ROR and breast cancer
mortality during years 0 – 9 versus no endocrine treatment. There was a significantly greater reduction in recurrence risk for aromatase inhibitors versus tamoxifen (by about 30%
[proportionately]) during years 0 – 4 but no difference in years 5 – 9 [13 ].
Breast cancer is a highly heterogeneous disease, with subtypes exhibiting substantial differences with regards to their disease presentation, patterns of metastasis, drug sensitivity, timing
of recurrence and prognosis. Treatment recommendations have been adapted to take this heterogeneity into account, being more biology centred, with treatment strategies differing according to
the tumour subtype. Therapeutic strategies for HR+, HER2− patients may focus on the need for treatment de-escalation to reduce their adverse effects and sequelae, with the identification of
patients who can safely be excluded from additional adjuvant chemotherapy and/or extended adjuvant endocrine therapy being a high priority.
The vast majority of ER+ patients do receive adjuvant endocrine therapy, which reduces the ROR and improves overall survival (OS); however, the recurrence rate remains ≥ 20% in the first
10 years in ER+ patients [14 ], [15 ].
For patients with HR+, HER2− early breast cancer, current standard treatments include a combination of surgery, with or without (neo)adjuvant chemotherapy, radiation therapy, and adjuvant
endocrine therapy; neoadjuvant or adjuvant chemotherapy is given to patients who are at increased ROR [16 ], [17 ]. It is also
recommended to treat postmenopausal women, or women receiving ovarian suppression therapy, with bisphosphonates to both prevent cancer treatment-induced bone loss and reduce the risk of
disease recurrence [18 ].
Risk of distant recurrence is the main determinant of recommendations for chemotherapy, and the risk of disease recurrence is significantly reduced with adjuvant chemotherapy [19 ]. Determining what constitutes high recurrence risk in early breast cancer is challenging because of individual patient factors and the large number of prognostic
methods that may or may not be available to a particular clinician. Consequently, most treatment guidelines do not recommend “one size fits all” risk thresholds [17 ], [20 ], [21 ]. In the MINDACT trial, high ROR was defined as high-risk gene expression signature results with no to
three involved lymph nodes, involvement of four or more lymph nodes or a > 10% risk of breast cancer-specific mortality at 10 years [22 ], but these
thresholds have not been established in clinical practice. Another approach to assessing whether chemotherapy should be recommended is by determining the absolute 10-year survival benefit from
chemotherapy. For example, the Cambridge Breast Unit (UK) considers that chemotherapy is not recommended if the 10-year survival benefit is < 3%; chemotherapy is discussed as a treatment
option when the 10-year survival benefit is 3 – 5%, and chemotherapy is recommended when the 10-year survival benefit is > 5% [23 ]. Some patients may
overestimate their ROR, which can have a negative impact on quality of life. Therefore, it is important to communicate recurrence risk to patients, particularly for those at lower risk [24 ]. Since the risk–benefit balance of chemotherapy can become unfavourable in low-risk patients because of treatment toxicity or in patients with a high benefit
from endocrine therapy [19 ], [25 ], low-risk patients benefit from de-escalation of chemotherapy [26 ].
As described, the ROR influences the treatment decisions for patients with HR+, HER2− early breast cancer. For example, the American Society of Clinical Oncology (ASCO) guidelines recommend
extended adjuvant endocrine therapy for patients at high ROR (although “high risk” is not defined) [27 ], [28 ]. The ASCO
guidelines were recently updated to recommend offering abemaciclib for 2 years plus endocrine therapy for ≥ 5 years to patients with resected, HR+, HER2−, node-positive early breast cancer at
high ROR [29 ] based on results from the phase III monarchE study [30 ]. Recently those results were updates with 5-year follow up
data and therapy effects were maintained over time [31 ]. Similarly, the National Comprehensive Cancer Network guidelines for breast cancer were recently updated
for patients with HR+/HER2−, high-risk breast cancer to recommend that 2 years of adjuvant abemaciclib be considered in combination with endocrine therapy. In these guidelines, high risk is
defined as patients with four or more positive lymph nodes, or one to three positive lymph nodes with one or more of the following: grade 3 disease, tumour size ≥ 5 cm, or a Ki-67 score of
≥ 20% [32 ].
Given the importance of estimating the ROR for treatment strategies, several molecular tests to classify patients with breast cancer into different recurrence risk groups have been developed
in recent years [33 ]. Some uncertainty remains as to which markers are reliable for identifying high-risk patients and for the differentiation of early and late
recurrence [34 ].
This review presents a summary of the currently available tools and patient characteristics that can be used by clinicians to identify patients with HR+, HER2− early breast cancer at high ROR
to guide optimised and balanced therapeutic decisions.
Review
Prognostic factors defining outcome
Prognostic and predictive factors
A prognostic biomarker is a characteristic (clinical or biological) that can be used to estimate the likely patient health outcome, regardless of any treatment. In oncology, a prognostic
factor is a marker that can be used to estimate the outcomes for a patient with a cancer, usually either recurrence, progression, or death [35 ]. A
predictive biomarker indicates the likely benefit of a treatment for a patient [36 ]. Some markers can be both prognostic and predictive, such as ER, PR, and
HER2 expression, whereas others are only prognostic, such as tumour size and lymph node status [37 ].
Prognostic biomarkers help determine which patients should receive particular treatments such as chemotherapy, hormone therapy, or neither. For patients selected to receive systemic
therapy, predictive biomarkers will help identify the most appropriate treatment or combination of treatments [38 ].
The current range of prognostic factors for recurrence in patients with early breast cancer include patient and tumour characteristics in addition to histological markers and gene
expression signatures. These prognostic factors are used to assess a patientʼs risk of an event; predictive factors help select treatment options that may reduce that risk. The main
classical clinicopathological and molecular prognostic factors in breast cancer are described below.
Routine clinicopathological prognostic factors
Tumour size (T), lymph node involvement (N), and the status concerning distant metastases (M) are among the factors with the greatest prognostic relevance. These three characteristics are
used to make up the TNM staging system, which has to be determined for every patient with breast cancer [39 ], [40 ].
Tumour size
Tumour size is one of the most important prognostic factors in breast cancer; tumours are categorised according to size as T0 (no evidence of primary tumour), T1 (≤ 20 mm in greatest
dimension), T2 (> 20 but ≤ 50 mm in greatest dimension), T3 (> 50 mm in greatest dimension), and T4 (any size with direct extension to the chest wall and/or to the skin) [41 ]. There is a well-established correlation between primary tumour size and the risk of developing metastases [42 ]. Tumour
size is an independent prognostic factor of recurrence in breast cancer, with patients who have larger tumours (T2–T4) having a greater ROR than those with smaller tumours (T0–T1) [43 ].
Lymph node involvement
Regional lymph node involvement has long been recognised as an important prognostic factor in early breast cancer. Prognosis worsens as the number of lymph nodes involved increases; the
higher the number of axillary lymph nodes involved, the higher the risk of local and distant recurrence [44 ] and the shorter the disease-free survival
(DFS) and OS times [45 ]. The status of the regional lymph nodes (N stage) is determined according to the extent of nodal involvement; axillary, internal
mammary, and ipsilateral supraclavicular lymph nodes can be involved, but the ipsilateral axillary lymph node is the most common site of involvement, making it a strong negative
prognostic factor significantly related to high recurrence risk [46 ], [47 ]. Axillary lymph node involvement is associated
with tumour size, histological grade and subtype in early breast cancer. Definition of axillary lymph node level is based on the relationship with the pectoralis minor muscle; level I
nodes are on the lateral border, level II nodes are between the medial and lateral borders, and level III lymph nodes are on the medial border of the pectoralis minor muscle [46 ]. Clinical lymph node assessment is categorised according to stage as N0 (no regional lymph node metastases), N1 (metastases to movable ipsilateral level I,
II axillary lymph node[s]), N2 (metastases in ipsilateral level I, II axillary lymph nodes that are clinically fixed or matted; or in clinically detected ipsilateral internal mammary
nodes in absence of clinically evident axillary lymph node metastases), N3 (metastases in ipsilateral infraclavicular [level III axillary] lymph node[s] with or without level I, II
axillary lymph node involvement; or in clinically detected ipsilateral internal mammary lymph node[s] with clinically evident level I, II axillary lymph node metastases; or metastases in
ipsilateral supraclavicular lymph node[s] with or without axillary or internal mammary lymph node involvement) [41 ], [46 ].
Pathological lymph node assessment is the lymph node assessment most commonly taken into consideration for prognosis and categorises findings according to stage as pN0 (no regional
lymph node metastasis identified or isolated tumour cell clusters only), pN1 (micrometastases, or metastases in one to three axillary lymph nodes and/or clinically negative internal
mammary nodes with micrometastases or macrometastases by sentinel lymph node biopsy), pN2 (metastases in four to nine axillary lymph nodes, or positive ipsilateral internal mammary nodes
by imaging in the absence of axillary lymph node metastases), pN3 (metastases in ten or more axillary lymph nodes, or in infraclavicular [level III axillary] lymph nodes, or positive
ipsilateral internal mammary lymph nodes by imaging in the presence of one or more positive level I, II axillary lymph nodes, or in more than three axillary lymph nodes and
micrometastases or macrometastases by sentinel lymph node biopsy in clinically negative ipsilateral mammary lymph nodes, or in ipsilateral supraclavicular lymph nodes) [41 ], [46 ].
Tumour grade
The Nottingham grading system (also called the Elston–Ellis modification of the Scarff–Bloom–Richardson grading system) is one of the most commonly used grading systems for breast
cancer. It stratifies tumours into low-, intermediate-, or high-grade categories by scoring three features:
the proportion of tubule (and gland) formation,
the nuclear pleomorphism, and
the mitotic counts.
Each feature is given a score of 1 – 3, then the individual scores are added together, giving a possible total score of 3 – 9 points. Tumours are classified as grade 1 (low grade, 3 – 5
points), grade 2 (intermediate grade, 6 – 7 points), or grade 3 (high grade, 8 – 9 points) [48 ]. Well-differentiated tumours are associated with the best
prognosis, with patients with poorly–moderately differentiated tumours having poorer survival [49 ]. Tumour grading is partly subjective, and this can lead
to some disagreements in grading between pathologists [50 ] and between different hospitals. Grading has demonstrated limited reproducibility, so whether
decisions on therapy should be based on this marker is under debate [51 ].
Both increasing tumour size and grade are associated with a significant long-term risk of distant recurrence. In a clinical trial of tamoxifen, patients with larger tumours and
higher-grade tumours had significantly worse long-term survival than patients with smaller tumours and lower-grade tumours [52 ]. Colleoni et al. [53 ] studied the ROR at 5-year intervals over 25 years and found that, although in the first 5 years patients with grade 3 tumours had a numerically higher
hazard of recurrence than patients with grade 1 or 2 tumours, this risk difference decreased beyond 5 years, with the hazard of recurrence for patients with grade 3 tumours being at
least equal to that of those with a grade 1 tumour [53 ].
ER and PR expression
ER and PR are involved in normal breast development and in mammary tumorigenesis [54 ]. A feasible method to measure the expression of both receptors is
immunohistochemistry (IHC), and breast cancer tissue samples are considered positive (hormone therapy responsive) for ER or PR if 1 – 100% of tumour nuclei are positive for the
receptors. ER and PR are commonly used as prognostic markers in breast cancer, with the presence of PR expression in ER+ patients being associated with a better prognosis [55 ]. However, patients who are ER-“low” positive (1 – 10% of cells staining ER+) can still be at high ROR, with some clinical and biological similarities with
triple-negative patients [55 ], [56 ]. Furthermore, low expression of PR (defined as < 20%) is associated with a high
risk of relapse [56 ]. Patients with ER+/PR+ tumours have lower risks of disease-related mortality than those with ER+/PR−, ER−/PR+, or ER−/PR− tumours
[57 ]. Compared with ER−/PR− tumours, ER+/PR+ breast cancers generally have less aggressive clinicopathological features and have a better prognosis since
they benefit from endocrine treatment [58 ], [59 ].
HER2 status
HER2 is a transmembrane tyrosine kinase receptor expressed in breast tissues. The amplification of its gene leads to HER2 overexpression and mammary tumorigenesis and has been
associated with an unfavourable prognosis [60 ]. HER2 is usually assessed to decide for or against an anti-HER2 therapy. An algorithm incorporating IHC and
in situ hybridization is used: stained breast cancer tissue samples are given a score of 0 to 3+, where a score of 0 – 1+ (no staining or incomplete, faint/barely perceptible staining)
is considered HER2− and a score of 3+ (complete, intense, circumferential staining) is considered HER2+. A score of 2+ (weak to moderate complete membrane staining) is considered
equivocal and requires analysis of the gene amplification by in situ hybridization [61 ], [62 ]. Patients with low HER2
expression levels that do not reach the HER2-positivity threshold may also benefit from some HER2-targeted therapies [63 ], [64 ], [65 ]. HER2 has been found to be overexpressed in approximately 15 – 20% of breast cancers [63 ]. HER2
overexpression is associated with an unfavourable prognosis and correlates with a lower frequency of hormone receptor expression and a higher histopathological grade [6 ], [66 ].
Clinical factors
Age
Both young age (normally defined as < 35 – 40 years) and older age (≥ 65 years) are associated with poor prognosis in breast cancer [4 ], [68 ]. Older patients tend to have more favourable disease characteristics (e.g. smaller tumours and lower grade) [68 ].
Furthermore, patients aged ≥ 75 years are less likely to be given adjuvant therapy and have a higher risk of death than treated patients [69 ], [70 ]. Older patients also have a greater likelihood of presenting with comorbidities, and this has a negative effect on their survival [69 ]. The influence of comorbidities in the OS estimates for elderly patients suggests that breast cancer-specific survival rates should be measured instead
[71 ]. Similarly, the survival benefits of chemotherapy may be masked when looking at OS in this population because of the competing causes of mortality,
and comorbidities may prompt the use of chemotherapy with lower toxicity than standard agents. However, more tolerable chemotherapy comes at a cost of reduced efficacy [72 ].
Additional biomarkers of potential use in clinical routine
Ki-67
Ki-67 is an immunohistochemical marker of cell proliferation, labelling nuclei of cycling cells from G1 to M phase of the cell cycle. Ki-67 correlates well with tumour grade [73 ] and with results from molecular profiling systems [74 ], [75 ].
Higher levels of Ki-67 expression are associated with poor prognosis in breast cancer. No consensus has yet been reached as to which threshold for Ki-67 labelling index defines high
risk [76 ]. Ki-67 values ≤ 10% are generally considered low risk. Different investigators have proposed a value of ≥ 20% to define high risk for DFS
prognosis [77 ], but it is not generally accepted. In a decentralised observational study [78 ] and a central setting within
the frame of a prospective study [79 ], unfavourable prognosis and high-risk recurrence score (RS) were encountered in tumours with a Ki-67 labelling index
exceeding 35%. High and low levels of Ki-67 correlate well with classical prognostic markers; however, correlation is less clear for intermediate levels of Ki-67 from 15 to 35% [73 ].
Ki-67 has been suggested to differentiate between types of breast cancer, with a cut-off Ki-67 index < 14% proposed to distinguish luminal A-like tumours from luminal B-like tumours
[80 ], [81 ]. The St. Gallen consensus temporally adopted this threshold but discarded it in subsequent meetings. No
consensus has yet been reached on a parameter to discriminate luminal A from luminal B tumours. Proliferation suppression by short-term pre-operative endocrine therapy has yielded a
novel prognostic and predictive marker based on Ki-67. Currently, Ki-67 response to endocrine therapy represents a predictive biomarker to indicate that endocrine therapy alone without
chemotherapy might be sufficient for treatment of premenopausal women with breast cancer. Furthermore, in the WSG-ADAPT-HR+/HER2− trial, Oncotype DX RS (discussed later) was combined
with 3-week Ki-67 response to endocrine therapy to guide systemic therapy, and it was found that Ki-67 can be used to identify patients with a higher RS that can be spared chemotherapy
[82 ].
Scoring of Ki-67 for routine clinical practice can be challenging, with manual scoring methods having high inter- and intra-user variability, so it is not routinely used in clinical
practice in all countries. Automated scoring methods have been developed to help with this issue [83 ]. As with ER, PR, and HER2 determination, quality
assurance trials for Ki-67 have demonstrated that regular participation will improve concordance rates among pathologists [84 ].
Together with the described prognostic markers, Ki-67 plays an important role in therapy recommendations for patients with early breast cancer [85 ].
Importantly, Ki-67 may help to provide a more accurate prognosis in patients with an intermediate prognosis according to clinical factors (particularly stage II disease) and identify
patients who are at high ROR and who may benefit from further therapy [86 ]. To improve inter-observer concordance and to provide a platform for both
training and testing of capacity to read Ki-67 labelling in breast cancer, a novel digital pathology tool has been established [87 ].
Prognostic biomarkers under investigation
Circulating tumour cells
Circulating tumour cells (CTCs) are intact, viable non-haematological cells with malignant features that have been shed from the primary tumour or a metastatic lesion into the
bloodstream as single cells or clusters. In breast cancer, CTCs can provide information about disease progression and response to therapy in patients with metastases [88 ]. Detection of CTCs is rare in early breast cancer, but it has been validated as a prognostic marker for metastasis. CTCs were shown to be a prognostic
marker for reduced DFS, distant DFS, breast cancer-specific survival, and OS before the start of systemic treatment and for DFS after completion of adjuvant chemotherapy in the SUCCESS
(Simultaneous Study of Gemcitabine-Docetaxel Combination adjuvant treatment, as well as Extended Bisphosphonate and Surveillance-Trial) prospective trial [89 ]. Detection of CTCs is not routinely used in clinical practice, and further studies, including prospective trials, are needed to assess its utility [90 ].
Bone marrow involvement – disseminated tumour cells
Although bone marrow involvement is not assessed in routine clinical practice, studies have shown that in patients with stage I – III breast cancer, the presence of occult
cytokeratin-positive metastatic cells in bone marrow detected by IHC is associated with an increased risk of relapse. Furthermore, the presence of these disseminated tumour cells (DTCs)
in the bone marrow following chemotherapy is associated with a poor prognosis in early and metastatic breast cancer [91 ], [92 ]. An international pooled analysis showed that DTC detection was an independent prognostic marker for OS, DFS, and distant DFS [93 ]. On the
other hand, the presence of DCTs in bone marrow was not significantly associated with the risk of locoregional relapse in some studies [92 ], [93 ], while another one established an association [94 ].
Circulating tumour DNA
Circulating tumour DNA (ctDNA) is fragmented cell-free DNA that is released from necrotic and apoptotic cancer cells and that may contain cancer-specific mutations that have occurred in
the originating cell. Elevated ctDNA levels have been associated with poorer outcomes in patients with early, locally advanced, and metastatic breast cancer [95 ]. A meta-analysis found that ctDNA is a strong prognostic marker in breast cancer; high levels of ctDNA and the presence of ctDNA were significantly associated with poor
DFS/recurrence-free survival (RFS) and OS in patients with breast cancer [96], and ctDNA is used to identify patients in the ongoing ZEST trial of niraparib [97 ]. However, like CTCs, ctDNA is not used in routine clinical practice.
Prognostic scoring systems and models
Several prognostic scoring systems have been developed for breast cancer, including the Nottingham Prognostic Index (NPI), Adjuvant! Online, PREDICT, Clinical Treatment Score post-5 years
(CTS5), IHC, and CanAssist Breast.
NPI
The NPI is a widely used, validated, and clinically relevant tool for classifying patients with early breast cancer into three or more prognostic groups [98 ]. The NPI is calculated based on tumour size and grade and the number of lymph nodes involved. It is used to predict 5-year survival [99 ] and
can be used to predict benefits of adjuvant therapy [100 ]. The NPI was devised in 1982 and validated in prospective cohorts and independent multicentre
studies. It has been refined over the years to include ER and HER2 status, which are now considered essential predictive factors, to create NPI+, which has been verified in independent
European studies [100 ], [101 ].
Adjuvant! Online
The Adjuvant! Online score is an open access computer program, developed using the Surveillance, Epidemiology, and End Results (SEER) registry, that assists decision making regarding
adjuvant therapy in patients with early breast cancer by predicting 10-year risks for recurrence, breast cancer-specific mortality, and mortality due to other causes, including the
expected benefit of adjuvant systemic treatments based on patient- and tumour-related factors [102 ]. However, this model was found to overestimate [103 ], [104 ] or underestimate [105 ] OS and breast cancer-specific and event-free survival and
is no longer available for clinical use [106 ].
PREDICT
The PREDICT breast cancer prognostication and treatment benefit prediction model (V1) was developed in 2010 using cancer registry data from the UK and was based on positive attributes
of the Adjuvant! model. The model predicts 5-year and 10-year OS in early breast cancer following surgery based on patient- and tumour-related characteristics. PREDICT also provides
information on the expected benefits of chemotherapy and endocrine therapy. The PREDICT model provides a high degree of discrimination across different prognostic groups [107 ]. A newer version of PREDICT (V2) has been developed with improved calibration in patients diagnosed before 40 years of age, and both V1 and V2 versions
have been validated [108 ], [109 ]. The latest version in development at the time of writing (V2.3) incorporates PR
status, which improved model performance and provides more accurate absolute treatment benefit predictions [110 ].
CTS5
The CTS5 is a tool based on clinicopathological information such as age, tumour size, histological grade, and lymph node involvement that estimates the risk of distant recurrence after
5 years and can help with the identification of patients who could benefit from extended endocrine therapy (> 5 years); the CTS5 stratifies patients into low risk (< 5%),
intermediate risk (5 – 10%), or high risk (> 10%) of late distant recurrence [111 ]. The CTS5 has been validated in postmenopausal patients but has not
been sufficiently validated in premenopausal patients; further calibration in this population is still required. CTS5 is recommend for postmenopausal women treated for ER+, HER2− breast
cancer who are free of distant recurrence at 5 years [112 ].
IHC
IHC is an immunohistochemical score that can be calculated from three to four of the following stains: ER, PR, HER2, and Ki-67 [113 ], [114 ]. IHC scores can be combined with a clinical score that includes factors such as patientʼs age, tumour grade, and nodal burden. IHC4 has been compared
with several multigene tests and can produce similar prognostic values to some of the multigene tests, specifically in nodal-negative patients [115 ].
CanAssist Breast
CanAssist Breast was first developed in patients of Indian origin and ethnicity and validated in the same population along with a wider exposure in Caucasian patients; it combines the
data of five immunohistochemical biomarkers (CD44 , N-cadherin, pan-cadherin, ABCC4 , and ABCC11 ) with tumour size, grade, and node status to calculate
recurrence risk. CanAssist Breast predicts risk of distant recurrence at 5 years from diagnosis by segregating the patients into low- and high-risk groups for distant recurrence [116 ].
Genomic profiling for prognosis
In recent years, several genomic prognostic assays have been developed to estimate recurrence risk in breast cancer in addition to classical clinicopathological factors and to help decide
treatment in the adjuvant setting [117 ]. The prognostic information provided by these assays is complementary to standard breast cancer clinicopathological
parameters. The UK National Institute for Health and Care Excellence recommends a molecular profiling system to lower adjuvant chemotherapy rates as the calculations from multigene systems
generally identify fewer patients with a high ROR than risk calculations based on patient demographics and tumour characteristics [118 ]. A prospective
study in Canada found that, when genomic profiling was used to guide treatment decisions for patients with early breast cancer, a smaller proportion of patients actually receive
chemotherapy (42%) than the proportion who were recommended chemotherapy before genomic profiling guidance (79%), which is a decrease of 36%. Increased confidence in the adjuvant treatment
recommendations was reported for 49% of physicians and 54% of patients [119 ]. A similar study in Lebanon reported that use of genomic profiling resulted in
a treatment change in 35% of patients, with 25% of patients having de-escalation of planned therapy to avoid chemotherapy [120 ].
Although choice of adjuvant treatments in treatment guidelines can differ among countries, the association between using molecular profiling assays and a change in treatment decisions and
an overall reduction in chemotherapy use is consistent [22 ], [121 ], [122 ], [123 ]. By avoiding chemotherapy treatments, improvements in patientsʼ quality of life and overall cost savings for health systems can be achieved, despite the
negative economic impact in the short term caused by the cost of the molecular tests [124 ], [125 ].
Oncotype DX
The Oncotype DX RS, also called the 21-gene RS, is one of the most well-established genomic profiling assays. The Oncotype DX RS stratifies the 5- or 10-year risk of distant relapse
into low risk (RS 0 – 10; expected small benefit from chemotherapy), intermediate risk (RS 11 – 25; uncertain whether the beneficial effects of chemotherapy outbalances the toxic effects
of chemotherapy), and high risk (RS > 25 in women aged < 50 years, RS > 15 and high clinical risk, or RS > 21, independent of the clinical risk; high probability of cancer
recurrence, and benefits of chemotherapy should surpass the risk of treatment toxicity) [33 ]. However, it is important to note that the patient data used
to define these risk categories were from studies from several years ago. This may therefore result in overestimation of the risk in the context of contemporary clinical practice. A
study into the influence of Oncotype DX in decision making found that treatment choices were changed for approximately 25% of patients with early breast cancer, with most of these
changes to use lower-intensity treatment regimens than originally chosen [126 ], reducing adjuvant chemotherapy use in routine clinical practice [127 ]. In contrast, a comparison of Oncotype DX RS with IHC4 found that the IHC4 score provided similar prognostic information [113 ]. A separate study comparing Oncotype DX RS with IHC4, MammaTyper, NexCourse Breast, PAM50, and MammaPrint (the latter two profiles are discussed below) found that, although
these tests provide similar prognostic information at the population level, there can be notable discordance in risk categorisation when individual patients are assessed with multiple
tests, with 61% of tumours in the study classified in more than one risk group across the six different prognostic tests [128 ].
MammaPrint
MammaPrint is a 70-gene signature endorsed by several practice guidelines that classifies patients with breast cancer into high-risk (patients with risk of developing distant metastases
within 5 years after diagnosis) and low-risk (patients with a high probability of metastasis-free survival) groups [129 ]. MammaPrint has been reported to
provide similar predictions to Oncotype DX, despite focusing on different genes [130 ] and is more widely used in some European countries. In the MINDACT
trial, MammaPrint was used to identify women classed as at high risk of relapse by Adjuvant! Online who could be spared chemotherapy, discussed later in this review [22 ].
EndoPredict
EndoPredict is a 12-gene molecular signature that measures the expression of eight cancer-related genes, three RNA reference genes, and one DNA reference gene. The EndoPredict score
ranges between 0 and 15 with a cut-off score of 5 to discriminate low and high risk. The EndoPredict risk score can be combined with tumour size and nodal status to allow the calculation
of a comprehensive risk score (Epclin). EndoPredict can be used to guide treatment decisions for both chemotherapy and extended anti-hormonal therapy [131 ], although it has not been validated in phase III studies.
Prosigna/PAM50
Prosigna/PAM50 was developed in premenopausal and postmenopausal women treated without any adjuvant systemic therapy and covers 50 genes (and five reference genes). The PAM50 ROR
combines the PAM50 profile and clinical features such as tumour size and proliferation. ROR stratifies the risk of recurrence into low (≤ 40), intermediate (41 – 60), and high (> 60)
[123 ]. When the PAM50 ROR was compared with the Oncotype DX RS for the prediction of risk of distant recurrence after endocrine therapy, more patients
were scored as at high risk and fewer as at intermediate risk by ROR than by RS [15 ].
Breast Cancer Index
The Breast Cancer Index (BCI) was developed in postmenopausal patients with ER+ lymph node-negative breast cancer as a predictive test for the likelihood of benefit from extended
adjuvant endocrine therapy. It is an algorithmic gene expression-based signature comprising two functional biomarker panels. Individuals are categorised as low risk (< 5.1),
intermediate risk (5.1 to < 6.5), or high risk (≥ 6.5) [132 ], [133 ].
A secondary analysis of the Translational Study of Anastrozole or Tamoxifen Alone or Combined (TransATAC) randomised clinical trial of patients with HR+ early-stage breast cancer
treated with 5 years of tamoxifen or anastrozole compared the prognostic value of Oncotype DX RS, PAM50 ROR, BCI, Epclin, CTS, and IHC4. For 5- to 10-year recurrence, there was
prognostic value in adding PAM50 ROR, BCI, or Epclin to the CTS. However, there was no prognostic value in adding IHC4 or Oncotype DX RS to the CTS [115 ].
Further details for the recommended genomic tests used to measure ROR are presented in [Table 1 ]. Molecular profiling provides a standardised and
reproducible estimate of the ROR. However, the results do not necessarily correlate with the prognosis given by classical clinicopathological markers. The values provided by these tests
must be carefully interpreted by the clinician on a case-by-case basis rather than simply being used as an indication of high or low risk. Rather than replacing clinicopathological
markers, molecular profiling tests are complementary to classical clinicopathological markers. Therefore, the integration of genomic assays with clinicopathological prognostic factors
may be the best approach [75 ], [134 ].
Table 1 Characteristics of the most used molecular profiling systems in early breast cancer.
Profiling system
Number of genes assessed
Patient characteristics suitable for adjuvant chemotherapy assessment
Clinical application
Recommended by
Classification
Adapted from Gluz et al. [147 ] and Puppe et al. [134 ].
AGO: German Gynaecological Oncology Group, ASCO: American Society of Clinical Oncology, EP: EndoPredict score, Epclin: EndoPredict score + tumour size and nodal status, ER:
oestrogen receptor, ESMO: European Society for Medical Oncology, HER2: human epidermal growth factor receptor 2, HR: hormone receptor, IQWiG: Institute for Quality and
Efficiency in Health Care, N: lymph nodes, NCCN: National Comprehensive Cancer Network, NICE: UK National Institute for Health and Care Excellence, PR: progesterone receptor,
ROR: risk of recurrence, RS: recurrence score
Oncotype DX Recurrence Score
21 genes (16 breast cancer-related genes and five reference genes)
ER/PR+, HER2− N−
ER/PR+, HER2−, N+
Prognostic, endocrine treated, HR+, HER2−, N±
NCCN
ASCO
NICE
St. Gallen
ESMO
AGO
IQWiG
RS0 – 100:
Low risk (0 – 10)
Intermediate risk (11 – 25)
High risk (> 25 in women aged < 50 years, > 15 and high clinical risk or > 21, independent of the clinical risk)
MammaPrint
70 genes
ER/PR+, HER2−, N−
ER/PR+, HER2−, N+
Prognostic, age < 70 years, HR+, HER2−, N±
NCCN
ASCO
St. Gallen
ESMO
AGO
MammaPrint Index:
Low risk
High risk
Prosigna/PAM50
50 genes (and five reference genes)
ER/PR+, HER2−, N−
Prognostic, endocrine treated, postmenopausal, HR+, HER2−, N±
NCCN
ASCO
NICE
St. Gallen
ESMO
AGO
ROR 0 – 100:
Low risk (< 40)
Intermediate risk (41 – 60)
High risk (61 – 100)
EndoPredict
12 genes (8 cancer-related, 3 reference genes, and 1 control gene for DNA contamination)
ER/PR+, HER2−, N−
Prognostic, endocrine treated, (pre-)
postmenopausal,
HR+, HER2−, N±
NCCN
ASCO
NICE
St. Gallen
ESMO
AGO
EP molecular score 0 – 15:
Low risk (< 5)
High risk (> 5)
Epclin score (with tumour size and nodal status):
Low risk (< 3.3)
High risk (≥ 3.3)
Breast Cancer Index
Two functional biomarker panels
ER/PR+, HER2−, N−
Prognostic, endocrine treated, HR+, HER2−, N±
NCCN
ASCO
St. Gallen
AGO
Low risk (< 5.1)
Intermediate risk (5.1– < 6.5)
High risk (≥ 6.5)
Multigene tests versus non-genomic scoring systems
Currently, genomic methods are a commonly recommended method to predict ROR. However, genomic methods can be expensive and require specialised equipment. Several studies have compared IHC
scoring systems with these newer genomic profiling methods and have found that the prognostic value of non-genomic scoring systems can be equivalent to that of the modern genomic methods
and can still be used as an alternative when genomic methods are not available or affordable [135 ]. For example, patients grouped into low-, intermediate-,
and high-risk categories by Oncotype DX RS were found to have significantly different tumour grade, PR expression, and Ki-67 index between risk categories [136 ]. An analysis of the TransATAC dataset found that, for ER+, HER2− patients, late distant recurrence risk was significantly predicted by PAM50 ROR and IHC4 but not by Oncotype DX
RS [43 ].
Prospective clinical trials with multigene tests
Although OS can be considered the most clinically relevant endpoint in cancer trials, DFS is usually used as a surrogate endpoint for breast cancer [137 ], [138 ], and DFS is the most commonly reported clinical endpoint in breast cancer studies [139 ]. This is
because of the relatively long expected survival time of treated patients with breast cancer, especially the patients with ER+, HER2− disease, which makes an OS endpoint impractical in
this patient population within the timeframe of a clinical trial [140 ].
Recurrence endpoints include DFS-ductal carcinoma in situ, invasive DFS (IDFS), RFS, recurrence-free interval (RFI), breast cancer-free interval, invasive breast cancer-free survival,
distant DFS, distant relapse-free survival, and distant RFI [141 ], [142 ].
In addition to these end points, quality of life is becoming an increasingly important end point in breast cancer. Health-related quality of life is a patient-reported measure that
demonstrates clinical benefit of a treatment. Quality of life can be used as a secondary end point to compare treatments that have similar beneficial effects on disease but differences in
toxicity [138 ].
TAILORx
The predictive ability of a combination of molecular profiling with a clinicopathological factor (nodal involvement) has been assessed in clinical studies. The prospective Trial
Assigning Individualized Options for Treatment (TAILOR-X) study evaluated the classification of recurrence risk in patients with early breast cancer with HR+, HER2−, and axillary
node-negative tumours to identify patients likely to benefit from chemotherapy [106 ], [143 ]. Treatment choice was
stratified according to Oncotype DX RS. Analysis of IDFS from TAILORx showed that endocrine therapy was noninferior to adjuvant chemotherapy plus endocrine (chemoendocrine) therapy in
HR+, HER2−, and axillary node-negative patients with an RS of 11 – 25 [143 ]. Exploratory analyses indicated that women aged ≤ 50 years with an RS of
16 – 25 might benefit from chemotherapy with respect to both locoregional and distant recurrences. The trial also showed a low percentage of women with distant recurrence (3%) at 9 years
with endocrine therapy alone if the RS was < 16, irrespective of age [106 ].
MINDACT
The MINDACT trial, which included women with early breast cancer who were lymph node negative or one to three lymph nodes positive, used MammaPrint for molecular profiling. Patients
received either endocrine therapy alone or adjuvant chemotherapy plus endocrine therapy. Patients were categorised as having high or low genomic risk and high or low clinical risk. In
the case of discordant results (i.e. high clinical risk and low genomic risk, or low clinical risk and high genomic risk), patients were randomly assigned to chemotherapy or no
chemotherapy based on either the clinical result or the genomic result. The primary outcome was distant RFI. This study showed that chemotherapy could be avoided in patients at high
clinical risk for relapse according to Adjuvant! Online but who had a low genomic risk for recurrence according to MammaPrint (corresponding to 46% of patients in the MINDACT trial [22 ].
RxPonder
The RxPonder trial used Oncotype DX to determine which patients with HR+, HER2− breast cancer and nodal positivity (one to three positive lymph nodes) and an RS of 0 – 25 would benefit
from chemotherapy and which could safely avoid it. The primary objective was to determine the effect of chemotherapy on IDFS. In RxPonder, data were analysed according to menopausal
status; results showed that postmenopausal women with an RS of ≤ 25 derived no further benefit from chemotherapy added to endocrine therapy and can safely avoid adjuvant treatment with
it. On the other hand, premenopausal patients with an RS of ≤ 25 benefited from the addition of chemotherapy at 5 years [144 ].
OPTIMA
The OPTIMA study is an ongoing international randomised controlled trial comparing standard treatment with chemotherapy followed by endocrine therapy versus undergoing Prosigna testing;
participants with high Prosigna score (> 60) tumours receive standard treatment, and those with low-score tumours receive endocrine therapy alone. The investigators aim to randomise
2250 patients in each arm to demonstrate non-inferiority of test-directed treatment, defined as not more than 3% below the estimated 85% 5-year IDFS for the control arm [145 ]. To date, more than 3000 patients have been recruited [146 ].
Decision making in HR+, HER2− breast cancer
A summary of recommendations on the use of prognostic markers to guide adjuvant therapy in patients with early breast cancer is presented in [Tables 2 ] to
[4 ].
Table 4 Summary of national and international guideline recommendations on prognostic testing in early breast cancer (ER+, HER2−): prognostic scoring
system/model.
According to the European Society for Medical Oncology clinical practice guidelines for diagnosis, treatment, and follow-up in early breast cancer, the decision to start chemotherapy should
in part be based on an individualʼs risk of relapse, which depends on tumour burden and tumour biology [17 ]. Guidelines vary in their recommendations for
identification of patients at high risk of relapse, particularly regarding Ki-67 testing. To summarise this, we provide a table of national and international guideline recommendations for
prognosis in early breast cancer ([Tables 2 ] to [4 ]). This highlights that no single test provides accurate prognosis in
isolation, and a combination of tools should therefore be used.
Conclusions
Our review highlights that no single test provides accurate prognosis in isolation, and a combination of tools should therefore be used. Risk thresholds are important to guide optimised and
balanced therapeutic decisions in HR+, HER2− early breast cancer. However, prognostic assessment should be performed on a case-by-case basis, considering patient factors as well as choice of
treatment, making patient-specific prognostic approaches essential to avoid over- or undertreatment. Prediction of recurrence risk and risk thresholds to guide treatment decisions continues
to evolve with the improving availability and accuracy of prognostic tests and the increased availability of treatments with more favourable efficacy-to-adverse-event balances.
