An Evolutionary Approach for Identifying Driver Mutations in Colorectal Cancer

The traditional view of cancer as a genetic disease that can successfully be treated with drugs targeting mutant onco-proteins has motivated whole-genome sequencing efforts in many human cancer types. However, only a subset of mutations found within the genomic landscape of cancer is likely to provide a fitness advantage to the cell. Distinguishing such “driver” mutations from innocuous “passenger” events is critical for prioritizing the validation of candidate mutations in disease-relevant models. We design a novel statistical index, called the Hitchhiking Index, which reflects the probability that any observed candidate gene is a passenger alteration, given the frequency of alterations in a cross-sectional cancer sample set, and apply it to a mutational data set in colorectal cancer. Our methodology is based upon a population dynamics model of mutation accumulation and selection in colorectal tissue prior to cancer initiation as well as during tumorigenesis. This methodology can be used to aid in the prioritization of candidate mutations for functional validation and contributes to the process of drug discovery.

Evolutionary dynamic models have been intensively studied to elucidate the process of tumorigenesis. One key aspect of studying tumorigenesis is to distinguish the “driver” mutations providing a fitness advantage to cancer cells against neutral “passenger” or “hitchhiking” mutations. Many statistical models to address this question have been developed. Evolutionary models, however, add another layer of complexity by taking into account the process of mutation accumulation and selection within the tissue. Here we present a novel approach combining both statistical and evolutionary thinking to identify driver mutations in cancer genomes using cross-sectional mutation data. Our method considers the process of mutation accumulation and selection before and during colorectal cancer initiation. This work demonstrates the importance of using evolutionary population dynamic models to study driver events of tumorigenesis.

Abstract 2847: High complexity barcoding to study clonal dynamics in response to cancer therapy

Targeted therapies, such as erlotinib and imatinib, lead to dramatic clinical responses, but the emergence of resistance presents a significant challenge. Recent studies have revealed intratumoral heterogeneity as a potential source for the emergence of therapeutic resistance. However, it is still unclear if relapse/resistance is driven predominantly by pre-existing or de novo acquired alterations. To address this question, we developed a high-complexity barcode library, ClonTracer, which contains over 27 million unique DNA barcodes and thus enables the high resolution tracking of cancer cells under drug treatment. Using this library in two clinically relevant resistance models, we demonstrate that the majority of resistant clones pre-exist as rare subpopulations that become selected in response to therapeutic challenge. Furthermore, our data provide direct evidence that both genetic and non-genetic resistance mechanisms pre-exist in cancer cell populations. The ClonTracer barcoding strategy, together with mathematical modeling, enabled us to quantitatively dissect the frequency of drug-resistant subpopulations and evaluate the impact of combination treatments on the clonal complexity of these cancer models. Hence, monitoring of clonal diversity in drug-resistant cell populations by the ClonTracer barcoding strategy described here may provide a valuable tool to optimize therapeutic regimens towards the goal of curative cancer therapies.

Citation Format: Hyo-eun C. Bhang, David A. Ruddy, Viveksagar Krishnamurthy Radhakrishna, Rui Zhao, Iris Kao, Daniel Rakiec, Pamela Shaw, Marissa Balak, Justina X. Caushi, Elizabeth Ackley, Nicholas Keen, Michael R. Schlabach, Michael Palmer, William R. Sellers, Franziska Michor, Vesselina G. Cooke, Joshua M. Korn, Frank Stegmeier. High complexity barcoding to study clonal dynamics in response to cancer therapy. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 2847. doi:10.1158/1538-7445.AM2015-2847

CXCR4 Regulates Extra-Medullary Myeloma through Epithelial-Mesenchymal-Transition-like Transcriptional Activation

Extra-medullary disease (EMD) in multiple myeloma (MM) is associated with poor prognosis and resistance to chemotherapy. However, molecular alterations that lead to EMD have not been well defined. We developed bone marrow (BM)- and EMD-prone MM syngeneic cell lines; identified that epithelial-to-mesenchymal transition (EMT) transcriptional patterns were significantly enriched in both clones compared to parental cells, together with higher levels of CXCR4 protein; and demonstrated that CXCR4 enhanced the acquisition of an EMT-like phenotype in MM cells with a phenotypic conversion for invasion, leading to higher bone metastasis and EMD dissemination in vivo. In contrast, CXCR4 silencing led to inhibited tumor growth and reduced survival. Ulocuplumab, a monoclonal anti-CXCR4 antibody, inhibited MM cell dissemination, supported by suppression of the CXCR4-driven EMT-like phenotype. These results suggest that targeting CXCR4 may act as a regulator of EMD through EMT-like transcriptional modulation, thus representing a potential therapeutic strategy to prevent MM disease progression.

The Impact of Microenvironmental Heterogeneity on the Evolution of Drug Resistance in Cancer Cells

Therapeutic resistance arises as a result of evolutionary processes driven by dynamic feedback between a heterogeneous cell population and environmental selective pressures. Previous studies have suggested that mutations conferring resistance to epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKI) in non-small-cell lung cancer (NSCLC) cells lower the fitness of resistant cells relative to drug-sensitive cells in a drug-free environment. Here, we hypothesize that the local tumor microenvironment could influence the magnitude and directionality of the selective effect, both in the presence and absence of a drug. Using a combined experimental and computational approach, we developed a mathematical model of preexisting drug resistance describing multiple cellular compartments, each representing a specific tumor environmental niche. This model was parameterized using a novel experimental dataset derived from the HCC827 erlotinib-sensitive and -resistant NSCLC cell lines. We found that, in contrast to in the drug-free environment, resistant cells may hold a fitness advantage compared to parental cells in microenvironments deficient in oxygen and nutrients. We then utilized the model to predict the impact of drug and nutrient gradients on tumor composition and recurrence times, demonstrating that these endpoints are strongly dependent on the microenvironment. Our interdisciplinary approach provides a model system to quantitatively investigate the impact of microenvironmental effects on the evolutionary dynamics of tumor cells.

Optimization of radiation dosing schedules for proneural glioblastoma

Glioblastomas are the most aggressive primary brain tumor. Despite treatment with surgery, radiation and chemotherapy, these tumors remain uncurable and few significant increases in survival have been observed over the last half-century. We recently employed a combined theoretical and experimental approach to predict the effectiveness of radiation administration schedules, identifying two schedules that led to superior survival in a mouse model of the disease (Leder et al., Cell 156(3):603-616, 2014). Here we extended this approach to consider fractionated schedules to best minimize toxicity arising in early- and late-responding tissues. To this end, we decomposed the problem into two separate solvable optimization tasks: (i) optimization of the amount of radiation per dose, and (ii) optimization of the amount of time that passes between radiation doses. To ensure clinical applicability, we then considered the impact of clinical operating hours by incorporating time constraints consistent with operational schedules of the radiology clinic. We found that there was no significant loss incurred by restricting dosage to an 8:00 a.m. to 5:00 p.m. window. Our flexible approach is also applicable to other tumor types treated with radiotherapy.

DNA copy number analysis of metastatic urothelial carcinoma with comparison to primary tumors

Background

To date, there have been no reports characterizing the genome-wide somatic DNA chromosomal copy-number alteration landscape in metastatic urothelial carcinoma. We sought to characterize the DNA copy-number profile in a cohort of metastatic samples and compare them to a cohort of primary urothelial carcinoma samples in order to identify changes that are associated with progression from primary to metastatic disease.

Methods

Using molecular inversion probe array analysis we compared genome-wide chromosomal copy-number alterations between 30 metastatic and 29 primary UC samples. Whole transcriptome RNA-Seq analysis was also performed in primary and matched metastatic samples which was available for 9 patients.

Results

Based on a focused analysis of 32 genes in which alterations may be clinically actionable, there were significantly more amplifications/deletions in metastases (8.6% vs 4.5%, p < 0.001). In particular, there was a higher frequency of E2F3 amplification in metastases (30% vs 7%, p = 0.046). Paired primary and metastatic tissue was available for 11 patients and 3 of these had amplifications of potential clinical relevance in metastases that were not in the primary tumor including ERBB2, CDK4, CCND1, E2F3, and AKT1. The transcriptional activity of these amplifications was supported by RNA expression data.

Conclusions

The discordance in alterations between primary and metastatic tissue may be of clinical relevance in the era of genomically directed precision cancer medicine.

Electronic supplementary material

The online version of this article (doi:10.1186/s12885-015-1192-2) contains supplementary material, which is available to authorized users.

JAK-STAT pathway activation in malignant and nonmalignant cells contributes to MPN pathogenesis and therapeutic response

The identification of JAK2/MPL mutations in patients with myeloproliferative neoplasms (MPN) has led to the clinical development of JAK kinase inhibitors, including ruxolitinib. Ruxolitinib reduces splenomegaly and systemic symptoms in myelofibrosis and improves overall survival; however, the mechanism by which JAK inhibitors achieve efficacy has not been delineated. Patients with MPN present with increased levels of circulating proinflammatory cytokines, which are mitigated by JAK inhibitor therapy. We sought to elucidate mechanisms by which JAK inhibitors attenuate cytokine-mediated pathophysiology. Single-cell profiling demonstrated that hematopoietic cells from myelofibrosis models and patient samples aberrantly secrete inflammatory cytokines. Pan-hematopoietic Stat3 deletion reduced disease severity and attenuated cytokine secretion, with similar efficacy as observed with ruxolitinib therapy. In contrast, Stat3 deletion restricted to MPN cells did not reduce disease severity or cytokine production. Consistent with these observations, we found that malignant and nonmalignant cells aberrantly secrete cytokines and JAK inhibition reduces cytokine production from both populations.

SIGNIFICANCE

Our results demonstrate that JAK-STAT3-mediated cytokine production from malignant and nonmalignant cells contributes to MPN pathogenesis and that JAK inhibition in both populations is required for therapeutic efficacy. These findings provide novel insight into the mechanisms by which JAK kinase inhibition achieves therapeutic efficacy in MPNs.

Abstract PR16: Spatio-temporal heterogeneity in the tumor microenvironment influences the evolutionary dynamics of drug resistance

Tumor growth is a complex evolutionary process driven by dynamic feedback between a heterogeneous cell population and selection pressures from the tumor microenvironment. Spatio-temporal heterogeneity in the microenvironment can create physical niches that facilitate cellular adaptation as seen in regions of hypoxia and acidosis where cells may up-regulate glycolysis and become resistant to acid-meditated toxicity in order to survive. In recent studies, using clinically prevalent subtypes of EGFR-related non-small cell lung cancer (NSCLC), we observe that nutrient and drug gradients resulting from a cells' proximity to vasculature, can produce selective pressures driving tumor evolution. We provide a detailed examination of the microenvironmental impact (i.e. oxygen, glucose, and drug) on growth rates of NSCLC cell lines that are either sensitive or resistant to the EGFR TKI, erlotinib. Often we consider drug resistance to be associated with a fitness cost to the cell in the absence of drug. However, here we demonstrate that the situation is more complex, with the local tumor microenvironment influencing the magnitude and the directionality of the selective effect. In fact, the resistant cells actually gain a selective advantage in nutrient-stressed environments compared to the sensitive cells. The resulting growth dynamics were used to inform a stochastic compartment-based tumor model of pre-existing drug resistance where each compartment represents a specific tumor environmental niche. This integrative modeling framework was then used to predict rebound growth kinetics and tumor composition (i.e. % resistance) and in particular, provide insight into the magnitude by which the microenvironment influences these results. These investigations strongly suggest that ignoring the microenvironment or using laboratory environmental conditions to inform tumor dynamics can lead to inaccurate conclusions. Therefore, knowledge of the selective advantage/disadvantage of different cell populations within different regions of the tumor will better guide model predictions, influence overall tumor dynamics, and impact treatment strategies.

This abstract is also presented as Poster A80.

Citation Format: Shannon Mumenthaler, Jasmine Foo, Nathan Choi, William Pao, David Agus, Franziska Michor, Parag Mallick. Spatio-temporal heterogeneity in the tumor microenvironment influences the evolutionary dynamics of drug resistance. [abstract]. In: Abstracts: AACR Special Conference on Cellular Heterogeneity in the Tumor Microenvironment; 2014 Feb 26-Mar 1; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2015;75(1 Suppl):Abstract nr PR16. doi:10.1158/1538-7445.CHTME14-PR16

Abstract 2355: Risk prediction for late-stage ovarian cancer by meta-analysis of 1,525 patient samples

Background: Ovarian cancer causes over 15,000 deaths per year in the United States. The survival of patients is quite heterogeneous, and accurate prognostic tools would help with the clinical management of these patients. However, existing ovarian cancer gene signatures did not make it to the clinic so far. Expression data of thousands of patients from a large number of different cohorts is now available in the public domain. Harvesting all this information for building signatures is a challenging “big data” problem, but has the potential to yield more robust and accurate signatures.

Methods: We developed and validated two gene expression signatures of potential clinical relevance in advanced stage serous ovarian cancer, the first for predicting survival and the second for predicting the outcome of initial debulking surgery. We integrated 13 publicly available datasets totaling 1,525 subjects. We trained prediction models using a meta-analysis variation on the Compound Covariate method, tested models via a “leave-one-dataset-out” procedure, and validated models in additional independent datasets. Selected genes from the debulking signature were validated by immunohistochemistry and qRT-PCR in two further independent cohorts of 179 and 78 patients, respectively.

Results: The survival signature stratified patients into high- and low-risk groups (HR=2.19; 95% CI, 1.84 to 2.61) significantly better than the best previously published overall survival signature (P = 0.039). POSTN, CXCL14, FAP, NUAK1, PTCH1, and TGFBR2) were validated by qRT-PCR (P &lt; 0.05) and POSTN, CXCL14 and phosphorylated Smad2/3 by immunohistochemistry (P &lt; 0.001) as independent predictors of debulking status. The sum of IHC intensities for these three proteins provided a tool that classified 92.8% of samples correctly in high- and low-risk groups for suboptimal debulking (AUC 0.89; 95% CI 0.84 to 0.93). We present the strongest evidence to date for the existence of a biologic basis of suboptimal outcome of debulking surgery.

Conclusions: Our survival signature provides the most accurate and validated prognostic model for early and advanced stage high-grade serous ovarian cancer. The debulking signature accurately (92.8%) predicts the outcome of cytoreductive surgery and will have clinical utility if the accuracy of the immunohistochemistry tool observed in our initial 179-patient validation cohort is confirmed in prospective validation. This tool will potentially allow for the identification of those patients who will not benefit from primary debulking surgery, sparing them the toxicity of extensive surgery and the delay until the initiation of chemotherapy. These patients can be triaged to neoadjuvant chemotherapy with interval debulking.

Citation Format: Markus Riester, Wei Wei, Aedin C. Culhane, Lorenzo Trippa, Franziska Michor, Curtis Huttenhower, Giovanni Parmigiani, Michael Birrer. Risk prediction for late-stage ovarian cancer by meta-analysis of 1,525 patient samples. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 2355. doi:10.1158/1538-7445.AM2014-2355

Abstract LB-310: Single cell genome sequencing reveals clonal stability and diversity in breast cancer

Human breast cancers often display intratumor genomic heterogeneity. This clonal diversity confounds the clinical diagnosis and basic research of cancer, because single samples may not represent that tumor as a whole. Sequencing breast tumor cohorts en masse has identified many prevalent mutations, but has limited ability for resolving subclonal diversity. Here, we developed a whole-genome and exome single-cell sequencing approach (Nuc-Seq) using G2/M cells. To validate our method, we applied Nuc-Seq to sequence the whole genomes of two single cells from a genetically monoclonal breast cancer cell line (SK-BR-3) at high coverage depth (61X ± 5 sem, n=2) and breadth (83.70% ± 3.40 sem, n=2) to detect somatic mutations. Our analysis suggests that Nuc-Seq generates low allelic dropout rates (9.73% ± 2.19%) and low false positive error rates for point mutations (FPR = 1.24e-6). We then applied this method to sequence single normal and tumor cells from an estrogen-receptor positive breast cancer and a triple-negative ductal carcinoma at base-pair resolution. In parallel, we performed single cell copy number profiling. In both tumors, we observed a large number of rare variants that were not detected by sequencing the bulk tumor en masse. In contrast, we find that single cell copy number profiles are highly similar. Our data suggest that aneuploid rearrangements occurred early in tumor evolution and remained highly stable as the tumor mass expanded. In contrast we find that point mutations evolved gradually, generating extensive clonal diversity. Many of the diverse mutations were shown to occur at low frequencies (0.03 -10%) in the tumor mass by targeted duplex sequencing. Mathematical modeling suggests that the triple-negative tumor cells have an increased mutation rate (13.3X), while the ER+ tumor cells do not. These findings have important implications for the diagnosis, therapeutic treatment and evolution of chemoresistance in breast cancer.

Citation Format: Yong Wang, Nicholas Navin, Jill Waters, Marco Leung, Anna Unruh, Xiuqing Shi, Whijae Roh, Ken Chen, Paul Scheet, Selina Vattathil, Han Liang, Asha Multani, Hong Zhang, Funda Meric-Bernstam, Franziska Michor, Rui Zhao. Single cell genome sequencing reveals clonal stability and diversity in breast cancer. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr LB-310. doi:10.1158/1538-7445.AM2014-LB-310

Abstract 2344: 3D extracellular stiffness cues drive localized changes in gene expression

The stiffness of the extracellular matrix (ECM) stimulates mechanotransduction pathways that regulate tissue development and tumor progression. We previously showed that a stiff ECM potentiates cell growth and survival, enhances cell migration to drive tumor cell invasion, and drives malignant progression of the mammary gland in culture and in vivo, but the specific transcriptional and molecular events that occur as cells acquire these phenotypic changes are not well understood. To clarify this process, we used expression microarrays, tandem mass spectrometry, and RNA sequencing to identify changes in gene expression levels, isoform usage, and protein abundance that occur as intact ascini respond to distinct stiffness environments. We found that in stiffer ECM conditions ascini acquire consistent changes in gene expression related to cell adhesion and mRNA splicing, and specifically induce the expression of a set of genes involved in epithelial cell differentiation that includes multiple SPRR and S100 proteins. Remarkably, these genes map to an apparent stiffness-mediated transcriptional hotspot on chromosome 1q21, a region containing elevated transcription in many cancers but whose activity has not been related to mechanotransduction. We then used a heuristic approach to identify additional candidate force-mediated transcriptional hotspots throughout the genome that contain multiple genes that are coordinately activated or silenced in response to elevated ECM stiffness. Provocatively, we find that genes whose expression levels are responsive to ECM stiffness cues are disproportionately located within chromosomal regions that associate with the nuclear lamina, suggesting that these transcriptional changes may be due in part to force-dependent alteration of genomic contacts with the nuclear envelope. These studies provide biological insight into the divergent cellular responses to distinct stiffness environments and suggest that genome regulatory responses to the force environment may specifically target distinct chromosomal regions via a mechanism that remains to be elucidated.

Citation Format: Russell Bainer, Yoshihiro Yui, Shannon Mumenthaler, Parag Mallick, Ondrej Podlaha, Franziska Michor, Jan Liphardt, Jonathan Licht, Valerie Weaver. 3D extracellular stiffness cues drive localized changes in gene expression. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 2344. doi:10.1158/1538-7445.AM2014-2344

Epigenetic diversity in hematopoietic neoplasms

Tumor cell populations display a remarkable extent of variability in non-genetic characteristics such as DNA methylation, histone modification patterns, and differentiation levels of individual cells. It remains to be elucidated whether non-genetic heterogeneity is simply a byproduct of tumor evolution or instead a manifestation of a higher-order tissue organization that is maintained within the neoplasm to establish a differentiation hierarchy, a favorable microenvironment, or a buffer against changing selection pressures during tumorigenesis. Here, we review recent findings on epigenetic diversity, particularly heterogeneity in DNA methylation patterns in hematologic malignancies. We also address the implications of epigenetic heterogeneity for the clonal evolution of tumors and discuss its effects on gene expression and other genome functions in cancer.

Dynamics of chronic myeloid leukemia response to dasatinib, nilotinib, and high-dose imatinib

Treatment with the tyrosine kinase inhibitor imatinib is the standard of care for newly diagnosed patients with chronic myeloid leukemia. In recent years, several second-generation inhibitors - such as dasatinib and nilotinib - have become available: these promise to overcome some of the mutations associated with acquired resistance to imatinib. Despite eliciting similar clinical responses, the molecular effects of these agents on different subpopulations of leukemic cells remain incompletely understood. Furthermore, the consequences of using high-dose imatinib therapy have not been investigated in detail. Here we utilized clinical data from patients treated with dasatinib, nilotinib, or high-dose imatinib, together with a statistical data analysis and mathematical modeling approach, to investigate the molecular treatment response of leukemic cells to these agents. We found that these drugs elicit very similar responses if administered front-line. However, patients display significantly different kinetics when treated second-line, both in terms of differences between front-line and second-line treatment for the same drug, and among agents when used as second-line. We then utilized a mathematical framework describing the behavior of four differentiation levels of leukemic cells during therapy to predict the treatment response kinetics for the different cohorts of patients. The dynamics of BCR-ABL1 clearance observed in our study suggest that the use of standard or high-dose imatinib or a second-generation tyrosine kinase inhibitor such as nilotinib or dasatinib elicits similar responses when administered as front-line therapy for patients with chronic myeloid leukemia in chronic phase.

Evolution of acquired resistance to anti-cancer therapy

Acquired drug resistance is a major limitation for the successful treatment of cancer. Resistance can emerge due to a variety of reasons including host environmental factors as well as genetic or epigenetic alterations in the cancer cells. Evolutionary theory has contributed to the understanding of the dynamics of resistance mutations in a cancer cell population, the risk of resistance pre-existing before the initiation of therapy, the composition of drug cocktails necessary to prevent the emergence of resistance, and optimum drug administration schedules for patient populations at risk of evolving acquired resistance. Here we review recent advances towards elucidating the evolutionary dynamics of acquired drug resistance and outline how evolutionary thinking can contribute to outstanding questions in the field.

Most human non-GCIMP glioblastoma subtypes evolve from a common proneural-like precursor glioma

To understand the relationships between the non-GCIMP glioblastoma (GBM) subgroups, we performed mathematical modeling to predict the temporal sequence of driver events during tumorigenesis. The most common order of evolutionary events is 1) chromosome (chr) 7 gain and chr10 loss, followed by 2) CDKN2A loss and/or TP53 mutation, and 3) alterations canonical for specific subtypes. We then developed a computational methodology to identify drivers of broad copy number changes, identifying PDGFA (chr7) and PTEN (chr10) as driving initial nondisjunction events. These predictions were validated using mouse modeling, showing that PDGFA is sufficient to induce proneural-like gliomas and that additional NF1 loss converts proneural to the mesenchymal subtype. Our findings suggest that most non-GCIMP mesenchymal GBMs arise as, and evolve from, a proneural-like precursor.

Clonal evolution in breast cancer revealed by single nucleus genome sequencing

Sequencing studies of breast tumor cohorts have identified many prevalent mutations, but provide limited insight into the genomic diversity within tumors. Here, we developed a whole-genome and exome single cell sequencing approach called Nuc-Seq that utilizes G2/M nuclei to achieve 91% mean coverage breadth. We applied this method to sequence single normal and tumor nuclei from an estrogen-receptor positive breast cancer and a triple-negative ductal carcinoma. In parallel, we performed single nuclei copy number profiling. Our data show that aneuploid rearrangements occurred early in tumor evolution and remained highly stable as the tumor masses clonally expanded. In contrast, point mutations evolved gradually, generating extensive clonal diversity. Many of the diverse mutations were shown to occur at low frequencies (<10%) in the tumor mass by targeted single-molecule sequencing. Using mathematical modeling we found that the triple-negative tumor cells had an increased mutation rate (13.3X) while the ER+ tumor cells did not. These findings have important implications for the diagnosis, therapeutic treatment and evolution of chemoresistance in breast cancer.

Histone modifications are associated with transcript isoform diversity in normal and cancer cells

Mechanisms that generate transcript diversity are of fundamental importance in eukaryotes. Although a large fraction of human protein-coding genes and lincRNAs produce more than one mRNA isoform each, the regulation of this phenomenon is still incompletely understood. Much progress has been made in deciphering the role of sequence-specific features as well as DNA-and RNA-binding proteins in alternative splicing. Recently, however, several experimental studies of individual genes have revealed a direct involvement of epigenetic factors in alternative splicing and transcription initiation. While histone modifications are generally correlated with overall gene expression levels, it remains unclear how histone modification enrichment affects relative isoform abundance. Therefore, we sought to investigate the associations between histone modifications and transcript diversity levels measured by the rates of transcription start-site switching and alternative splicing on a genome-wide scale across protein-coding genes and lincRNAs. We found that the relationship between enrichment levels of epigenetic marks and transcription start-site switching is similar for protein-coding genes and lincRNAs. Furthermore, we found associations between splicing rates and enrichment levels of H2az, H3K4me1, H3K4me2, H3K4me3, H3K9ac, H3K9me3, H3K27ac, H3K27me3, H3K36me3, H3K79me2, and H4K20me, marks traditionally associated with enhancers, transcription initiation, transcriptional repression, and others. These patterns were observed in both normal and cancer cell lines. Additionally, we developed a novel computational method that identified 840 epigenetically regulated candidate genes and predicted transcription start-site switching and alternative exon splicing with up to 92% accuracy based on epigenetic patterning alone. Our results suggest that the epigenetic regulation of transcript isoform diversity may be a relatively common genome-wide phenomenon representing an avenue of deregulation in tumor development.

Traditionally, the regulation of gene expression was thought to be largely based on DNA and RNA sequence motifs. However, this dogma has recently been challenged as other factors, such as epigenetic patterning of the genome, have become better understood. Sparse but convincing experimental evidence suggests that the epigenetic background, in the form of histone modifications, acts as an additional layer of regulation determining how transcripts are processed. Here we developed a computational approach to investigate the genome-wide prevalence and the level of association between the enrichment of epigenetic marks and transcript diversity generated via alternative transcription start sites and splicing. We found that the role of epigenetic patterning in alternative transcription start-site switching is likely to be the same for all genes whereas the role of epigenetic patterns in splicing is likely gene-specific. Furthermore, we show that epigenetic data alone can be used to predict the inclusion pattern of an exon. These findings have significant implications for a better understanding of the regulation of transcript diversity in humans as well as the modifications arising during tumor development.

FGFR3 expression in primary and metastatic urothelial carcinoma of the bladder

While fibroblast growth factor receptor 3 (FGFR3) is frequently mutated or overexpressed in nonmuscle-invasive urothelial carcinoma (UC), the prevalence of FGFR3 protein expression and mutation remains unknown in muscle-invasive disease. FGFR3 protein and mRNA expression, mutational status, and copy number variation were retrospectively analyzed in 231 patients with formalin-fixed paraffin-embedded primary UCs, 33 metastases, and 14 paired primary and metastatic tumors using the following methods: immunohistochemistry, NanoString nCounterTM, OncoMap or Affymetrix OncoScanTM array, and Gain and Loss of Analysis of DNA and Genomic Identification of Significant Targets in Cancer software. FGFR3 immunohistochemistry staining was present in 29% of primary UCs and 49% of metastases and did not impact overall survival (P = 0.89, primary tumors; P = 0.78, metastases). FGFR3 mutations were observed in 2% of primary tumors and 9% of metastases. Mutant tumors expressed higher levels of FGFR3 mRNA than wild-type tumors (P < 0.001). FGFR3 copy number gain and loss were rare events in primary and metastatic tumors (0.8% each; 3.0% and 12.3%, respectively). FGFR3 immunohistochemistry staining is present in one third of primary muscle-invasive UCs and half of metastases, while FGFR3 mutations and copy number changes are relatively uncommon.

Risk prediction for late-stage ovarian cancer by meta-analysis of 1525 patient samples

BACKGROUND

Ovarian cancer causes more than 15000 deaths per year in the United States. The survival of patients is quite heterogeneous, and accurate prognostic tools would help with the clinical management of these patients.

METHODS

We developed and validated two gene expression signatures, the first for predicting survival in advanced-stage, serous ovarian cancer and the second for predicting debulking status. We integrated 13 publicly available datasets totaling 1525 subjects. We trained prediction models using a meta-analysis variation on the compound covariable method, tested models by a "leave-one-dataset-out" procedure, and validated models in additional independent datasets. Selected genes from the debulking signature were validated by immunohistochemistry and quantitative reverse-transcription polymerase chain reaction (qRT-PCR) in two further independent cohorts of 179 and 78 patients, respectively. All statistical tests were two-sided.

RESULTS

The survival signature stratified patients into high- and low-risk groups (hazard ratio = 2.19; 95% confidence interval [CI] = 1.84 to 2.61) statistically significantly better than the TCGA signature (P = .04). POSTN, CXCL14, FAP, NUAK1, PTCH1, and TGFBR2 were validated by qRT-PCR (P < .05) and POSTN, CXCL14, and phosphorylated Smad2/3 were validated by immunohistochemistry (P < .001) as independent predictors of debulking status. The sum of immunohistochemistry intensities for these three proteins provided a tool that classified 92.8% of samples correctly in high- and low-risk groups for suboptimal debulking (area under the curve = 0.89; 95% CI = 0.84 to 0.93).

CONCLUSIONS

Our survival signature provides the most accurate and validated prognostic model for early- and advanced-stage high-grade, serous ovarian cancer. The debulking signature accurately predicts the outcome of cytoreductive surgery, potentially allowing for stratification of patients for primary vs secondary cytoreduction.

Risk prediction for late-stage ovarian cancer by meta-analysis of 1525 patient samples

BACKGROUND

Ovarian cancer causes more than 15000 deaths per year in the United States. The survival of patients is quite heterogeneous, and accurate prognostic tools would help with the clinical management of these patients.

METHODS

We developed and validated two gene expression signatures, the first for predicting survival in advanced-stage, serous ovarian cancer and the second for predicting debulking status. We integrated 13 publicly available datasets totaling 1525 subjects. We trained prediction models using a meta-analysis variation on the compound covariable method, tested models by a "leave-one-dataset-out" procedure, and validated models in additional independent datasets. Selected genes from the debulking signature were validated by immunohistochemistry and quantitative reverse-transcription polymerase chain reaction (qRT-PCR) in two further independent cohorts of 179 and 78 patients, respectively. All statistical tests were two-sided.

RESULTS

The survival signature stratified patients into high- and low-risk groups (hazard ratio = 2.19; 95% confidence interval [CI] = 1.84 to 2.61) statistically significantly better than the TCGA signature (P = .04). POSTN, CXCL14, FAP, NUAK1, PTCH1, and TGFBR2 were validated by qRT-PCR (P < .05) and POSTN, CXCL14, and phosphorylated Smad2/3 were validated by immunohistochemistry (P < .001) as independent predictors of debulking status. The sum of immunohistochemistry intensities for these three proteins provided a tool that classified 92.8% of samples correctly in high- and low-risk groups for suboptimal debulking (area under the curve = 0.89; 95% CI = 0.84 to 0.93).

CONCLUSIONS

Our survival signature provides the most accurate and validated prognostic model for early- and advanced-stage high-grade, serous ovarian cancer. The debulking signature accurately predicts the outcome of cytoreductive surgery, potentially allowing for stratification of patients for primary vs secondary cytoreduction.

Integrative analysis of 1q23.3 copy-number gain in metastatic urothelial carcinoma

PURPOSE

Metastatic urothelial carcinoma of the bladder is associated with multiple somatic copy-number alterations (SCNAs). We evaluated SCNAs to identify predictors of poor survival in patients with metastatic urothelial carcinoma treated with platinum-based chemotherapy.

EXPERIMENTAL DESIGN

We obtained overall survival (OS) and array DNA copy-number data from patients with metastatic urothelial carcinoma in two cohorts. Associations between recurrent SCNAs and OS were determined by a Cox proportional hazard model adjusting for performance status and visceral disease. mRNA expression was evaluated for potential candidate genes by NanoString nCounter to identify transcripts from the region that are associated with copy-number gain. In addition, expression data from an independent cohort were used to identify candidate genes.

RESULTS

Multiple areas of recurrent significant gains and losses were identified. Gain of 1q23.3 was independently associated with a shortened OS in both cohorts [adjusted HR, 2.96; 95% confidence interval (CI), 1.35-6.48; P = 0.01 and adjusted HR, 5.03; 95% CI, 1.43-17.73; P < 0.001]. The F11R, PFDN2, PPOX, USP21, and DEDD genes, all located on 1q23.3, were closely associated with poor outcome.

CONCLUSIONS

1q23.3 copy-number gain displayed association with poor survival in two cohorts of metastatic urothelial carcinoma. The identification of the target of this copy-number gain is ongoing, and exploration of this finding in other disease states may be useful for the early identification of patients with poor-risk urothelial carcinoma. Prospective validation of the survival association is necessary to demonstrate clinical relevance.

Inference of tumor evolution during chemotherapy by computational modeling and in situ analysis of genetic and phenotypic cellular diversity

Cancer therapy exerts a strong selection pressure that shapes tumor evolution, yet our knowledge of how tumors change during treatment is limited. Here, we report the analysis of cellular heterogeneity for genetic and phenotypic features and their spatial distribution in breast tumors pre- and post-neoadjuvant chemotherapy. We found that intratumor genetic diversity was tumor-subtype specific, and it did not change during treatment in tumors with partial or no response. However, lower pretreatment genetic diversity was significantly associated with pathologic complete response. In contrast, phenotypic diversity was different between pre- and posttreatment samples. We also observed significant changes in the spatial distribution of cells with distinct genetic and phenotypic features. We used these experimental data to develop a stochastic computational model to infer tumor growth patterns and evolutionary dynamics. Our results highlight the importance of integrated analysis of genotypes and phenotypes of single cells in intact tissues to predict tumor evolution.