Higher ER load is not associated with better outcome in stage 1–3 breast cancer: a descriptive overview of quantitative HR analysis in operable breast cancer

Breast cancer is the most common type of cancer amongst women worldwide and the leading cause of cancer-specific death for women in Europe [1]. The oestrogen receptor (ER) and progesterone receptor (PR) expression are the oldest biomarkers in breast cancer [2, 3].

Different methods exist for determining the expression of hormone receptors (HRs). The tissue can be analysed using enzyme immunoassays (EIA), in which the amount of HRs is expressed in fmol/mg, defining HR positive as 15 fmol/mg or more [4, 5]. More recently, however, immunohistochemistry (IHC) has been the preferred method of staining hormone receptors. The number of cells expressing HRs is counted, generating a percentage of positive cells [6]. Different cut-off levels are used to determine whether a tumour is considered HR positive. Usually, a tumour is considered HR positive when more than 10% of the tumour cells express HRs [7, 8].

Furthermore, nuclei can be grouped into categories of negative, weak, moderate and strong nuclear staining to generate a continuous histoscore ranging from 0 to 300, calculated by multiplying the sum of the percentage of weakly stained cells times 1, moderately stained cells times 2, and strongly stained cells times 3 [9]. Tumours with a histoscore of 50 or more are usually considered HR positive.

Additionally, the Allred scoring system has been used, which is a semi-quantitative measure that takes into consideration the proportion of positive cells (scored on a scale of 0–5) and staining intensity (scored on a scale of 0–3). The sum of these produces a score between 0 and 8, and tumours with a score of 3 or more are usually considered HR positive [10].

Another semi-quantitative measure is the ER immunoreactive score (IRS), which also relies on the proportion and intensity. This produces a score between 0 and 12, considering tumours with a score of 2 or higher HR positive [11].

Already more than 15 years ago, IHC was proposed as the reference method by different boards and peer committees [12] and a dichotomous, qualitative scale of HR expression (i.e. “positive” or “negative”) was unanimously adopted [13]. This method remains the gold standard for HR expression evaluation [14].

Although it is generally claimed that tumours with strong ER and/or PR expression are more sensitive to endocrine therapy (ET), there is no clear definition of weak or strong ER/PR expression. The most recent guidelines as proposed by the St. Gallen International Expert Consensus Conference on the Primary Therapy of Early Breast Cancer in 2017, briefly mention high ER expression as a characteristic of a low risk tumour and vice versa, but fail to provide any definition or cut-off value to determine which tumours are in fact high in ER expression [14]. As there is no consensus on the value of quantitative HR expression analysis, it is not (yet) common practice to report on HR load in the clinical setting.

This systematic review gives an overview of the methods to quantitatively assess HR load, the predictive and prognostic value of determining the HR load, gives recommendations for clinical practice and discusses future developments for HR analysis and endocrine treatment.

Many efforts have been made to identify biomarkers or profiles in breast cancer patients capable of predicting sensitivity to endocrine treatment and the risk of recurrence after treatment is discontinued [35, 36]. One of these methods is the quantitative assessment of ER and PR expression, i.e. the ER and PR load, instead of merely assigning tumours an ER and PR-positive or negative status [37].

This review concludes that in patients with an ER-positive tumour (defined as ER > 10%), a higher ER load as assessed by IHC is not correlated to better outcome, and no evidence could be found for using quantitative ER load as a prognostic marker. In other words, patients with a higher ER load (e.g. 100%) do not inherently have a better prognosis than patients with a lower ER load (e.g. 20%). Furthermore, no evidence could be found for using quantitative ER load as a predictive marker, i.e. patients with a higher ER load do not have more benefit of ET than patients with a lower ER load.

This review also concludes that in patients with an HR-positive tumour, higher PR load does not seem to be correlated to better outcome. Based on the included studies, quantitative PR load is not a suitable prognostic marker; patients with a higher PR load do not inherently have a better prognosis than patients with a lower PR load, nor is it suitable as a predictive marker. Furthermore, PR load seems to be interacted with ER load and is therefore not recognised as an independent predictor.

One of the included studies, by Esslimani-Sahla [23], found an unusually high number of recurrences and only found an association between recurrence and ER load when examining ERβ, not when examining ERα. As this is the only study that specifically examines ERβ, it is somewhat of an outlier, and its results should be interpreted with caution.

In this analysis, only studies that examined the HR expression using IHC were included. This method is the gold standard for determining HR status and other methods, such as EIA or mRNA expression profiles are not routinely used in clinical practice. Specifically, we have not focussed on articles studying EIA to determine HR status, as this method is outdated and is not routinely used in the current clinical practice. This also ensures that the included studies create a homogenous cohort. Even still, different methods were used to stain the HRs, such as staining on whole-section slides or on TMAs, though this did not seem to influence the outcome; studies staining on TMAs were not less likely to find a correlation between HR load and outcome than studies staining on whole-section slides.

Studies also differed in their way of quantitatively measuring HR load; some studies used a continuous percentage or histoscore, some studies used groups of HR-negative, low HR expression and high HR expression and some divided patients in four or more groups based on Allred score or percentage. This does have an influence on the outcome. Studies were more likely to find a positive association between HR load and outcome if a continuous score was used. However, using a continuous quantitative measure to assess HR expression is questioned by several articles. Interobserver variability is high, and samples get assigned different HR percentages depending on the pathologist and the lab it was reviewed in [38]. Most importantly, staining breast cancer tissue using IHC does not allow for precise enough measurement of HR load to generate a continuous score and can only quantify into negative, weak positive and strong positive [7, 39, 40]. The problem with this approach is defining “weak” and “strong”. A lack of generally accepted definition results in pathologists and papers choosing their own definition, making it difficult to compare multiple studies. Furthermore, and as-mentioned previously, the St. Gallen Consensus makes a distinction between high and low ER expression but fails to provide any definition or cut-off value to determine which tumours are in fact high in ER expression [14]. The St. Gallen Consensus does not mention high and low PR expression at all.

Based on the results of this review, we propose using both ER and PR expression only as a qualitative measure; defining tumours with < 1–10% of cells expressing this receptor as negative, and tumours with more than 1–10% of cells expressing the receptor as positive [17, 41]. Using a continuous quantitative measure does not seem feasible without centralised, unambiguous and clear pathological measurement. The implications for the daily clinical practise of pathologists are that more detailed information on the HR status beyond “positive” or “negative” should no longer be provided, to prevent oncologists subconsciously or instinctively making different treatment decisions based on this information. Since there is no evidence for different treatment strategies, providing extra information is both unnecessary and undesirable.

Simultaneously, one can speculate whether there is any added value of measuring the PR status at all. It is generally accepted that there is no such thing as an ER-negative/PR-positive tumour [42, 43]. Since the quantitative PR load is correlated to the quantitative ER load and PR load is inversely correlated to the histological grade of the tumour [19, 24], the question arises whether PR status provides any additional prognostic information, when ER status, grade and potentially a proliferation factor such as ki-67 is known. Likewise, when examining guidelines on adjuvant treatment, they do not propose different treatment strategies for tumours that are ER positive/PR positive compared to tumours that are ER positive/PR negative [8, 14]. Therefore, if the PR status is unlikely to change the course of treatment, it could be considered wasteful and excessive to continue measuring it [44, 45]. It might be worthwhile to focus future research on the independent contribution of PR status using multivariable models.

Gene expression profiling can be used to identify two inherently different entities within breast cancer, known as luminal-A and luminal-B. These are intrinsic molecular subtypes that reflect a different tumour biology and disease prognosis. Unfortunately, neither subtype is predictive for a better response to ET [29, 46‐50]. Moreover, the gene expression profiles used to differentiate between these subtypes are expensive and not universally available and the added value for daily clinical practice, in particular for which subgroup of patients, is still debated [35]. For these reasons, researchers have tried to approach the distinction between these molecular subtypes using IHC, which resulted in subtypes called luminal-A-like and luminal-B-like [29]. When defining IHC-based luminal-A-like tumours as HR positive, HER2 negative and ki-67 below 14%, approximately 81–85% of luminal-A tumours were correctly identified as luminal-A-like. However, approximately 35–52% of luminal-B tumours were incorrectly identified as luminal-A-like. When expanding the definition of luminal-A-like to HR positive, HER2 negative, ki-67 below 14% and PR above 20%, the specificity improves somewhat but not enough to accurately discriminate between the two subtypes [29, 48, 50].

With these considerations, and the lack of prognostic and predictive value of IHC assessed quantitative ER and PR load as shown in this review, the distinction between IHC-based luminal-A-like and luminal-B-like tumours should not be used to tailor treatment decisions for women with HR-positive stage 1–3 breast cancer.

There is no clear evidence for using quantitative ER and PR load assessed by immunohistochemistry as a prognostic measure nor as a predictive marker for response to ET in patients with stage 1–3 breast cancer. Immunohistochemistry is the gold standard for measuring HR status but should only be used to distinguish HR-negative and HR-positive tumours. Gene expression profiles have prognostic value for women with ER-positive disease, early response evaluation to neoadjuvant therapy holds promise in the prediction of long-term response to endocrine therapy.