Controlling technical variation amongst 6693 patient microarrays of the randomized MINDACT trial.

Gene expression data obtained in large studies hold great promises for discovering disease signatures or subtypes through data analysis. It is also prone to technical variation, whose removal is essential to avoid spurious discoveries. Because this variation is not always known and can be confounded with biological signals, its removal is a challenging task. Here we provide a step-wise procedure and comprehensive analysis of the MINDACT microarray dataset. The MINDACT trial enrolled 6693 breast cancer patients and prospectively validated the gene expression signature MammaPrint for outcome prediction. The study also yielded a full-transcriptome microarray for each tumor. We show for the first time in such a large dataset how technical variation can be removed while retaining expected biological signals. Because of its unprecedented size, we hope the resulting adjusted dataset will be an invaluable tool to discover or test gene expression signatures and to advance our understanding of breast cancer.

Equivalence of MammaPrint array types in clinical trials and diagnostics.

MammaPrint is an FDA-cleared microarray-based test that uses expression levels of the 70 MammaPrint genes to assess distant recurrence risk in early-stage breast cancer. The prospective RASTER study proved that MammaPrint Low Risk patients can safely forgo chemotherapy, which is further subject of the prospective randomized MINDACT trial. While MammaPrint diagnostic results are obtained from mini-arrays, clinical trials may be performed on whole-genome arrays. Here we demonstrate the equivalence and reproducibility of the MammaPrint test. MammaPrint indices were collected for breast cancer samples: (i) on both customized certified array types (n = 1,897 sample pairs), (ii) with matched fresh and FFPE tissues (n = 552 sample pairs), iii) for control samples replicated over a period of 10 years (n = 11,333), and iv) repeated measurements (n = 280). The array type indicated a near perfect Pearson correlation of 0.99 (95 % CI: 0.989-0.991). Paired fresh and FFPE samples showed an excellent Pearson correlation of 0.93 (95 % CI 0.92-0.94), in spite of the variability introduced by intratumoral tissue heterogeneity. Control samples showed high consistency over 10 year's time (overall reproducibility of 97.4 %). Precision and repeatability are overall 98.2 and 98.3 %, respectively. Results confirm that the combination of the near perfect correlation between array types, excellent equivalence between tissue types, and a very high stability, precision, and repeatability demonstrate that results from clinical trials (such as MINDACT and I-SPY 2) are equivalent to current MammaPrint FFPE and fresh diagnostics, and can be used interchangeably.

Use of 70-gene signature to predict prognosis of patients with node-negative breast cancer: a prospective community-based feasibility study (RASTER).

BACKGROUND: A microarray-based 70-gene prognosis signature might improve the selection of patients with node-negative breast cancer for adjuvant systemic treatment. The main aims of this MicroarRAy PrognoSTics in Breast CancER (RASTER) study were to assess prospectively the feasibility of implementation of the 70-gene prognosis signature in community-based settings and its effect on adjuvant systemic treatment decisions when considered with treatment advice formulated from the Dutch Institute for Healthcare Improvement (CBO) and other guidelines. METHODS: Between January, 2004 and December, 2006, 812 women aged under 61 years with primary breast carcinoma (clinical T1-4N0M0) were enrolled. Fresh tumour samples were collected in 16 hospitals in the Netherlands within 1 h after surgery. Clinicopathological factors were collected and microarray analysis was done with a custom-designed array chip that assessed the mRNA expression index of the 70 genes previously identified for the prognostic signature. Patients with a "good" signature were deemed to have a good prognosis and, therefore, could be spared adjuvant systemic treatment with its associated adverse effects, whereas patients with a "poor" signature were judged to have a poor prognosis and should be considered for adjuvant systemic treatment. Concordance between risk predicted by the prognosis signature and risk predicted by commonly used clinicopathological guidelines (ie, St Gallen guidelines, Nottingham Prognostic Index, and Adjuvant! Online) was assessed. FINDINGS: Of 585 eligible patients, 158 patients were excluded because of sampling failure (n=128) and incorrect procedure (n=30). Prognosis signatures were assessed in 427 patients. The 70-gene prognosis signature identified 219 (51%) patients with good prognosis and 208 (49%) patients with poor prognosis. The Dutch CBO guidelines identified 184 patients (43%) with poor prognosis, which was discordant with those findings obtained with the prognosis signature in 128 (30%) patients. Oncologists recommended adjuvant treatment in 203 (48%) patients based on Dutch CBO guidelines, in 265 (62%) patients if the guidelines were used with the prognosis signature, and in 259 (61%) patients if Dutch CBO guidelines, prognosis signature, and patients' preferences for treatment were all taken into account. Adjuvant! Online guidelines identified more patients with poor prognosis than did the signature alone (294 [69%]), and discordance with the signature occurred in 160 (37%) patients. St Gallen guidelines identified 353 (83%) patients with poor prognosis with the signature and discordance in 168 (39%) patients. Nottingham Prognostic Index recorded 179 (42%) patients with poor prognosis with the signature and discordance in 117 (27%) patients. INTERPRETATION: Use of the prognosis signature is feasible in Dutch community hospitals. Adjuvant systemic treatment was advised less often when the more restrictive Dutch CBO guidelines were used compared with that finally given after use of the prognosis signature. For the other guidelines assessed, less adjuvant chemotherapy would be given when the data based on prognosis signature alone are used, which might spare patients from adverse effects and confirms previous findings. Future studies should assess whether use of the prognosis signature could improve survival or equal survival while avoiding unnecessary adjuvant systemic treatment without affecting patients' survival, and further assess the factors that physicians use to recommend adjuvant systemic treatment.