multiSNV: a probabilistic approach for improving detection of somatic point mutations from multiple related tumour samples

Somatic variant analysis of a tumour sample and its matched normal has been widely used in cancer research to distinguish germline polymorphisms from somatic mutations. However, due to the extensive intratumour heterogeneity of cancer, sequencing data from a single tumour sample may greatly underestimate the overall mutational landscape. In recent studies, multiple spatially or temporally separated tumour samples from the same patient were sequenced to identify the regional distribution of somatic mutations and study intratumour heterogeneity. There are a number of tools to perform somatic variant calling from matched tumour-normal next-generation sequencing (NGS) data; however none of these allow joint analysis of multiple same-patient samples. We discuss the benefits and challenges of multisample somatic variant calling and present multiSNV, a software package for calling single nucleotide variants (SNVs) using NGS data from multiple same-patient samples. Instead of performing multiple pairwise analyses of a single tumour sample and a matched normal, multiSNV jointly considers all available samples under a Bayesian framework to increase sensitivity of calling shared SNVs. By leveraging information from all available samples, multiSNV is able to detect rare mutations with variant allele frequencies down to 3% from whole-exome sequencing experiments.     

Tumour class prediction and discovery by microarray-based DNA methylation analysis

Aberrant DNA methylation of CpG sites is among the earliest and most frequent alterations in cancer. Several studies suggest that aberrant methylation occurs in a tumour type-specific manner. However, large-scale analysis of candidate genes has so far been hampered by the lack of high throughput assays for methylation detection. We have developed the first microarray-based technique which allows genome-wide assessment of selected CpG dinucleotides as well as quantification of methylation at each site. Several hundred CpG sites were screened in 76 samples from four different human tumour types and corresponding healthy controls. Discriminative CpG dinucleotides were identified for different tissue type distinctions and used to predict the tumour class of as yet unknown samples with high accuracy using machine learning techniques. Some CpG dinucleotides correlate with progression to malignancy, whereas others are methylated in a tissue-specific manner independent of malignancy. Our results demonstrate that genome-wide analysis of methylation patterns combined with supervised and unsupervised machine learning techniques constitute a powerful novel tool to classify human cancers.     

Variation in Cell Signaling Protein Expression May Introduce Sampling Bias in Primary Epithelial Ovarian Cancer

Although the expression of cell signaling proteins is used as prognostic and predictive biomarker, variability of protein levels within tumors is not well studied. We assessed intratumoral heterogeneity of protein expression within primary ovarian cancer. Full-length proteins were extracted from 88 formalin-fixed and paraffin-embedded tissue samples of 13 primary high-grade serous ovarian carcinomas with 5–9 samples each. In addition, 14 samples of normal fallopian tube epithelium served as reference. Quantitative reverse phase protein arrays were used to analyze the expression of 36 cell signaling proteins including HER2, EGFR, PI3K/Akt, and angiogenic pathways as well as 15 activated (phosphorylated) proteins. We found considerable intratumoral heterogeneity in the expression of proteins with a mean coefficient of variation of 25% (range 17–53%). The extent of intratumoral heterogeneity differed between proteins (p<0.005). Interestingly, there were no significant differences in the extent of heterogeneity between phosphorylated and non-phosphorylated proteins. In comparison, we assessed the variation of protein levels amongst tumors from different patients, which revealed a similar mean coefficient of variation of 21% (range 12–48%). Based on hierarchical clustering, samples from the same patient clustered more closely together compared to samples from different patients. However, a clear separation of tumor versus normal tissue by clustering was only achieved when mean expression values of all individual samples per tumor were analyzed. While differential expression of some proteins was detected independently of the sampling method used, the majority of proteins only demonstrated differential expression when mean expression values of multiple samples per tumor were analyzed. Our data indicate that assessment of established and novel cell signaling proteins as diagnostic or prognostic markers may require sampling of serous ovarian cancers at several distinct locations to avoid sampling bias.     

qpure: A Tool to Estimate Tumor Cellularity from Genome-Wide Single-Nucleotide Polymorphism Profiles

Tumour cellularity, the relative proportion of tumour and normal cells in a sample, affects the sensitivity of mutation detection, copy number analysis, cancer gene expression and methylation profiling. Tumour cellularity is traditionally estimated by pathological review of sectioned specimens; however this method is both subjective and prone to error due to heterogeneity within lesions and cellularity differences between the sample viewed during pathological review and tissue used for research purposes. In this paper we describe a statistical model to estimate tumour cellularity from SNP array profiles of paired tumour and normal samples using shifts in SNP allele frequency at regions of loss of heterozygosity (LOH) in the tumour. We also provide qpure, a software implementation of the method. Our experiments showed that there is a medium correlation 0.42 (-value = 0.0001) between tumor cellularity estimated by qpure and pathology review. Interestingly there is a high correlation 0.87 (-value 2.2e-16) between cellularity estimates by qpure and deep Ion Torrent sequencing of known somatic KRAS mutations; and a weaker correlation 0.32 (-value = 0.004) between IonTorrent sequencing and pathology review. This suggests that qpure may be a more accurate predictor of tumour cellularity than pathology review. qpure can be downloaded from https://sourceforge.net/projects/qpure/.     

Proteome analysis of membrane fractions in colorectal carcinomas by using 2D-DIGE saturation labeling

Because of its high content in receptors and signaling proteins, the analysis of membrane fractions is critical for the study of neoplastic diseases as colorectal cancer. Here, we have used the new saturation labeling for 2D-DIGE analysis of the membrane proteome of colorectal cancer mucosal tissues. Samples from 6 patients (tumoral and normal paired biopsies) were included in this study. Twelve analytical gels were performed and considered for the quantitative study and statistical analysis. A spot pattern analysis, by using an unsupervised clustering algorithm, allowed the classification of the samples according to similar expression patterns in tumoral and normal samples. Those proteins whose expression changed significantly (Student's t-test, p < 0.05) were further digested and characterized by mass spectrometry. Among the differentially expressed proteins: annexin A2, annexin A4, annexin A5, annexin A7, lamin B, calponin 1 and VDAC were analyzed by immunohistochemistry using tissue microarrays. Annexin A2, annexin A4 and VDAC appear as potential markers of interest for colorectal cancer diagnosis and, presumably, therapy. In summary, saturation labeling provides a new and sensitive tool for the analysis of scarce amounts of samples, allowing sample classification and direct identification of deregulated proteins.     

Comparative expression analysis of four breast cancer subtypes versus matched normal tissue from the same patients

Gene expression studies have been widely used in an effort to identify signatures that can predict clinical progression of cancer. In this study we focused instead on identifying gene expression differences between breast tumors and adjacent normal tissue, and between different subtypes of tumor classified by clinical marker status. We have collected a set of 20 breast cancer tissues, matched with the adjacent pathologically normal tissue from the same patient. The cancer samples representing each subtype of breast cancer identified by estrogen receptor ER(+/-) and Her2(+/-) status and divided into four subgroups (ER+/Her2+, ER+/Her2-, ER-/Her2+, and ER-/Her2-) were hybridized on Affymetrix HG-133 Plus 2.0 microarrays. By comparing cancer samples with their matched normal controls we have identified 3537 overall differentially expressed genes using data analysis methods from Bioconductor. When we looked at the genes in common of the four subgroups, we found 151 regulated genes, some of them encoding known targets for breast cancer treatment. Unique genes in the four subgroups instead suggested gene regulation dependent on the ER/Her2 markers selection. In conclusion, the results indicate that microarray studies using robust analysis of matched tumor and normal samples from the same patients can be used to identify genes differentially expressed in breast cancer tumor subtypes even when small numbers of samples are considered and can further elucidate molecular features of breast cancer.     

Multiplex detection of surface molecules on colorectal cancers

A technique of fluorescence multiplexing is described for analysis of the plasma membrane proteome of colorectal cancer cells from surgically resected specimens, enabling detection and immunophenotyping when the cancer cells are in the minority. A single-cell suspension was prepared from a colorectal tumour, and the mixed population of cells was captured on a CD antibody microarray. The cancer cells were detected using a fluorescently tagged antibody for carcinoembryonic antigen (CEA-Alexa647) or epithelial cell adhesion marker (EpCAM-Alexa488). Using this multiplexing procedure, dot patterns from colorectal cancers were distinct from those of adjacent normal tissue. Subtraction of the expression levels for each antigen from normal tissue from those for the cancer shows differential expression in the cancer of CD66c, CD15s, CD55, CD45, CD71, CD45RO, CD11b and CEA, in descending order. Cells captured on the same microarray were also labelled with fluorescent CD3-phycoerythrin antibody revealing the presence of tumour-infiltrating lymphocytes. The immunophenotypes of T lymphocytes from the tumour samples showed differential expression of HLA-DR, TCR alpha/beta, CD49d, CD52, CD49e, CD5, CD95, CD28, CD38 and CD71, in descending order. Fluorescence multiplexing of mixed cell populations captured on a single antibody microarray enables expression profiling of multiple sub-populations of cells within a tumour sample.     

Heterogeneity mapping of protein expression in tumors using quantitative immunofluorescence

Morphologic heterogeneity within an individual tumor is well-recognized by histopathologists in surgical practice. While this often takes the form of areas of distinct differentiation into recognized histological subtypes, or different pathological grade, often there are more subtle differences in phenotype which defy accurate classification (Figure 1). Ultimately, since morphology is dictated by the underlying molecular phenotype, areas with visible differences are likely to be accompanied by differences in the expression of proteins which orchestrate cellular function and behavior, and therefore, appearance. The significance of visible and invisible (molecular) heterogeneity for prognosis is unknown, but recent evidence suggests that, at least at the genetic level, heterogeneity exists in the primary tumor(1,2), and some of these sub-clones give rise to metastatic (and therefore lethal) disease. Moreover, some proteins are measured as biomarkers because they are the targets of therapy (for instance ER and HER2 for tamoxifen and trastuzumab (Herceptin), respectively). If these proteins show variable expression within a tumor then therapeutic responses may also be variable. The widely used histopathologic scoring schemes for immunohistochemistry either ignore, or numerically homogenize the quantification of protein expression. Similarly, in destructive techniques, where the tumor samples are homogenized (such as gene expression profiling), quantitative information can be elucidated, but spatial information is lost. Genetic heterogeneity mapping approaches in pancreatic cancer have relied either on generation of a single cell suspension(3), or on macrodissection(4). A recent study has used quantum dots in order to map morphologic and molecular heterogeneity in prostate cancer tissue(5), providing proof of principle that morphology and molecular mapping is feasible, but falling short of quantifying the heterogeneity. Since immunohistochemistry is, at best, only semi-quantitative and subject to intra- and inter-observer bias, more sensitive and quantitative methodologies are required in order to accurately map and quantify tissue heterogeneity in situ. We have developed and applied an experimental and statistical methodology in order to systematically quantify the heterogeneity of protein expression in whole tissue sections of tumors, based on the Automated QUantitative Analysis (AQUA) system(6). Tissue sections are labeled with specific antibodies directed against cytokeratins and targets of interest, coupled to fluorophore-labeled secondary antibodies. Slides are imaged using a whole-slide fluorescence scanner. Images are subdivided into hundreds to thousands of tiles, and each tile is then assigned an AQUA score which is a measure of protein concentration within the epithelial (tumor) component of the tissue. Heatmaps are generated to represent tissue expression of the proteins and a heterogeneity score assigned, using a statistical measure of heterogeneity originally used in ecology, based on the Simpson's biodiversity index(7). To date there have been no attempts to systematically map and quantify this variability in tandem with protein expression, in histological preparations. Here, we illustrate the first use of the method applied to ER and HER2 biomarker expression in ovarian cancer. Using this method paves the way for analyzing heterogeneity as an independent variable in studies of biomarker expression in translational studies, in order to establish the significance of heterogeneity in prognosis and prediction of responses to therapy.     

Detergent addition to tryptic digests and ion mobility separation prior to MS/MS improves peptide yield and protein identification for in situ proteomic investigation of frozen and formalin‐fixed paraffin‐embedded adenocarcinoma tissue sections

The identification of proteins involved in tumour progression or which permit enhanced or novel therapeutic targeting is essential for cancer research. Direct MALDI analysis of tissue sections is rapidly demonstrating its potential for protein imaging and profiling in the investigation of a range of disease states including cancer. MALDI‐mass spectrometry imaging (MALDI‐MSI) has been used here for direct visualisation and in situ characterisation of proteins in breast tumour tissue section samples. Frozen MCF7 breast tumour xenograft and human formalin‐fixed paraffin‐embedded breast cancer tissue sections were used. An improved protocol for on‐tissue trypsin digestion is described incorporating the use of a detergent, which increases the yield of tryptic peptides for both fresh frozen and formalin‐fixed paraffin‐embedded tumour tissue sections. A novel approach combining MALDI‐MSI and ion mobility separation MALDI‐tandem mass spectrometry imaging for improving the detection of low‐abundance proteins that are difficult to detect by direct MALDI‐MSI analysis is described. In situ protein identification was carried out directly from the tissue section by MALDI‐MSI. Numerous protein signals were detected and some proteins including histone H3, H4 and Grp75 that were abundant in the tumour region were identified.     

SNP arrays in heterogeneous tissue: highly accurate collection of both germline and somatic genetic information from unpaired single tumor samples

SNP arrays provide reliable genotypes and can detect chromosomal aberrations at a high resolution. However, tissue heterogeneity is currently a major limitation for somatic tissue analysis. We have developed SOMATICs, an original program for accurate analysis of heterogeneous tissue samples. Fifty-four samples (42 tumors and 12 normal tissues) were processed through Illumina Beadarrays and then analyzed with SOMATICs. We demonstrate that tissue heterogeneity-related limitations not only can be overcome but can also be turned into an advantage. First, admixture of normal cells with tumor can be used as an internal reference, thereby enabling highly sensitive detection of somatic deletions without having corresponding normal tissue. Second, the presence of normal cells allows for discrimination of somatic from germline aberrations, and the proportion of cells in the tissue sample that are harboring the somatic events can be assessed. Third, relatively early versus late somatic events can also be distinguished, assuming that late events occur only in subsets of cancer cells. Finally, admixture by normal cells allows inference of germline genotypes from a cancer sample. All this information can be obtained from any cancer sample containing a proportion of 40-75% of cancer cells. SOMATICs is a ready-to-use open-source program that integrates all of these features into a simple format, comprehensively describing each chromosomal event.     

Corruption of the Intra-Gene DNA Methylation Architecture Is a Hallmark of Cancer

Epigenetic processes - including DNA methylation - are increasingly seen as having a fundamental role in chronic diseases like cancer. It is well known that methylation levels at particular genes or loci differ between normal and diseased tissue. Here we investigate whether the intra-gene methylation architecture is corrupted in cancer and whether the variability of levels of methylation of individual CpGs within a defined gene is able to discriminate cancerous from normal tissue, and is associated with heterogeneous tumour phenotype, as defined by gene expression. We analysed 270985 CpGs annotated to 18272 genes, in 3284 cancerous and 681 normal samples, corresponding to 14 different cancer types. In doing so, we found novel differences in intra-gene methylation pattern across phenotypes, particularly in those genes which are crucial for stem cell biology; our measures of intra-gene methylation architecture are a better determinant of phenotype than measures based on mean methylation level alone (K-S test in all 14 diseases tested). These per-gene methylation measures also represent a considerable reduction in complexity, compared to conventional per-CpG beta-values. Our findings strongly support the view that intra-gene methylation architecture has great clinical potential for the development of DNA-based cancer biomarkers.     

Multi-class clustering of cancer subtypes through SVM based ensemble of pareto-optimal solutions for gene marker identification

With the advancement of microarray technology, it is now possible to study the expression profiles of thousands of genes across different experimental conditions or tissue samples simultaneously. Microarray cancer datasets, organized as samples versus genes fashion, are being used for classification of tissue samples into benign and malignant or their subtypes. They are also useful for identifying potential gene markers for each cancer subtype, which helps in successful diagnosis of particular cancer types. In this article, we have presented an unsupervised cancer classification technique based on multiobjective genetic clustering of the tissue samples. In this regard, a real-coded encoding of the cluster centers is used and cluster compactness and separation are simultaneously optimized. The resultant set of near-Pareto-optimal solutions contains a number of non-dominated solutions. A novel approach to combine the clustering information possessed by the non-dominated solutions through Support Vector Machine (SVM) classifier has been proposed. Final clustering is obtained by consensus among the clusterings yielded by different kernel functions. The performance of the proposed multiobjective clustering method has been compared with that of several other microarray clustering algorithms for three publicly available benchmark cancer datasets. Moreover, statistical significance tests have been conducted to establish the statistical superiority of the proposed clustering method. Furthermore, relevant gene markers have been identified using the clustering result produced by the proposed clustering method and demonstrated visually. Biological relationships among the gene markers are also studied based on gene ontology. The results obtained are found to be promising and can possibly have important impact in the area of unsupervised cancer classification as well as gene marker identification for multiple cancer subtypes.     

Quantification of Normal Cell Fraction and Copy Number Neutral LOH in Clinical Lung Cancer Samples Using SNP Array Data

Background

Technologies based on DNA microarrays have the potential to provide detailed information on genomic aberrations in tumor cells. In practice a major obstacle for quantitative detection of aberrations is the heterogeneity of clinical tumor tissue. Since tumor tissue invariably contains genetically normal stromal cells, this may lead to a failure to detect aberrations in the tumor cells.

Principal Finding

Using SNP array data from 44 non-small cell lung cancer samples we have developed a bioinformatic algorithm that accurately models the fractions of normal and tumor cells in clinical tumor samples. The proportion of normal cells in combination with SNP array data can be used to detect and quantify copy number neutral loss-of-heterozygosity (CNNLOH) in the tumor cells both in crude tumor tissue and in samples enriched for tumor cells by laser capture microdissection.

Conclusion

Genome-wide quantitative analysis of CNNLOH using the CNNLOH Quantifier method can help to identify recurrent aberrations contributing to tumor development in clinical tumor samples. In addition, SNP-array based analysis of CNNLOH may become important for detection of aberrations that can be used for diagnostic and prognostic purposes.     

Comparative DNA methylation analysis in normal and tumour tissues and in cancer cell lines using differential methylation hybridisation

Immortalized human cancer cell lines are widely used as tools and model systems in cancer research but their authenticity with regard to primary tissues remains a matter of debate. We have used differential methylation hybridisation to obtain comparative methylation profiles from normal and tumour tissues of lung and colon, and permanent cancer cell lines originally derived from these tissues. Average methylation differences only larger than 25% between sample groups were considered for the profiles and with this criterion approximately 1000 probesets, around 2% of the sites represented on the array, indicated differential methylation between normal lung and primary lung cancer tissue, and approximately 700 probesets between normal colon and primary colon cancer tissue. Both hyper- and hypomethylation was found to differentiate normal tissue from cancer tissue. The profiles obtained from these tissue comparisons were found to correspond largely to those from the corresponding cancer cell lines, indicating that the cell lines represent the methylation pattern of the primary tissue rather well. Moreover, the cancer specific profiles were found to be very similar for the two tumour types studied. Tissue specific differential methylation between lung and colon tissues, in contrast, was found to be preserved to a larger extent only in the malignant tissue, but was not preserved well in the cancer cell lines studied. Overall, our data therefore provide further evidence that permanent cell lines are good model systems for cancer specific methylation patterns, but deviate with regard to tissue-specific methylation.     

Discrimination of normal, inflammatory, premalignant, and malignant oral tissue: a Raman spectroscopy study

Optical spectroscopy methods are fast emerging as potential alternatives for early diagnosis of cancer. A Raman spectroscopy method for discrimination of normal and malignant oral tissues has been developed by us earlier. It is necessary to evaluate and establish the validity of the approach before it can be routinely used. In the present study, our Raman spectroscopy investigations are extended further to evaluate the efficacy of the technique to discriminate between normal, inflammatory, premalignant, and malignant conditions in oral tissue. Spectral profiles of normal, malignant, premalignant, and inflammatory conditions show pronounced differences between one another. Spectra of normal tissues can be attributed mainly to lipids whereas pathological tissue spectra are dominated by proteins. Principal components analysis (PCA) of the spectral data sets belonging to the four different categories showed that scores of factors differentiated between normal and all pathological conditions but gave only poor discrimination among the three pathological states. PCA combined with multiparameter limit tests allow match/mismatch criteria to be applied to test samples when pathologically certified calibration sets are available in each class. It is shown that by this method all the four tissue types could be discriminated and diagnosed correctly. The biochemical differences between normal and pathological conditions of oral tissue are also discussed from spectral differences of the different classes of spectra.     

Vitamin D receptor levels in colorectal cancer. Possible role of BsmI polymorphism

A high expression of vitamin D receptor (VDR) in colorectal cancer (CRC) tumoral tissue has been related to a good prognosis and it has been proposed that it could be a good biological marker of CRC progression. Nevertheless, there are no previous studies that compare the VDR expression in tumoral towards normal tissue of the same CRC patient in relation to VDR BsmI genotype. We collected normal and tumoral tissue samples, as well as blood samples, from CRC patients (n=170) and controls (n=122). VDR genotyping was performed and BsmI homozygous patients were selected (CRC=50, Cont=32). VDR mRNA and protein levels were analyzed. We also measured 25-Hydroxyvitamin D serum levels. We found no differences in the polymorphism distribution in tumoral versus normal tissue (control: BB=15.7%, bb=41.3%, Bb=43%; CRC: BB=14.2%, bb=41.9%, Bb=43.9%). Furthermore, VDR levels decreased in colonic cancer tissue (mean: 3.03) versus normal mucosa (11.62) from the same patient (p<0.001), but this decrease was similar in both genotypes. There were differences in 25-Hydroxyvitamin D(3) levels between the CRC and the control group (CRC=8.65 ng/ml, Cont=18.15 ng/ml). In conclusion, we found a decrease in VDR levels in tumoral compared with normal mucosa from the same patient. This difference is independent of the BsmI polymorphism.