Combining biological tumour markers into surrogate molecular subtypes has been shown to add prognostic information [
6‐
8,
10,
11] which may be of importance for recommendation of systemic therapy. Unlike the analyses of individual biomarkers in the present cohort of patients [
17] which showed high concordance between primary tumours and corresponding lymph node metastases, the molecular subtypes identify a subgroup of patients with ER positive disease as luminal B, with a worse prognosis, who may benefit from adjuvant chemotherapy alongside endocrine treatment [
10,
14,
20]. We found the prognosis according to the molecular subtypes to be superior for the luminal A subtype in primary tumours as well as in synchronous lymph node metastases. A subset of patients shifting from a luminal A subtype in the primary tumour to a non-luminal A subtype in the metastatic lymph node can constitute a subgroup where adjuvant chemotherapy would have improved prognosis.
The present cohort, with patients included in two prospective trials of adjuvant tamoxifen, was initiated decades ago. The distribution of molecular subtypes in the primary tumour is, however, similar to today’s distribution with 13% of the tumours being HER2 overexpressed and more than 50% having a luminal A phenotype [
11]. The finding of a shift in molecular subtype from the primary tumour to the metastases is thus not necessarily influenced by the draw-backs of including a cohort not offered modern treatment. The prognosis, however, is dependent not only on the phenotype of the tumour and the metastases, but also on the calendar-period including the treatment offered at that time. Survival analyses were adjusted also for calendar-period with similar results. The study only includes 85 patients and is not powered to find any difference in presentation of four molecular subtypes in the primary tumour
versus metastases. Hence, the shift of a molecular subtype towards a more aggressive subtype in the metastatic lymph node is a hypothesis-generating finding in line with recent publications [
21]. In the recently published study from our group [
22] comparison of molecular subtypes in primary tumour and synchronous lymph node metastases also revealed a shift in individual patients. The shift was observed from luminal A to non-luminal A in the metastatic node as well as the reversed shift, from non-luminal A to luminal A in the metastatic node. In the present study, only shifts to a molecular subtype with worse prognosis were observed. The number of patients in the present study cohort is limited (N = 85) and the inclusion was restricted to patients with stage II breast cancer whom all received adjuvant treatment with tamoxifen irrespective of expression of ER as opposed to the patients in the more recent study [
22] which constitutes an unselected cohort where patients were offered adjuvant treatment according to modern guidelines. The analyses of HER2 also differ between the studies, where assessment according to immunohistochemistry (IHC) (present) or silver
in situ hybridization (SISH) [
22] could affect the results. Interestingly, shifts are observed in individual patients in both patient cohorts according to molecular subtypes, proposing a molecular event in the metastatic niche during tumour cell progression with influence on prognosis.
Tissue analysis
The individual biomarker discordance between primary tumours and metastases may reflect tumour progression, although test artefacts have also been proposed. For HER2 analysis, a recent meta-analysis including 26 primary publications has suggested that limitations of test reproducibility are less likely to explain the discordance in HER2 status found between primary tumours and metastatic sites [
23]. The authors found a low HER2 discordant proportion for synchronous lymph node metastases compared to metachronous distant metastasis, supporting that tumour progression plays a major role. In the present study, biopsies were obtained by a manual arrayer from lymph nodes corresponding to the primary tumour and further processed as described previously for analyses of HER2 and Ki67. The method has limitations because a small area of one of the metastatic lymph nodes is examined. Sampling may therefore contribute to bias in representative areas of evaluation.
In the present study, 8/85 tumours were classified as HER2 2+ according to IHC analyses. In a national survey performed by our group, 12% of HER2 2+ tumours were amplified according to fluorescence
in situ hybridization FISH [
24] and in another study [
25] the concordance was up to 24%. This would result in 1–2 patients of HER2 2+ tumours as amplified in the present cohort, thus patients with HER2 2+ tumours were included as HER2-negative.
Cut-off values
The previously defined cut-off values for biomarker expression are based on accepted guidelines [
26,
27] in which Ki67 is the least studied with few validated guidelines available. In the present study, representative areas for the TMAs were examined to identify cancerous regions within a tissue sample. Areas in the region with increased number of Ki67 positive cells, hot spots, were identified and the number of positive cells was assessed and index calculated. The present study used a predefined 20% cut-off point based on the population sectioning, distinguishing the one third of the patients in the population with the highest proliferation from the remaining two thirds [
28,
29]. The prognostic value of Ki67 has been investigated in several recent publications [
6,
28,
30] but the assessment of the cut-off value of Ki67 is not settled and the reliability of the measures varies in different geographic settings [
10]. The cut-off value of ER responsiveness in clinical practice is traditionally 10%. This cut-point was chosen also in the present study, although there is support for a lower cut-off value of 1% for endocrine treatment and thus the detection of any ER positive cell in the tumour will define it as an ER responsive tumour [
10]. ASCO/PAP guidelines support the 1% cut-off [
27] but the guidelines are questioned in a recent study [
31].
The results in this study indicate tumour instability in clinically used markers in combination classified according to the St Gallen molecular subtypes between primary breast cancer and synchronous matched lymph node metastases. Furthermore, the survival analyses show that the St. Gallen molecular subtypes have similar prognostic implications in primary tumours and matched lymph node metastases. Node status is still a powerful prognostic factor in primary breast cancer despite advanced molecular techniques. A shift in molecular characteristics to a more aggressive phenotype in synchronous nodal metastases compared to the primary tumour suggests that tumour progression occurs already at time of diagnosis in a fraction of breast cancer patients with node positive disease. The selection of more aggressive cell clones in lymph node metastases can be an additional explanation to the prognostic information gained by nodal involvement in primary breast cancer, besides a more advanced stage of the disease.