Correction: PIK3CA Genotype and a PIK3CA Mutation-Related Gene Signature and Response to Everolimus and Letrozole in Estrogen Receptor Positive Breast Cancer.

[This corrects the article DOI: 10.1371/journal.pone.0053292.].

Pathway-based subnetworks enable cross-disease biomarker discovery.

Biomarkers lie at the heart of precision medicine. Surprisingly, while rapid genomic profiling is becoming ubiquitous, the development of biomarkers usually involves the application of bespoke techniques that cannot be directly applied to other datasets. There is an urgent need for a systematic methodology to create biologically-interpretable molecular models that robustly predict key phenotypes. Here we present SIMMS (Subnetwork Integration for Multi-Modal Signatures): an algorithm that fragments pathways into functional modules and uses these to predict phenotypes. We apply SIMMS to multiple data types across five diseases, and in each it reproducibly identifies known and novel subtypes, and makes superior predictions to the best bespoke approaches. To demonstrate its ability on a new dataset, we profile 33 genes/nodes of the PI3K pathway in 1734 FFPE breast tumors and create a four-subnetwork prediction model. This model out-performs a clinically-validated molecular test in an independent cohort of 1742 patients. SIMMS is generic and enables systematic data integration for robust biomarker discovery.

Tumor PIK3CA Genotype and Prognosis in Early-Stage Breast Cancer: A Pooled Analysis of Individual Patient Data.

Purpose Phosphatidylinositol-4, 5-bisphosphate 3-kinase catalytic subunit alpha ( PIK3CA) mutations are frequently observed in primary breast cancer. We evaluated their prognostic relevance by performing a pooled analysis of individual patient data. Patients and Methods Associations between PIK3CA status and clinicopathologic characteristics were tested by applying Cox regression models adjusted for age, tumor size, nodes, grade, estrogen receptor (ER) status, human epidermal growth factor receptor 2 (HER2) status, treatment, and study. Invasive disease-free survival (IDFS) was the primary end point; distant disease-free survival (DDFS) and overall survival (OS) were also assessed, overall and by breast cancer subtypes. Results Data from 10,319 patients from 19 studies were included (median OS follow-up, 6.9 years); 1,787 patients (17%) received chemotherapy, 4,036 (39%) received endocrine monotherapy, 3,583 (35%) received both, and 913 (9%) received none or their treatment was unknown. PIK3CA mutations occurred in 32% of patients, with significant associations with ER positivity, increasing age, lower grade, and smaller size (all P < .001). Prevalence of PIK3CA mutations was 18%, 22%, and 37% in the ER-negative/HER2-negative, HER2-positive, and ER-positive/HER2-negative subtypes, respectively. In univariable analysis, PIK3CA mutations were associated with better IDFS (HR, 0.77; 95% CI, 0.71 to 0.84; P < .001), with evidence for a stronger effect in the first years of follow-up (0 to 5 years: HR, 0.73; 95% CI, 0.66 to 0.81; P < .001; 5 to 10 years: HR, 0.82; 95% CI, 0.68 to 0.99; P = .037); > 10 years: (HR, 1.15; 95% CI, 0.84 to 1.58; P = .38; P heterogeneity = .02). In multivariable analysis, PIK3CA genotype remained significant for improved IDFS ( P = .043), but not for the DDFS and OS end points. Conclusion In this large pooled analysis, PIK3CA mutations were significantly associated with a better IDFS, DDFS, and OS, but had a lesser prognostic effect after adjustment for other prognostic factors.

Tumour sampling method can significantly influence gene expression profiles derived from neoadjuvant window studies.

Patient-matched transcriptomic studies using tumour samples before and after treatment allow inter-patient heterogeneity to be controlled, but tend not to include an untreated comparison. Here, Illumina BeadArray technology was used to measure dynamic changes in gene expression from thirty-seven paired diagnostic core and surgically excised breast cancer biopsies obtained from women receiving no treatment prior to surgery, to determine the impact of sampling method and tumour heterogeneity. Despite a lack of treatment and perhaps surprisingly, consistent changes in gene expression were identified during the diagnosis-surgery interval (48 up, 2 down; Siggenes FDR 0.05) in a manner independent of both subtype and sampling-interval length. Instead, tumour sampling method was seen to directly impact gene expression, with similar effects additionally identified in six published breast cancer datasets. In contrast with previous findings, our data does not support the concept of a significant wounding or immune response following biopsy in the absence of treatment and instead implicates a hypoxic response following the surgical biopsy. Whilst sampling-related gene expression changes are evident in treated samples, they are secondary to those associated with response to treatment. Nonetheless, sampling method remains a potential confounding factor for neoadjuvant study design.

Mutational analysis of PI3K/AKT signaling pathway in tamoxifen exemestane adjuvant multinational pathology study.

PURPOSE: Deregulation of key PI3K/AKT pathway genes may contribute to endocrine resistance in breast cancer (BC). PIK3CA is the most frequently mutated gene in luminal BC (35%); however, the effect of mutations in helical versus kinase domains remains controversial. We hypothesize that improved outcomes occur in patients with estrogen receptor­positive (ER positive) BC receiving endocrine therapy and possessing PIK3CA mutations. MATERIALS AND METHODS: DNA was extracted from 4,540 formalin-fixed paraffin-embedded BC samples from the Exemestane Versus Tamoxifen-Exemestane pathology study. Mutational analyses were performed for 25 mutations (PIK3CAx10, AKT1x1, KRASx5, HRASx3, NRASx2 and BRAFx4). RESULTS: PIK3CA mutations were frequent (39.8%), whereas RAS/RAF mutations were rare (1%). In univariable analyses PIK3CA mutations were associated with significantly improved 5-year distant relapse-free survival (DRFS; HR, 0.76; 95% CI, 0.63 to 0.91; P = .003). However, a multivariable analysis correcting for known clinical and biologic prognostic factors failed to demonstrate that PIK3CA mutation status is an independent prognostic marker for DRFS (HR, 0.92; 95% CI, 0.75 to 1.12; P = .4012). PIK3CA mutations were more frequent in low-risk luminal BCs (e.g., grade 1 nodev 3, node-negative v -positive), confounding the relationship between mutations and outcome. CONCLUSION: PIK3CA mutations are present in approximately 40% of luminal BCs but are not an independent predictor of outcome in the context of endocrine therapy, whereas RAS/RAF mutations are rare inluminal BC. A complex relationship between low-risk cancers and PIK3CA mutations was identified. Although the PI3K/AKT pathway remains a viable therapeutic target as the result of ahigh mutation frequency, PIK3CA mutations do not seem to affect residual risk following treatment with endocrine therapy.

PIK3CA genotype and a PIK3CA mutation-related gene signature and response to everolimus and letrozole in estrogen receptor positive breast cancer.

The phosphatidylinositol 3' kinase (PI3K) pathway is commonly activated in breast cancer and aberrations such as PI3K mutations are common. Recent exciting clinical trial results in advanced estrogen receptor-positive (ER) breast cancer support mTOR activation is a major means of estrogen-independent tumor growth. Hence the means to identify a responsive breast cancer population that would most benefit from these compounds in the adjuvant or earlier stage setting is of high interest. Here we study PIK3CA genotype as well as a previously reported PI3K/mTOR-pathway gene signature (PIK3CA-GS) and their ability to estimate the level of PI3K pathway activation in two clinical trials of newly diagnosed ER-positive breast cancer patients- a total of 81 patients- one of which was randomized between letrozole and placebo vs letrozole and everolimus. The main objectives were to correlate the baseline PIK3CA genotype and GS with the relative change from baseline to day 15 in Ki67 (which has been shown to be prognostic in breast cancer) and phosphorylated S6 (S240) immunohistochemistry (a substrate of mTOR). In the randomized dataset, the PIK3CA-GS could identify those patients with the largest relative decreases in Ki67 to letrozole/everolimus (R = -0.43, p = 0.008) compared with letrozole/placebo (R = 0.07, p = 0.58; interaction test p = 0.02). In a second dataset of pre-surgical everolimus alone, the PIK3CA-GS was not significantly correlated with relative change in Ki67 (R = -0.11, p = 0.37) but with relative change in phosphorlyated S6 (S240) (R = -0.46, p = 0.028). PIK3CA genotype was not significantly associated with any endpoint in either datasets. Our results suggest that the PIK3CA-GS has potential to identify those ER-positive BCs who may benefit from the addition of everolimus to letrozole. Further evaluation of the PIK3CA-GS as a predictive biomarker is warranted as it may facilitate better selection of responsive patient populations for mTOR inhibition in combination with letrozole.

Relative impact of key sources of systematic noise in Affymetrix and Illumina gene-expression microarray experiments.

BACKGROUND: Systematic processing noise, which includes batch effects, is very common in microarray experiments but is often ignored despite its potential to confound or compromise experimental results. Compromised results are most likely when re-analysing or integrating datasets from public repositories due to the different conditions under which each dataset is generated. To better understand the relative noise-contributions of various factors in experimental-design, we assessed several Illumina and Affymetrix datasets for technical variation between replicate hybridisations of Universal Human Reference (UHRR) and individual or pooled breast-tumour RNA. RESULTS: A varying degree of systematic noise was observed in each of the datasets, however in all cases the relative amount of variation between standard control RNA replicates was found to be greatest at earlier points in the sample-preparation workflow. For example, 40.6% of the total variation in reported expressions were attributed to replicate extractions, compared to 13.9% due to amplification/labelling and 10.8% between replicate hybridisations. Deliberate probe-wise batch-correction methods were effective in reducing the magnitude of this variation, although the level of improvement was dependent on the sources of noise included in the model. Systematic noise introduced at the chip, run, and experiment levels of a combined Illumina dataset were found to be highly dependent upon the experimental design. Both UHRR and pools of RNA, which were derived from the samples of interest, modelled technical variation well although the pools were significantly better correlated (4% average improvement) and better emulated the effects of systematic noise, over all probes, than the UHRRs. The effect of this noise was not uniform over all probes, with low GC-content probes found to be more vulnerable to batch variation than probes with a higher GC-content. CONCLUSIONS: The magnitude of systematic processing noise in a microarray experiment is variable across probes and experiments, however it is generally the case that procedures earlier in the sample-preparation workflow are liable to introduce the most noise. Careful experimental design is important to protect against noise, detailed meta-data should always be provided, and diagnostic procedures should be routinely performed prior to downstream analyses for the detection of bias in microarray studies.

Response to mTOR inhibition: activity of eIF4E predicts sensitivity in cell lines and acquired changes in eIF4E regulation in breast cancer.

BACKGROUND: Inhibitors of the kinase mTOR, such as rapamycin and everolimus, have been used as cancer therapeutics with limited success since some tumours are resistant. Efforts to establish predictive markers to allow selection of patients with tumours likely to respond have centred on determining phosphorylation states of mTOR or its targets 4E-BP1 and S6K in cancer cells. In an alternative approach we estimated eIF4E activity, a key effector of mTOR function, and tested the hypothesis that eIF4E activity predicts sensitivity to mTOR inhibition in cell lines and in breast tumours. RESULTS: We found a greater than three fold difference in sensitivity of representative colon, lung and breast cell lines to rapamycin. Using an assay to quantify influences of eIF4E on the translational efficiency specified by structured 5'UTRs, we showed that this estimate of eIF4E activity was a significant predictor of rapamycin sensitivity, with higher eIF4E activities indicative of enhanced sensitivity. Surprisingly, non-transformed cell lines were not less sensitive to rapamycin and did not have lower eIF4E activities than cancer lines, suggesting the mTOR/4E-BP1/eIF4E axis is deregulated in these non-transformed cells. In the context of clinical breast cancers, we estimated eIF4E activity by analysing expression of eIF4E and its functional regulators within tumour cells and combining these scores to reflect inhibitory and activating influences on eIF4E. Estimates of eIF4E activity in cancer biopsies taken at diagnosis did not predict sensitivity to 11-14 days of pre-operative everolimus treatment, as assessed by change in tumour cell proliferation from diagnosis to surgical excision. However, higher pre-treatment eIF4E activity was significantly associated with dramatic post-treatment changes in expression of eIF4E and 4E-binding proteins, suggesting that eIF4E is further deregulated in these tumours in response to mTOR inhibition. CONCLUSIONS: Estimates of eIF4E activity predict sensitivity to mTOR inhibition in cell lines but breast tumours with high estimated eIF4E activity gain changes in eIF4E regulation in order to enhance resistance.

Gene expression profiling of response to mTOR inhibitor everolimus in pre-operatively treated post-menopausal women with oestrogen receptor-positive breast cancer.

There is growing evidence that uncontrolled activation of the PI3K/Akt/mTOR pathway contributes to the development and progression of breast cancer. Inhibition of this pathway has antitumour effects in preclinical studies and efficacy in combination with other agents in breast cancer patients. The aim of this study is to characterise the effects of pre-operative everolimus treatment in primary breast cancer patients and to identify potential molecular predictors of response. Twenty-seven patients with oestrogen receptor (ER)-positive breast cancer completed 11-14 days of neoadjuvant treatment with 5-mg everolimus. Core biopsies were taken before and after treatment and analysed using Illumina HumanRef-8 v2 Expression BeadChips. Changes in proliferation (Ki67) and phospho-AKT were measured on diagnostic core biopsies/resection samples embedded in paraffin by immunohistochemistry to determine response to treatment. Patients that responded to everolimus treatment with significant reductions in proliferation (fall in % Ki67 positive cells) also had significant decreases in the expression of genes involved in cell cycle (P = 8.70E-09) and p53 signalling (P = 0.01) pathways. Highly proliferating tumours that have a poor prognosis exhibited dramatic reductions in the expression of cell cycle genes following everolimus treatment. The genes that most clearly separated responding from non-responding pre-treatment tumours were those involved with protein modification and dephosphorylation, including DYNLRB2, ERBB4, PTPN13, ULK2 and DUSP16. The majority of ER-positive breast tumours treated with everolimus showed a significant reduction in genes involved with proliferation, these may serve as markers of response and predict which patients will derive most benefit from mTOR inhibition.

Correcting for intra-experiment variation in Illumina BeadChip data is necessary to generate robust gene-expression profiles.

BACKGROUND: Microarray technology is a popular means of producing whole genome transcriptional profiles, however high cost and scarcity of mRNA has led many studies to be conducted based on the analysis of single samples. We exploit the design of the Illumina platform, specifically multiple arrays on each chip, to evaluate intra-experiment technical variation using repeated hybridisations of universal human reference RNA (UHRR) and duplicate hybridisations of primary breast tumour samples from a clinical study. RESULTS: A clear batch-specific bias was detected in the measured expressions of both the UHRR and clinical samples. This bias was found to persist following standard microarray normalisation techniques. However, when mean-centering or empirical Bayes batch-correction methods (ComBat) were applied to the data, inter-batch variation in the UHRR and clinical samples were greatly reduced. Correlation between replicate UHRR samples improved by two orders of magnitude following batch-correction using ComBat (ranging from 0.9833-0.9991 to 0.9997-0.9999) and increased the consistency of the gene-lists from the duplicate clinical samples, from 11.6% in quantile normalised data to 66.4% in batch-corrected data. The use of UHRR as an inter-batch calibrator provided a small additional benefit when used in conjunction with ComBat, further increasing the agreement between the two gene-lists, up to 74.1%. CONCLUSION: In the interests of practicalities and cost, these results suggest that single samples can generate reliable data, but only after careful compensation for technical bias in the experiment. We recommend that investigators appreciate the propensity for such variation in the design stages of a microarray experiment and that the use of suitable correction methods become routine during the statistical analysis of the data.

Expression of tumor necrosis factor alpha converting enzyme in endocrine cancers.

Tumor necrosis factor alpha converting enzyme (TACE) mediates shedding of human epidermal growth factor receptor-4 (HER4). Recent data suggest that released HER4 intracellular domain (4ICD) induces apoptosis in breast cancer. TACE expression, as measured by immunohistochemical analysis, was observed in 183 of 383 breast carcinomas, 39 of 217 ovarian carcinomas, and 16 of 24 and 17 of 24 hormonesensitive and hormone-insensitive prostate carcinomas, respectively. HER4 expression was detected in breast carcinomas by using 2 antibodies recognizing an extracellular or intracellular epitope. TACE expression was predominantly seen in tumors with high levels of 4ICD and membranous HER4. Apoptotic activity was measured by the terminal deoxynucleotidyl transferase-mediated nick-end labeling (TUNEL) assay and cleaved caspase-3 staining in breast carcinomas. There was no significant association between cleaved caspase-3 or TUNEL positivity and 4ICD, whereas TUNEL positivity was seen predominantly in tumors with high levels of internalized HER4. The data presented herein show TACE expression in endocrine cancers and further support a role for TACE in breast cancer apoptosis.