The significance of highlighting the oestrogen receptor low category in breast cancer

Abstract

The latest ASCO/CAP guideline has recommended to report oestrogen receptor (ER) low cases (ER^lo; 1–10%) as “ER low positive category”, prompting us to compare the clinicopathologic features, biomarkers, survival and treatment of the ER^lo cases with other subgroups (ER negative (ER^neg) and ER high (ER^hi)). ER^lo cases revealed more similar clinicopathologic and biomarker profiles (including younger age, larger tumour, high proliferation, HER2 and basal markers expression) to ER^neg than ER^hi cancers. The ER^lo cases receiving hormonal therapy showed a similarly poor outcome as ER^neg cancers. However, majority of ER^lo cases were downstaged to stage I in the 8th AJCC pathological prognostic staging, highlighting a risk of potential under treatment. Overall, our data highlighted the differences of ER^lo from other ER^pos cases and their management should be considered separately.

Background

Oestrogen receptor (ER) expression as assessed by immunohistochemistry is an important predictive marker for hormonal therapy (HT) in breast cancers. Most guidelines set a positive threshold at 1% or above of any staining intensity and such patients are eligible for HT. However, there is a concern that the cases with low ER expression (ER^lo, i.e. 1–10%) may be biologically distinct from the high ER expressing cases (ER^hi, i.e. >10%) and not benefit from HT.¹ The latest ASCO/ CAP guideline has acknowledged this and recommended reporting ER^lo tumours as “ER Low Positive” category.² A standardised comment is suggested to acknowledge the limited data on the overall benefit of HT for ER^lo patients, the heterogeneity in their behaviour and biology as well as their similarity to ER-negative (ER^neg) cancers in gene profiling. Decision regarding appropriate treatment should be made after considering the totality of the information about the individual case. Here, we re-evaluated the characteristic of these ER^lo cases in comparison with ER^neg and ER^hi cases and the relevance of the new ER^lo category in breast cancer management.

Methods

The study included 1824 consecutive cases of breast cancer diagnosed between 2002 and 2009 from three involved hospitals (Prince of Wales Hospital, Kwong Wah Hospital and Tuen Mun Hospital). Baseline demographic, clinical, treatment and outcome information were retrieved from the medical records. All patients were managed according to the NCCN guidelines. Histologic features including grade, necrosis and stromal tumour infiltrating lymphocytes (sTIL) were assessed on H&E slides as described previously.³ Immunohistochemical data on routine breast markers (including ER, progesterone receptor (PR), HER2 and Ki67), basal markers (CK5/6, CK14, EGFR, p63 and c-Kit) and androgen receptor (AR) were retrieved from our previous tissue microarray analysis.³ ER expression was also verified with medical records, and all cases were categorised into ER^neg (0%), ER^lo (1–10%) and ER^hi (>10%). Disease‐free survival (DFS) and breast-cancer-specific survival (BCSS) were analysed using the Kaplan–Meier method and compared between groups using the log‐rank test. Comparison of features between different ER groups was performed using Chi‐square or Fisher’s exact test. All statistical analyses were performed in SPSS version 23.

Results

Table 1 summarised the clinical, histologic and biomarker features of the cohort. Fifty-four cases (3%) were ER^lo, 503 (27.6%) cases were ER^neg and 1266 (69.4%) cases were ER^hi. Compared to ER^hi cases, ER^lo cases were associated with larger tumour, higher grade, more necrosis, more sTIL, higher pN stage, high KI67, HER2, EGFR and CK5/6 positivity but PR and AR negativity (p ≤ 0.039). Among the ER^lo cases, only 24 cases (46.0%) had high expression of PR, much lower than the 73.7% (928/1259) in ER^hi cases. Compared to ER^neg cases, ER^lo cases were associated with higher PR but lower grade, lower expression of CK5/6 and CK14 (p ≤ 0.014). ER^lo cases were associated with more LVI (p = 0.024 and 0.011, respectively) and younger age (p = 0.005 and 0.004, respectively) than both ER^neg and ER^hi groups.

Table 1 Correlation of ER expression with clinicopathologic features and biomarkers.

Among the cases with treatment data, 14.5% (56/385), 66.0% (33/50) and 94.8% (957/1010), respectively, of the ER^neg, ER^lo and ER^hi cases received HT while 96.8% (299/309), 86.0% (43/50) and 64.8% (637/983) received chemotherapy, respectively. Most ER^neg cases received HT were PR^pos and some others had coexisting ER^pos DCIS. Some ER^lo/hi cases did not receive HT because of the higher ER diagnostic threshold applied for the earlier cases, patients’ comorbidity and refusal of treatment. Compared to ER^hi cases, more ER^lo cases received chemotherapy (p = 0.002), but less received HT (p < 0.001). Follow-up information was available for 1597 patients (median follow-up of 73 months, range 1–210 months). Among them, 3.1% and 66.1% have follow-up of ≤1 year and >5 years, respectively. The rate of relapse/mortality of ER^lo cases (25.0%) was higher than ER^hi cases (13.8%), but similar to ER^neg cases (23.8%) (Table 1). Notably, high sTIL level showed a reduced mortality rate in both ER^lo (low Vs high sTIL: 9.1 Vs 24.1%) and ER^neg (14.4 Vs 24.0%) cancers. Additionally, the DFS of ER^lo cases was significantly worse than ER^hi cases (log-rank = 5.884, p = 0.015), but comparable to ER^neg cases (log-rank = 0.020, p = 0.887) (supplementary Figure). Among patients receiving HT, those with ER^lo cancers showed a significantly worse outcome than the ER^hi cancer patients, but no difference from those ER^neg cancer patients (without HT). If AJCC staging criteria were applied, 37 of the 49 (75.5%) ER^lo cases would be downstaged by at least one substage (17 cases with ≥2 substages), and more ER^lo cases were in pathological prognostic staging (PPS) stage I (55.1% (27/49)) compared to 18.0% stage I in anatomical staging (AS), and this may pose significant prognostic implication. Among all the PPS IA cases in the current cohort, cases of ER^lo showed a significantly poorer DFS than ER^hi cases (log-rank = 3.849, p = 0.050) (supplementary Figure).

Discussion

Here, we observed a similar proportion of ER^lo cases as reported from a large series,⁴ but lower than some other smaller series.^5,6 The ER^lo cases, as reported,^4,6 showed distinct clinicopathologic and biomarker profiles from ER^hi cases. They were associated with expression of some basal markers, albeit at a lower rate than ER^neg cases, and this was concordant with previous report that most ER^lo cases were of basal molecular subtype.⁶ Thus, ER^lo cases were heterogeneous, with some cases similar to ER^neg cancers. Little apparent benefit from adjuvant HT for patients with low ER has been suggested.⁷ For treatment, it has been speculated that interventions for ER^neg tumours may be appropriate for some ER^lo cases. In fact, ER^lo cases could behave clinically like triple negative breast cancer (TNBC) in terms of pathological complete response to neoadjuvant chemotherapy.⁸ Recently, immune blockade has been approved for treating TNBC.⁹ Despite the traditional view of ER^pos cancers as non-immunogenic, the enrichment of sTIL and the associated reduced mortality within ER^lo cases may suggest an active role of antitumour immunity and the potential of immunotherapy.

Recently, ER has been incorporated into the AJCC PPS together with PR, HER2 and grade. PPS shows a superior prognostication power than the traditional TNM AS.¹⁰ In the PPS, compared to the corresponding AS, if a breast cancer expresses ER and/or HER2, it will be downstaged. There is, however, a caveat as in the AJCC guideline, ER positivity was defined as expression in 1% or more of the tumour cells, without segregation into ER^lo and ER^hi. As we demonstrated, ER^lo cases were biologically more similar to ER^neg and showed worse survival in the downstaged cases. Thus, using the approach in AJCC staging, there is a real risk of downstaging ER^lo cases that behave more like ER^neg cases biologically, resulting in potentially under treatment.

This analysis confirmed that distinct behaviour of ER^lo tumours, which are heterogeneous, with some resembling ER^neg tumours biologically (higher likelihood of being basal-like, and worse prognosis), thus lending support to the new ASCO guideline. Furthermore, downstaging as per AJCC guideline for ER^lo cases may incur a real possibility of risk underestimation and under treatment.

Limitations of the analysis included retrospective nature of the study and the small number of ER^lo cases, which limited the power of statistical analysis. Moreover, some IHC data were obtained from TMA analysis. For PPS staging, no results from OncotypeDX were available.