Expression-based Classification Research Articles

Analytical Performance of a Formalin-fixed Paraffin-embedded Tissue-based 634-probe Prognostic Assay for Predicting Outcome of Patients With Stage II Colon Cancer

A formalin-fixed paraffin-embedded tissue-based prognostic assay to assess the risk for recurrence in stage II colon cancer has recently been clinically validated. This study describes the analytical performance and quality control measures of the assay. The reportable range was determined to be [-1.129, 1.414] in risk score units. The accuracy was evaluated with a split sample comparison within the production lab and between the production lab and a reference lab. The concordance between the replicates within the production lab was 79% (95% confidence interval, 64%-91%). There was no evidence of bias, and the concordance was 78% (95% confidence interval, 61%-90%) between the labs. The lab-to-lab concordance was further evaluated by simulating risk scores from the full reportable range. The simulation suggested a higher concordance. The sensitivity study demonstrated that the percentage of tumor tissue did not impact the risk score and that RNA concentration of 9.5 ng/μL was a conservative determination of the analyte lower limit of quantification. From the precision study, the repeatability and reproducibility estimates were 0.1267 and 0.0548 in risk score units, respectively. Furthermore, multifaceted quality control measures were implemented, such as proper tissue processing steps, high-risk and low-risk controls, nontemplate control, and a gene expression-based classifier to evaluate the cDNA amplification kit, a key reagent in the assay. In conclusion, this study demonstrates the strong analytical performance of the assay and further supports its use as an objective standardized prognostic test for stage II colon cancer.

Learning regular expressions for clinical text classification

Natural language processing (NLP) applications typically use regular expressions that have been developed manually by human experts. Our goal is to automate both the creation and utilization of regular expressions in text classification. We designed a novel regular expression discovery (RED) algorithm and implemented two text classifiers based on RED. The RED+ALIGN classifier combines RED with an alignment algorithm, and RED+SVM combines RED with a support vector machine (SVM) classifier. Two clinical datasets were used for testing and evaluation: the SMOKE dataset, containing 1091 text snippets describing smoking status; and the PAIN dataset, containing 702 snippets describing pain status. We performed 10-fold cross-validation to calculate accuracy, precision, recall, and F-measure metrics. In the evaluation, an SVM classifier was trained as the control. The two RED classifiers achieved 80.9-83.0% in overall accuracy on the two datasets, which is 1.3-3% higher than SVM's accuracy (p<0.001). Similarly, small but consistent improvements have been observed in precision, recall, and F-measure when RED classifiers are compared with SVM alone. More significantly, RED+ALIGN correctly classified many instances that were misclassified by the SVM classifier (8.1-10.3% of the total instances and 43.8-53.0% of SVM's misclassifications). Machine-generated regular expressions can be effectively used in clinical text classification. The regular expression-based classifier can be combined with other classifiers, like SVM, to improve classification performance.

A Novel Subtype-Classifier Of Pediatric Acute Lymphoblastic Leukemia Using Gexp Multiplexed System

▪ ObjectiveAcute lymphoblastic leukemia (ALL) is the most frequent malignant neoplasm in children. The lack of reasonable and accurate classification has becomes the main obstructive factor, which makes normative treatment more difficult to carry out. Morphology, immunology, cytogenetics and molecular biology (MICM) classification is widely used clinically for pediatric leukemia. However, it is a time-consuming and expensive process and only available in a few major medical centers in some developing countries, which heavily impact the accurate classification. Our previous microarray analysis of gene expression profiles in 100 pediatric ALL bone marrow samples screened out 62 classification markers (mapped to 61 ENTREZ genes), which could classify pediatric ALL into 6 major ALL subtypes. In this study, the GenomeLab Gene Expression Profiler Genetic Analysis System (GeXP) was applied here to custom design a multiplex of 57-gene classifier to validate the feasibility of these genes as classification markers, and then establish a new and practical method for ALL classification, which will guide the risk classification and stratified treatment of leukemia. MethodsSixty-two classification markers were divided into 3 panels with each panel containing 20∼21 genes. Each primer is a chimeric primer consists of a gene-specific sequence fused to a universal tag sequence at the 5' terminal. The GeXP technology was used to measure the expression levels of these genes. The peak areas of each target gene and endogenous reference controls were normalized, from which, the relative quantification of each gene in a sample was determined. For each sample, we ranked the gene expression values from low to high and normalized these rank values by Z-score. Super k-means method was used to do the cluster by genes and samples. ResultsA novel gene expression-based ALL subtype classification using GeXP multiplexed assay was optimized and developed, which led to a rapid, reliable, and cost-effective classifier. Using a subset of 61 genes, we totally got 57 marker genes' expression data in 85 pediatric ALL samples. From the cluster results, 1) we got 4 clusters that mainly represented by 4 subtypes of ALL, including TEL-AML1+ ALL, BCR-ABL+ ALL, E2A-PBX1+ ALL and T-cell ALL, which obtained a high accuracy (98.4%, 60/61) with MICM classification (Fig 1); 2) we found the marker genes for each subtype and especially for those samples without any subtype, which provided the prediction according to which cluster it located in; 3) this classifier could take a single sample and made a prediction based solely on the relative expression ranks among the marker genes. ConclusionsWe develop a novel multiplexed 57-gene classifier to identify the major subtypes of pediatric ALL. This promising molecular test allows for a high-throughput, robust and reproducible assessment of multiplexed gene expression analysis. It offers a powerful diagnostic tool for the rapid classification using a minimal amount of samples. It may be usefully applied in the future clinical work and assist the physician in a more rapid and accurate classification of the subtypes in pediatric ALL. [Display omitted] Disclosures:No relevant conflicts of interest to declare.

Modeling the next generation sequencing sample processing pipeline for the purposes of classification.

BackgroundA key goal of systems biology and translational genomics is to utilize high-throughput measurements of cellular states to develop expression-based classifiers for discriminating among different phenotypes. Recent developments of Next Generation Sequencing (NGS) technologies can facilitate classifier design by providing expression measurements for tens of thousands of genes simultaneously via the abundance of their mRNA transcripts. Because NGS technologies result in a nonlinear transformation of the actual expression distributions, their application can result in data that are less discriminative than would be the actual expression levels themselves, were they directly observable.ResultsUsing state-of-the-art distributional modeling for the NGS processing pipeline, this paper studies how that pipeline, via the resulting nonlinear transformation, affects classification and feature selection. The effects of different factors are considered and NGS-based classification is compared to SAGE-based classification and classification directly on the raw expression data, which is represented by a very high-dimensional model previously developed for gene expression. As expected, the nonlinear transformation resulting from NGS processing diminishes classification accuracy; however, owing to a larger number of reads, NGS-based classification outperforms SAGE-based classification.ConclusionsHaving high numbers of reads can mitigate the degradation in classification performance resulting from the effects of NGS technologies. Hence, when performing a RNA-Seq analysis, using the highest possible coverage of the genome is recommended for the purposes of classification.

Highlights from Recent Cancer Literature

Highlights from Recent Cancer Literature

Predicting time to ovarian carcinoma recurrence using protein markers

Patients with ovarian cancer are at high risk of tumor recurrence. Prediction of therapy outcome may provide therapeutic avenues to improve patient outcomes. Using reverse-phase protein arrays, we generated ovarian carcinoma protein expression profiles on 412 cases from TCGA and constructed a PRotein-driven index of OVARian cancer (PROVAR). PROVAR significantly discriminated an independent cohort of 226 high-grade serous ovarian carcinomas into groups of high risk and low risk of tumor recurrence as well as short-term and long-term survivors. Comparison with gene expression-based outcome classification models showed a significantly improved capacity of the protein-based PROVAR to predict tumor progression. Identification of protein markers linked to disease recurrence may yield insights into tumor biology. When combined with features known to be associated with outcome, such as BRCA mutation, PROVAR may provide clinically useful predictions of time to tumor recurrence.

A novel treatment stratification system for current low- and intermediate-risk neuroblastoma patients using gene expression-based classification

To improve risk estimation of low- and intermediate-risk neuroblastoma patients, we determined gene expression profiles of 709 primary neuroblastoma samples using 44 K oligonucleotide-microarrays. A prognostic gene expression-based classifier was generated using a training set of 75 samples and evaluated on the remaining 634 test set patients. In the low- and intermediate subgroup of patients (n = 413), the classifier predicted event-free (EFS) and overall survival (OS) accurately (EFS, 0.846 ± 0.020 vs. 0.376 ± 0.071; OS 0.996 ± 0.004 vs. 0.699 ± 0.070; both p < 0.001). In univariate Cox regression models based on EFS and OS, the classifier was a significant prognostic marker (EFS, hazard ratio 5.0, 95% CI 3.2 to 7.8; OS, hazard ratio 40.5, 95% CI 11.8 to 139.3; both p < 0.001). We therefore suggest to implementing the classifier in the upcoming German neuroblastoma trial. Based on subgroup analyses, we propose to reduce treatment intensity of favourably classified patients currently considered to be at intermediate risk, and to intensify treatment of unfavourably classified patients > 18 months at diagnosis with localized disease and patients with stage 4S neuroblastoma.

The transcription factor activating protein 2 beta (TFAP2B) mediates neuronal differentiation in neuroblastoma cells

We here aimed to investigate the role of the transcription factor activating protein 2 beta (TFAP2B) in neuroblastoma pathogenesis. TFAP2 is a family of transcription factors that play a crucial role in embryonic differentiation and development. Oligonucleotide-microarray analysis of 649 neuroblastoma specimens demonstrated that TFAP2B downregulation is associated with unfavourable prognostic markers such as disseminated stage 4, age > 18 months, MYCN amplification, unfavourable gene expression-based classification and adverse patient outcome.

Use of the 92-gene assay to identify tumor type and direct predictive biomarker testing in limited tissue specimens.

e19072 Background: Limited tissue availability from biopsies can limit comprehensive diagnostic testing and individualized care. The 92-gene assay is a gene expression-based cancer classifier that requires a minimal number of tumor cells (~300) and may improve the efficiency of tissue usage as a molecular complement to IHC in cases with limited diagnostic material. In this study, the reporting rate and utility of molecular classification in limited tissue cytology specimens, as well as the frequency of directed predictive biomarker testing were examined. Methods: Four hundred thirty eight FFPE cytopathology cases (ie, fine needle aspirates, pleural effusions, ascites) with indeterminate diagnosis where the 92-gene assay (CancerTYPE ID, bioTheranostics Inc.) was used in clinical care were included in the analysis. The 92-gene assay reporting rate and tissue availability to perform downstream biomarker testing were analyzed. Results: The 92-gene assay provided a molecular diagnosis in 86% (n=378) of cytopathology specimens submitted for diagnostic testing. Twenty-four distinct tumor types were predicted including: 9% lung (n=41), 8% urinary bladder (n=34), 8% pancreaticobiliary (n=34), 8% ovary (n=34), 7% gastroesophageal (n=32), 5% intestine (n=20), and 1% melanoma (n=5). In 66 patients with molecular diagnoses where reflex biomarker testing was applicable (lung, CRC, melanoma), 33% of cases had additional predictive biomarker testing performed, and 2 or more biomarker tests were performed on 15% of cases. The most common tests ordered were EGFR mutations (n=16), KRAS mutations (n=8), BRAF mutation (n=8), and ALK rearrangement (n=5). Conclusions: These findings demonstrate the suitability and high reporting rate of the 92-gene assay in limited tissue cytologic specimens and successful predictive biomarker testing in appropriate cases. With the growing number of molecularly-targeted therapies requiring predictive biomarker testing, use of the 92-gene assay may have increased utility in indeterminate cancer cases where limited tissue specimens are the only option for diagnosis and treatment planning.

Hox-C9 activates the intrinsic pathway of apoptosis and is associated with spontaneous regression in neuroblastoma

Neuroblastoma is an embryonal malignancy of the sympathetic nervous system. Spontaneous regression and differentiation of neuroblastoma is observed in a subset of patients, and has been suggested to represent delayed activation of physiologic molecular programs of fetal neuroblasts. Homeobox genes constitute an important family of transcription factors, which play a fundamental role in morphogenesis and cell differentiation during embryogenesis. In this study, we demonstrate that expression of the majority of the human HOX class I homeobox genes is significantly associated with clinical covariates in neuroblastoma using microarray expression data of 649 primary tumors. Moreover, a HOX gene expression-based classifier predicted neuroblastoma patient outcome independently of age, stage and MYCN amplification status. Among all HOX genes, HOXC9 expression was most prominently associated with favorable prognostic markers. Most notably, elevated HOXC9 expression was significantly associated with spontaneous regression in infant neuroblastoma. Re-expression of HOXC9 in three neuroblastoma cell lines led to a significant reduction in cell viability, and abrogated tumor growth almost completely in neuroblastoma xenografts. Neuroblastoma growth arrest was related to the induction of programmed cell death, as indicated by an increase in the sub-G1 fraction and translocation of phosphatidylserine to the outer membrane. Programmed cell death was associated with the release of cytochrome c from the mitochondria into the cytosol and activation of the intrinsic cascade of caspases, indicating that HOXC9 re-expression triggers the intrinsic apoptotic pathway. Collectively, our results show a strong prognostic impact of HOX gene expression in neuroblastoma, and may point towards a role of Hox-C9 in neuroblastoma spontaneous regression.

Linkage of Genomic Biomarkers to Whole Organism End Points in a Toxicity Identification Evaluation (TIE)

Aquatic organisms are exposed to many toxic chemicals and interpreting the cause and effect relationships between occurrence and impairment is difficult. Toxicity Identification Evaluation (TIE) provides a systematic approach for identifying responsible toxicants. TIE relies on relatively uninformative and potentially insensitive toxicological end points. Gene expression analysis may provide needed sensitivity and specificity aiding in the identification of primary toxicants. The current work aims to determine the added benefit of integrating gene expression end points into the TIE process. A cDNA library and a custom microarray were constructed for the marine amphipod Ampelisca abdita. Phase 1 TIEs were conducted using 10% and 40% dilutions of acutely toxic sediment. Gene expression was monitored in survivors and controls. An expression-based classifier was developed and evaluated against control organisms, organisms exposed to low or medium toxicity diluted sediment, and chemically selective manipulations of highly toxic sediment. The expression-based classifier correctly identified organisms exposed to toxic sediment even when little mortality was observed, suggesting enhanced sensitivity of the TIE process. The ability of the expression-based end point to correctly identify toxic sediment was lost concomitantly with acute toxicity when organic contaminants were removed. Taken together, this suggests that gene expression enhances the performance of the TIE process.

Abstract P5-01-03: Discordant Estrogen and Progesterone Receptor Status in Breast Cancer

Abstract Introduction: Hormone receptor status is used to guide clinical therapy in breast cancer. Estrogen receptor (ER) assays from pathology reports have been shown to be in agreement with ER measured by a centralized laboratory on tissue microarray 87% of the time (Kappa = 0.64); however, little is known of reproducibility within subsets of patients defined by ER/PR status (1). Methods: We computed percent agreement and Kappa statistics for ER/PR status as reported in pathology reports from Nurse's Health Study (NHS) participants with breast cancer from 1976 to 2000, and as scored by immunohistochemistry (IHC) on tissue microarrays at a central laboratory (n = 2011). Statistics were computed separately for 4 subsets of patients stratified by ER/PR status as defined by pathology reports (ER+/PR+, ER+/PR−, ER−/PR+, ER−/PR−). We also used a curated compilation of publicly available gene expression data sets providing both IHC and microarray data on ER and PR expression (n = 1236). Results: We observed significant heterogeneity in Kappa values across the four groups of patients, with ER+/PR+ and ER−/PR- showing the strongest agreement, ER+/PR- showing fair agreement, and ER−/PR+ showing no significant agreement: 89.1% of ER+/PR+ cases from medical record were ER+/PR+ by TMA (κ=.60), 69.4% of ER−/PR- cases from medical record were ER−/PR- by TMA (κ=.63), 42% of ER+/PR- cases from medical record were ER+/PR- by TMA (κ=.37), and only 6% of cases of ER−/PR+ cases from medical record were ER−/PR+ by TMA (κ=.06). Using publicly available microarray data, we observed similar results comparing IHC to gene expression data in the same samples. Expression values were scaled across patients, and a patient was classified as positive for the marker by microarray if the scaled expression value was greater than zero. The agreement between IHC data and gene expression-based classification was 50.0% (k = .50), 80.8% (κ=.54), 42.0% (κ=.15), and 7.2% (κ = −.006) for ER+/PR+, ER−/PR−, ER+/PR−, and ER−/PR+, respectively. Conclusion: ER−/PR+ is by far the most rare and least reproducible subset of breast cancer cases. Given the lack of reproducibility of the ER−/PR+ group and the minimal reported benefit from hormonal therapy in these patients (2), these data question the clinical utility of assessment of PR, particularly in ER− breast cancer. 1. Collins LC, Marotti JD, Baer HJ, Tamimi RM. Comparison of estrogen receptor results from pathology reports with results from central laboratory testing. Journal of the National Cancer Institute. 2008;100(3):218–21. 2. Anderson H, Hills M, Zabaglo L, A'Hern R, Leary AF, Haynes BP, et al. Relationship between estrogen receptor, progesterone receptor, HER-2 and Ki67 expression and efficacy of aromatase inhibitors in advanced breast cancer. Annals of oncology: official journal of the European Society for Medical Oncology/ESMO. 2011;22(8):1770–6. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P5-01-03.

SOCS1 Mutation Subtypes Predict Divergent Outcomes in DLBCL Patients

AbstractAbstract 419Suppressor of cytokine signaling 1 (SOCS1) is frequently mutated in Hodgkin, primary mediastinal and diffuse large B-cell lymphomas (DLBCL). In the primary mediastinal B-cell lymphoma line MedB-1, mutated SOCS1 abnormally stabilizes phospho-JAK2, thereby enhances STAT signaling leading to continuous proliferation. Here, we evaluated the prognostic value of SOCS1 mutations by full-length gene sequencing of SOCS1 in 154 comprehensively characterized DLBCL cases. By sequence analysis, we identified 90 SOCS1 mutations in 16% of lymphomas. We defined two distinct subtypes with respect to putative mutational consequences: those predicting the full-length (minor) and a truncated protein (major), respectively. Neither the SOCS1 mutation group, nor minor/major subgroups can be distinguished by clinical phenotype; however, assignment of four established expression-based classifiers revealed significant associations of SOCS1 major cases with germinal center- and specific pathway activation pattern signatures. Above all, SOCS1 major cases had an excellent overall survival, even better than the GCB-like subgroup (see Figure). SOCS1 minor cases had a dismal survival, even worse than the ABC gene signature group (see Figure). SOCS1 mutation subsets retain prognostic significance in uni- and multivariate analyses. Thus, if a SOCS1 mutation is present, the mutation type is an important single gene prognostic biomarker in DLBCL. [Display omitted] Disclosures:No relevant conflicts of interest to declare.

Detection of disseminated tumor cells in the bone marrow of breast cancer patients using multiplex gene expression measurements identifies new therapeutic targets in patients at high risk for the development of metastatic disease

Disseminated tumor cells (DTCs) detected in the bone marrow (BM) of breast cancer patients identify women at high risk of recurrence. DTCs are traditionally detected by immunocytochemical staining for cytokeratins or single gene expression measurements, which limit both specificity and sensitivity. We evaluated the Nanostring nCounter™ platform for multi-marker, gene expression-based detection and classification of DTCs in the BM of breast cancer patients. Candidate genes exhibiting tumor cell-specific expression were identified from microarray datasets and validated by qRT-PCR analysis in non-malignant human BM and identical samples spiked with predefined numbers of molecularly diverse breast tumor cell lines. Thirty-eight validated transcripts were designed for the nCounter™ platform and a subset of these transcripts was technically validated against qRT-PCR measurements using identical spiked BM controls. Bilateral iliac crest BM aspirates were collected and analyzed from twenty breast cancer patients, prior to neoadjuvant therapy, using the full 38-gene nCounter™ code set. Tumor cell-specific gene expression by nCounter™ was detected with a sensitivity of one cancer cell per 1 × 10(6) nucleated BM cells after optimization. Measurements were quantitative, log linear over a 20-fold range, and correlated with qRT-PCR measurements. Using the nCounter™ 38-gene panel, 6 of 8 patients (75 %) who developed metastatic disease had detectable expression of at least one transcript. Notably, three of these patients had detectable expression of ERBB2 in their BM, despite the fact that their corresponding primary tumors were HER2/ERBB2 negative and therefore did not receive trastuzumab therapy. Four of these patients also expressed the PTCH1 receptor, a newly recognized therapeutic target based on hedgehog signaling pathway inhibition. The presumptive detection and classification of DTCs in the BM of breast cancer patients, based on sensitive and quantitative multi-marker detection of gene expression using the nCounter™ platform, provide an opportunity to both predict early distant recurrence and, more importantly, identify opportunities for preventing the spread of disease based on the expression of unique, therapeutically actionable gene targets. This study demonstrates the application of a new technology for multiplexed gene expression-based detection of DTCs in the BM of breast cancer patients and identifies at least two therapeutically targetable genes that are frequently expressed in the BM of patients who develop metastatic disease.

Gene expression-based diagnostics for molecular cancer classification of difficult to diagnose tumors

Standardized methods for accurate tumor classification are of critical importance for cancer diagnosis and treatment, particularly in diagnostically-challenging cases where site-directed therapies are an option. Molecular diagnostics for tumor classification, subclassification and site of origin determination based on advances in gene expression profiling have translated into clinical practice as complementary approaches to clinicopathological evaluations. In this review, the foundational science of gene expression-based cancer classification, technical and clinical considerations for clinical translation, and an overview of molecular signatures of tumor classification that are available for clinical use will be discussed. Proposed approaches will also be described for further integration of molecular tests for cancer classification into the diagnostic paradigm using a tissue-based strategy as a key component to direct evaluation. Increasing evidence of improved patient outcomes with the application of site and molecularly-targeted cancer therapy through use of molecular tools highlights the growing potential for these gene expression-based diagnostics to positively impact patient management. Looking forward, the availability of adequate tissue will be a significant issue and limiting factor as cancer diagnosis progresses; when the tumor specimen is limited, use of molecular classification may be a reasonable early step in the evaluation, particularly if the tumor is poorly-differentiated and has atypical features.

Identification of a 14-gene signature that predicts survival in colorectal cancer with liver metastasis.

e14037 Background: Colorectal cancer (CRC) is the third leading cause of cancer death in men and women combined in the USA. Colorectal liver metastasis (CLM), the most common metastasis of CRC, accounts for at least two thirds of CRC deaths. The purpose of this study is to identify a gene signature that predicts patient survival in patients with CLM. Methods: We analyzed the gene expression profiles of specimens from 24 CLM patients (M:F = 14:10) who underwent metastatic liver resection and unmatched primary CRC specimens from an independent cohort of 30 patients (M:F = 14:16). The association between gene expression levels and survival outcome was evaluated using Cox proportional hazards regression. Random Forests and risk scores were used to construct a gene expression-based survival classifier. Results: Based on survival classifier of CLM patients, a 14-gene signature was developed. According to leave-one-out cross validation, all 24 CLM patients (median follow-up time of 25 months) were correctly assigned into high-risk or low-risk groups (p=0.001). This 14-gene signature was then validated in an independent cohort of 30 primary CRC patients (median follow-up time of 42.2 months; p= 0.03) and a subset of 11 patients who were diagnosed at presentation or follow-up with liver metastasis (M:F = 5:6; median follow-up time of 27.6 months; p=0.04). Conclusions: We have identified a 14-gene signature that predicts the survival of CLM patients after liver resection, with validation in an independent cohort. Although sample size is small, the significance level achieved with our survival analysis warrants further investigation of this 14-gene signature in a larger sample size.

Identification of a Poor-PrognosisBRAF-Mutant–Like Population of Patients With Colon Cancer

Our purpose was development and assessment of a BRAF-mutant gene expression signature for colon cancer (CC) and the study of its prognostic implications. A set of 668 stage II and III CC samples from the PETACC-3 (Pan-European Trails in Alimentary Tract Cancers) clinical trial were used to assess differential gene expression between c.1799T>A (p.V600E) BRAF mutant and non-BRAF, non-KRAS mutant cancers (double wild type) and to construct a gene expression-based classifier for detecting BRAF mutant samples with high sensitivity. The classifier was validated in independent data sets, and survival rates were compared between classifier positive and negative tumors. A 64 gene-based classifier was developed with 96% sensitivity and 86% specificity for detecting BRAF mutant tumors in PETACC-3 and independent samples. A subpopulation of BRAF wild-type patients (30% of KRAS mutants, 13% of double wild type) showed a gene expression pattern and had poor overall survival and survival after relapse, similar to those observed in BRAF-mutant patients. Thus they form a distinct prognostic subgroup within their mutation class. A characteristic pattern of gene expression is associated with and accurately predicts BRAF mutation status and, in addition, identifies a population of BRAF mutated-like KRAS mutants and double wild-type patients with similarly poor prognosis. This suggests a common biology between these tumors and provides a novel classification tool for cancers, adding prognostic and biologic information that is not captured by the mutation status alone. These results may guide therapeutic strategies for this patient segment and may help in population stratification for clinical trials.

C-KIT receptor expression is strictly associated with the biological behaviour of thyroid nodules

BackgroundA large amount of information has been collected on the molecular tumorigenesis of thyroid cancer. A low expression of c-KIT gene has been reported during the transformation of normal thyroid epithelium to papillary carcinoma suggesting a possible role of the gene in the differentiation of thyroid tissue rather than in the proliferation. The initial presentation of thyroid carcinoma is through a nodule and the best way nowadays to evaluate it is by fine-needle aspiration (FNA). However many thyroid FNAs are not definitively benign or malignant, yielding an indeterminate or suspicious diagnosis which ranges from 10 to 25% of FNAs. BRAF mutational analysis is commonly used to assess the malignancy of thyroid nodules but unfortunately it still leaves indeterminate diagnoses. The development of molecular initial diagnostic tests for evaluating a thyroid nodule is needed in order to define optimal surgical approach for patients with uncertain diagnosis pre- and intra-operatively.MethodsIn this study we extracted RNA from 82 FNA smears, 46 malignant and 36 benign at the histology, in order to evaluate by quantitative Real Time PCR the expression levels of c-KIT gene.ResultsWe have found a highly preferential decrease rather than increase in transcript of c-KIT in malignant thyroid lesions compared to the benign ones. To explore the diagnostic utility of c-KIT expression in thyroid nodules, its expression values were divided in four arbitrarily defined classes, with class I characterized by the complete silencing of the gene. Class I and IV represented the two most informative groups, with 100% of the samples found malignant or benign respectively. The molecular analysis was proven by ROC (receiver operating characteristic) analysis to be highly specific and sensitive improving the cytological diagnostic accuracy of 15%.ConclusionWe propose the use of BRAF test (after uncertain cytological diagnosis) to assess the malignancy of thyroid nodules at first, then the use of the c-KIT expression to ultimately assess the diagnosis of the nodules that otherwise would remain suspicious. The c-KIT expression-based classification is highly accurate and may provide a tool to overcome the difficulties in today's preoperative diagnosis of thyroid suspicious malignancies.

Sepsis in transit: from clinical to molecular classification

In the previous issue of Critical Care, Maslove and colleagues studied circulating neutrophil transcriptional expression to discover and validate a molecular subclassification of adult patients with sepsis. The authors divided patients into small derivation (n = 55) and validation (n = 71) cohorts. Their complex methodology included partitioning around medoid and hierarchical clustering methods to define two transcriptionally distinct subtypes of sepsis. Pathway analysis found that chemokine and cytokine pathways as well as Toll-like receptor signaling were enhanced. Investigation of specific drug target genes relevant to sepsis found significantly different expression levels between the two molecular subtypes. Interestingly, most patient characteristics did not differ between groups, except for an increase in the proportion of severe sepsis in molecular subtype 1. Possible confounders of this study were the small sample size, population stratification, and lack of information regarding drug interventions, all of which support the need for more studies with larger cohorts that include transcriptional profiles. This thought-provoking hypothesis-generating study could lead to a new neutrophil expression-based molecular classification of adult sepsis.

Molecular heterogeneity of luminal breast cancer

Luminal (or clinically defined as estrogen receptor (ER)-positive and Her2-negative) breast cancer has long been successfully treated with anti-estrogen therapy, the first targeted anti-cancer agent in breast cancer. Recently, molecular profiling approaches have allowed better identification of a poor prognostic subgroup; however, the biological mechanisms which contribute this phenotype are currently unclear. With regards to defining prognosis, it is clear that proliferation markers can clearly separate ER+/HER2- breast cancer into at least two prognostic groups. Immunohistochemistry using Ki67 at the protein level and prognostic gene signatures such as Mammaprint™, the 21-gene Recurrence Score, the two-gene ratio and genomic grade all provide quantitative measurements of proliferation activity. However, a biologically relevant cut-off does not exist. Molecular subtype definitions using PAM50 or other gene expression-based classifiers do not provide a more concordant or reproducible luminal A or B definition. Improved definition and clinical management of luminal subtypes will come from an increased understanding of the molecular drivers of the phenotype. Lately, PIK3CA and AKT1 mutations have been shown to be associated with the good prognosis luminal A phenotype whilst FGFR1 and ZNF703 amplifications are responsible for about 25% of the luminal B phenotype. It is hoped that new genomic technologies such as next-generation sequencing will offer new insights into the biology of ER-positive breast cancer. Recent next-generation sequencing studies have identified MAP3K1, ATR and MYST3 mutations in around 10% of ER+ breast cancer which may be associated with de novo endocrine therapy resistance. These, if shown to be driver oncogenes, may shed new light on the biology of endocrine nonresponsive breast cancer and inspire new treatment strategies. Finally, recent results from the BOLERO-2 trial suggest that metastatic ER+ disease may be effectively treated with the addition of a mTORC1 inhibitor, which suggests for many patients with acquired endocrine therapy resistance, mTOR pathway activation plays a significant part in their tumor biology.