Heterogeneity in global gene expression profiles between biopsy specimens taken peri-surgically from primary ER-positive breast carcinomas.

BACKGROUND: Gene expression is widely used for the characterisation of breast cancers. Variability due to tissue heterogeneity or measurement error or systematic change due to peri-surgical procedures can affect measurements but is poorly documented. We studied the variability of global gene expression between core-cuts of primary ER+ breast cancers and the impact of delays to tissue stabilisation due to sample X-ray and of diagnostic core cutting. METHODS: Twenty-six paired core-cuts were taken immediately after tumour excision and up to 90 minutes delay due to sample X-ray; 57 paired core-cuts were taken at diagnosis and 2 weeks later at surgical excision. Whole genome expression analysis was conducted on extracted RNA. Correlations and differences were assessed between the expression of individual genes, gene sets/signatures and intrinsic subtypes. RESULTS: Twenty-three and 56 sample pairs, respectively, were suitable for analysis. The range of correlations for both sample sets were similar with the majority being >0.97 in both. Correlations between pairs for 18 commonly studied genes were also similar between the studies and mainly with Pearson correlation coefficients >0.6 except for a small number of genes, which had a narrow-dynamic range (e.g. MKI67, SNAI2). There was no systematic difference in intrinsic subtyping between the first and second sample of either set but there was c.15 % discordance between the subtype assignments between the pairs, mainly where the subtyping of individual samples was less certain. Increases in the expression of several stress/early-response genes (e.g. FOS, FOSB, JUN) were found in both studies and confirmed findings in earlier smaller studies. Increased expression of IL6, IGFBP2 and MYC (by 17 %, 14 % and 44 %, respectively) occurred between the samples taken 2 weeks apart and again confirmed findings from an earlier study. CONCLUSIONS: There is generally good correlation in gene expression between pairs of core-cuts except where genes have a narrow dynamic range. Similar correlation coefficients to the average gene expression profiles of intrinsic subtype, particularly LumA and LumB, can lead to discordances between assigned subtypes. Substantial changes in expression of early-response genes occur within an hour after surgery and in IL6, IGFB2 and MYC as a result of diagnostic core-cut biopsy. TRIAL REGISTRATION: Trial number CRUK/07/015 . Study start date September 2008.

ER and HER2 expression are positively correlated in HER2 non-overexpressing breast cancer.

INTRODUCTION: Estrogen receptor-α (ER) and human epidermal growth factor receptor 2 (HER2) positivity are inversely correlated by standard criteria. However, we investigated the quantitative relation between ER and HER2 expression at both RNA and protein levels in HER2+ve and HER2-ve breast carcinomas. METHODS: ER and HER2 levels were assessed with immunohistochemistry (IHC) and (for HER2) fluorescent in situ hybridization (FISH) and by quantitative reverse transcription-polymerase chain reaction (q-RT-PCR) in formalin-fixed primary breast cancers from 448 patients in the National Cancer Research Institute (NCRI) Adjuvant Breast Cancer Trial (ABC) tamoxifen-only arm. Relations at the RNA level were assessed in 1,139 TransATAC tumors. RESULTS: ER and HER2 RNA levels were negatively correlated as expected in HER2+ve (IHC 3+ and/or FISH-amplified) tumors (r = -0.45; P = 0.0028). However, in HER2-ve tumors (ER+ve and ER-ve combined), a significant positive correlation was found (r = 0.43; P < 0.0001), HER2 RNA levels being 1.74-fold higher in ER+ve versus ER-ve tumors. This correlation was maintained in the ER+veHER2-ve subgroup (r = 0.24; P = 0.0023) and confirmed in this subgroup in 1,139 TransATAC tumours (r = 0.25; P < 0.0001). The positive relation extended to IHC-detected ER in ABC: mean ± 95% confidence interval (CI) H-scores were 90 ± 19 and 134 ± 19 for 0 and 1+ HER2 IHC categories, respectively (P = 0.0013). A trend toward lower relapse-free survival (RFS) was observed in patients with the lowest levels of ER and HER2 RNA levels within the ER+veHER2-ve subgroup both for ABC and TransATAC cohorts. CONCLUSIONS: ER and HER2 expression is positively correlated in HER2-ve tumors. The distinction between HER2+ve and HER2-ve is greater in ER-ve than in ER+ve tumors. These findings are important to consider in clinical trials of anti-HER2 and anti-endocrine therapy in HER2-ve disease. TRIAL REGISTRATION: Clinical trial identifier: ISRCTN31514446.

Extreme loss of immunoreactive p-Akt and p-Erk1/2 during routine fixation of primary breast cancer.

INTRODUCTION: Very few studies have investigated whether the time elapsed between surgical resection and tissue fixation or the difference between core-cut and excision biopsies impact on immunohistochemically measured biomarkers including phosphorylated proteins in primary breast cancer. The aim of this study was to characterize the differences in immunoreactivity of common biomarkers that may occur (a) due to tissue handling at surgery and (b) between core-cuts and resected tumours. METHODS: Core-cuts taken from surgical breast cancer specimens immediately after resection (sample A) and after routine X-ray of the excised tumour (sample B) were formalin-fixed and paraffin-embedded and compared to the routinely fixed resection specimen (sample C). The variation in immunohistochemical expression of Ki67, oestrogen receptor (ER), progesterone receptor (PgR), human epidermal growth factor 2 (HER2), p-Akt and p-Erk were investigated. RESULTS: Twenty-one tissue sets with adequate tumour were available. Median time between collection of core-cuts A and B was 30 minutes (range 20 to 80). None of the markers showed significant differences between samples A and B. Similarly, Ki67, ER, PgR and HER2 did not differ significantly between core-cuts and main resection specimen although there was a trend for lower resection values for ER (P=0.06). However, p-Akt and p-Erk1/2 were markedly lower in resections than core-cuts (median 27 vs 101 and 69 vs 193, respectively; both P<0.0001 [two-sided]). This difference was significantly greater in mastectomy than lumpectomy specimens for p-Erk1/2 (P=0.01). CONCLUSIONS: The delay in fixation in core-cuts taken after post-operative X-ray of resection specimens has no significant impact on expression of Ki67, ER, PgR, HER2, p-Akt or p-Erk1/2. However extreme loss of phospho-staining can occur during routine fixation of resection specimens. These differences are likely attributable to suboptimal fixation and may have major repercussions for clinical research involving these markers.

Endocrine therapy, new biologicals, and new study designs for presurgical studies in breast cancer.

The preoperative setting is increasingly popular for the clinical investigation of hormonal agents and new biological drugs. The effectiveness of endocrine agents is well established for estrogen receptor-positive disease, and the emphasis in preoperative studies is on their combination with agents targeted at resistance mechanisms over 3 or more months. New agents are also being assessed for early evidence of clinical efficacy in shorter-term window-of-opportunity studies. The establishment of Ki67 as an intermediate marker of treatment benefit and of long-term outcome, with endocrine drugs, provides the opportunity for new trial designs with Ki67 as the primary endpoint. The PeriOperative Endocrine Therapy for Individualizing Care (POETIC) trial is randomizing (2:1) 4000 estrogen receptor-positive patients to 2 weeks presurgical treatment with a nonsteroidal aromatase inhibitor or no presurgical treatment. It provides a unique opportunity for detailed study of the determinants of response and resistance to estrogen deprivation as well as testing the role of presurgical therapy for improved biomarker-based estimates of prognosis.