Search

A key goal of computational personalized medicine is to systematically utilize genomic and other molecular features of samples to predict drug responses for a previously unseen sample. Such predictions are valuable for developing hypotheses for selecting therapies tailored for individual patients. This is especially valuable in oncology, where molecular and genetic heterogeneity of the cells has a major impact on the response. However, the prediction task is extremely challenging, raising the need for methods that can effectively model and predict drug responses. In this study, we propose a novel formulation of multi-task matrix factorization that allows selective data integration for predicting drug responses. To solve the modeling task, we extend the state-of-the-art kernelized Bayesian matrix factorization (KBMF) method with component-wise multiple kernel learning. In addition, our approach exploits the known pathway information in a novel and biologically meaningful fashion to learn the drug response associations. Our method quantitatively outperforms the state of the art on predicting drug responses in two publicly available cancer data sets as well as on a synthetic data set. In addition, we validated our model predictions with lab experiments using an in-house cancer cell line panel. We finally show the practical applicability of the proposed method by utilizing prior knowledge to infer pathway-drug response associations, opening up the opportunity for elucidating drug action mechanisms. We demonstrate that pathway-response associations can be learned by the proposed model for the well known EGFR and MEK inhibitors., Comment: Accepted in European Conference in Computational Biology, to be published in Bioinformatics 2016

Many commercial and custom-made microarray formats are routinely used for large-scale gene expression surveys. Here, we sought to determine the level of concordance between microarray platforms by analyzing breast cancer cell lines with in situ synthesized oligonucleotide arrays (Affymetrix HG-U95v2), commercial cDNA microarrays (Agilent Human 1 cDNA), and custom-made cDNA microarrays from a sequence-validated 13K cDNA library. Gene expression data from the commercial platforms showed good correlations across the experiments (r = 0.78–0.86), whereas the correlations between the custom-made and either of the two commercial platforms were lower (r = 0.62–0.76). Discrepant findings were due to clone errors on the custom-made microarrays, old annotations, or unknown causes. Even within platform, there can be several ways to analyze data that may influence the correlation between platforms. Our results indicate that combining data from different microarray platforms is not straightforward. Variability of the data represents a challenge for developing future diagnostic applications of microarrays.

The most effective targeted cancer therapies have arisen from research into genetically altered oncogenes, including BCR-ABL, HER2, RAS and EGFR. Recent advances in cancer genetics have identified many regions of the genome that undergo amplification (increase in copy number) but, in most cases, the key oncogenic targets driving the growth and survival of cancer cells remain unknown. In this review, we discuss high-throughput technologies for the discovery of putative oncogenes, and clinical and functional validation of these genes as targets for therapy. New technologies in translational genomics facilitate the identification, validation and prioritization of candidate molecular targets for anti-cancer therapy.

Background: Human serum paraoxonase eliminates carcinogenic lipid-soluble radicals. Because expression of the main human paraoxonase gene PON1 varies widely in humans, certain PON1 polymorphisms might be associated with increased risks of cancer. We sought new functional mutations in PON1 and determined whether known or new PON1 mutations were associated with the risk for prostate cancer in a prospective, random, population-based sample of Finnish men and in a case–control study. Methods: Serum paraoxonase activity was measured in 835 healthy men in the Kuopio Ischaemic Heart Disease Risk Factor Study. PON1 mutations were identified by hierarchical phenotype-targeted sequencing in DNAs from the 100 men with the lowest paraoxonase activity in this cohort, and 1595 men in the cohort were genotyped for PON1 mutations by restriction fragment length polymorphism. Multivariable analysis was used to investigate the association of known and new PON1 mutations with incident prostate cancer in 1569 cancer-free men in the cohort followed for 9–14 years. In a case–control study of Finnish men, the association of prostate cancer with the PON1 mutation identified in the cohort study was investigated in 69 case patients with familial prostate cancer and 69 unmatched healthy control subjects. Results: We identified a new single-nucleotide PON1 polymorphism associated with decreased serum paraoxonase activity that caused an isoleucine→valine change at codon 102 in exon 4 (I102V). Of the 1569 men cancer-free at baseline, 56 (3.6%) were carriers of the I102V mutation. After adjusting for age and cholesterol-lowering medications, the relative risk for developing prostate cancer during follow-up was 6.3 (95% confidence interval [CI] = 2.1 to 19.2) among 102V allele carriers compared with noncarriers. Other PON1 alleles were not statistically significantly associated with prostate cancer. In the case–control study, patients with familial prostate cancer were more likely to be carriers of the PON1 I102V mutation than control subjects (odds ratio = 4.3, 95% CI = 0.9 to 21.5). Conclusion: The PON1 102V allele appears to be associated with an increased risk for prostate cancer. [J Natl Cancer Inst 2003;95:812–8]

cDNA microarrays permit massively parallel gene expression analysis and have spawned a new paradigm in the study of molecular biology. One of the significant challenges in this genomic revolution is to develop sophisticated approaches to facilitate the visualization, analysis, and interpretation of the vast amounts of multi-dimensional gene expression data. We have applied self-organizing map (SOM) in order to meet these challenges. In essence, we utilize U-matrix and component planes in microarray data visualization and introduce general procedure for assessing significance for a cluster detected from U-matrix. Our case studies consist of two data sets. First, we have analyzed a data set containing 13,824 genes in 14 breast cancer cell lines. In the second case we show an example of the SOM in drug treatment of prostate cancer cells. Our results indicate that (1) SOM is capable of helping finding certain biologically meaningful clusters, (2) clustering algorithms could be used for finding a set of potential predictor genes for classification purposes, and (3) comparison and visualization of the effects of different drugs is straightforward with the SOM. In summary, the SOM provides an excellent format for visualization and analysis of gene microarray data, and is likely to facilitate extraction of biologically and medically useful information.

In breast cancer, several chromosomal sites frequently undergo amplification, implicating the location of genes important for tumor development and progression. Here we cloned two novel genes, breast carcinoma amplified sequence 3 (BCAS3) and 4 (BCAS4), from the two most common amplification sites in breast cancer, 17q23 and 20q13. The BCAS3 gene at 17q23 spans more than 600 kb at the genomic level and was predicted to encode a 913 amino acid nuclear protein. The BCAS4 gene at 20q13.2 encodes a 211 amino acid cytoplasmic protein. Both BCAS3 and BCAS4 represent novel genes with no homologies to any other known gene or protein. In the MCF7 breast cancer cell line, the BCAS3 and BCAS4 genes were co-amplified, and cloning of a highly overexpressed 1.3-kb transcript revealed a rearrangement fusing the last two exons of BCAS3 with BCAS4. The fusion led to a novel message in which only the first exon of BCAS4 and part of exon 23 of BCAS3 were transcribed. The BCAS4-BCAS3 fusion transcript was detected only in MCF7 cells, but the BCAS4 gene was also overexpressed in nine of 13 breast cancer cell lines. In conclusion, our results indicate that these novel genes, BCAS3 at 17q23 and BCAS4 at 20q13.2, undergo amplification, overexpression, and fusion in breast cancer and therefore may have a role in the frequent chromosomal alterations affecting these two loci.

CHEK2 (previously known as “CHK2”) is a cell-cycle–checkpoint kinase that phosphorylates p53 and BRCA1 in response to DNA damage. A protein-truncating mutation, 1100delC in exon 10, which abolishes the kinase function of CHEK2, has been found in families with Li-Fraumeni syndrome (LFS) and in those with a cancer phenotype that is suggestive of LFS, including breast cancer. In the present study, we found that the frequency of 1100delC was 2.0% among an unselected population-based cohort of 1,035 patients with breast cancer. This was slightly, but not significantly (P=.182), higher than the 1.4% frequency found among 1,885 population control subjects. However, a significantly elevated frequency was found among those 358 patients with a positive family history (11/358 [3.1%]; odds ratio [OR] 2.27; 95% confidence interval [CI] 1.11–4.63; P=.021, compared with population controls). Furthermore, patients with bilateral breast cancer were sixfold more likely to be 1100delC carriers than were patients with unilateral cancer (95% CI 1.87–20.32; P=.007). Analysis of the 1100delC variant in an independent set of 507 patients with familial breast cancer with no BRCA1 and BRCA2 mutations confirmed a significantly elevated frequency of 1100delC (28/507 [5.5%]; OR 4.2; 95% CI 2.4–7.2; P=.0002), compared with controls, with a high frequency also seen in patients with only a single affected first-degree relative (18/291 [6.2%]). Finally, tissue microarray analysis indicated that breast tumors from patients with 1100delC mutations show reduced CHEK2 immunostaining. The results suggest that CHEK2 acts as a low-penetrance tumor-suppressor gene in breast cancer and that it makes a significant contribution to familial clustering of breast cancer—including families with only two affected relatives, which are more common than families that include larger numbers of affected women.

A vast number of recurrent chromosomal alterations have been implicated in cancer development and progression. However, most of the genes involved in recurrent chromosomal alterations in solid tumors remain unknown, despite the recent substantial progress in genomic research and availability of high-throughput technologies. For example, it is now possible to quickly identify large numbers of differentially expressed genes in cancer specimens using cDNA microarrays. Integration of this ‘functional genomic view’ of the cancer genome with the ‘cytogenetic view’ could lead to the identification of genes playing a critical role in cancer development and progression. In this review, we illustrate how the combination of three different microarray technologies, cDNA, CGH, and tissue microarrays, makes it possible to directly identify genes involved in chromosomal rearrangements in cell line model systems and then rapidly explore their significance as potential diagnostic and therapeutic targets in human primary breast cancer progression.

Tissue microarray (TMA) technology allows rapid visualization of molecular targets in thousands of tissue specimens at a time, either at the DNA, RNA or protein level. The technique facilitates rapid translation of molecular discoveries to clinical applications. By revealing the cellular localization, prevalence and clinical significance of candidate genes, TMAs are ideally suitable for genomics-based diagnostic and drug target discovery. TMAs have a number of advantages compared with conventional techniques. The speed of molecular analyses is increased by more than 100-fold, precious tissues are not destroyed and a very large number of molecular targets can be analyzed from consecutive TMA sections. The ability to study archival tissue specimens is an important advantage as such specimens are usually not applicable in other high-throughput genomic and proteomic surveys. Construction and analysis of TMAs can be automated, increasing the throughput even further. Most of the applications of the TMA technology have come from the field of cancer research. Examples include analysis of the frequency of molecular alterations in large tumor materials, exploration of tumor progression, identification of predictive or prognostic factors and validation of newly discovered genes as diagnostic and therapeutic targets.

A rapidly increasing number of genes are being suspected to play a role in cancer biology. To evaluate the clinical significance of newly detected potential cancer genes, it is usually required to examine a high number of well-characterized primary tumors. Using traditional methods of molecular pathology, this is a time consuming endeavor rapidly exhausting precious tissue resources. To allow for a high throughput tissue analysis we have developed a “tissue chip” approach (Kononen et al., Nat. Med. 1998;4:844–7). Using this tissue microarray (TMA) technology, samples from up to 1,000 different tumors are arrayed in one recipient paraffin block, sections of which can be used for all kind of in situ analyses. Section from TMA blocks can then be utilized for the simultaneous analysis of up to 1,000 different tumors on the DNA, RNA or protein level. TMAs allow a high throughput molecular analysis of thousands of tumors within a few hours. All currently available data have suggested that minute arrayed tissue specimens are highly representative of their donor tissues. There are multiple different types of TMAs that can be utilized in cancer research including multi tumor arrays (containing different tumor types), tumor progression arrays (tumors of different stages) and prognostic arrays (tumors with clinical endpoints). The combination of multiple different TMAs allows a very quick but comprehensive characterization of biomarkers of interest. We anticipate that the use of TMAs will greatly accelerate the transition of basic research findings to clinical applications. © 2001 Wiley-Liss, Inc.

Development of high-throughput 'biochip' technologies has dramatically enhanced our ability to study biology and explore the molecular basis of disease. Biochips enable massively parallel molecular analyses to be carried out in a miniaturized format with a very high throughput. This review will highlight applications of the various biochip technologies in cancer research, including analysis of 1) disease predisposition by using single-nucleotide polymorphism (SNP) microarrays, 2) global gene expression patterns by cDNA microarrays, 3) concentrations, functional activities or interactions of proteins with proteomic biochips, and 4) cell types or tissues as well as clinical endpoints associated with molecular targets by using tissue microarrays. One can predict that individual cancer risks can, in the future, be estimated accurately by a microarray profile of multiple SNPs in critical genes. Diagnostics of cancer will be facilitated by biochip readout of activity levels of thousands of genes and proteins. Biochip diagnostics coupled with informatics solutions will form the basis of individualized treatment decisions for cancer patients.

Objective: Five to ten percent of prostate cancers may be caused by inherited genetic defects. In order to explore the nature of inherited cancer risks in the genetically homogeneous Finnish population, we investigated the incidence of prostate cancer and other cancers in first-degree relatives of prostate cancer patients by linking the population-based parish records on relatives with the Finnish Cancer Registry (FCR) data. Methods: The study population was composed of first-degree relatives of two groups of prostate cancer patients diagnosed in Finland during 1988–1993: (1) all early-onset (≤60years) patients (n=557) from the entire country, (2) a sample (n=989) of prostate cancer patients diagnosed at an age of >60years. A total of 11,427 first-degree relatives were identified through parish records, and their cancer incidence was determined based on a total of 299,970 person-years. Standardized incidence ratios (SIR) were calculated based on expected cancer rates in the general population. Results: The SIR of prostate cancer was increased in both Cohort 1 (2.5, 95% CI 1.9–3.2) and Cohort 2 (1.7, 95% CI 1.4–2.1). The risk of prostate cancer was high for relatives of patients diagnosed at an early age, and then leveled off for patients in the median age of prostate cancer diagnosis (70–79 years). However, the prostate cancer risk for relatives of patients diagnosed ≥80years was again statistically significantly elevated (SIR 1.8, 95% CI 1.3–2.6), suggesting a contribution of genetic factors to prostate cancer also at a late age of onset. Gastric cancer was the only other cancer type with a significantly elevated risk among the relatives. Increased risk of gastric cancer was seen only in male relatives of prostate cancer patients diagnosed at an early age, with the highest risk detected for the male relatives of prostate cancer patients diagnosed at an age of 55 years or less (SIR 5.0, 95% CI 2.8–8.2). Conclusions: Our population-based study indicates that hereditary factors may play an important role in the development of prostate cancer among the relatives of men diagnosed both at younger and older ages. This finding is relevant in the context of our observations that HPCX (hereditary prostate cancer susceptibility locus on Xq27-28) linkage in Finland is found exclusively among families with late age of onset. The association of gastric cancer with prostate cancer has not been reported previously, and may reflect the effects of a novel predisposition locus, which increases the risk to both of these common tumor types.

Background Germline mutations in recessive cancer predisposition genes are uncovered by somatic genetic deletions during tumor development. Analysis of genetic changes in tumor tissues from patients with an inherited predisposition may therefore highlight regions of the genome containing susceptibility or modifier genes. Our aim was to characterize genetic changes in familial prostate cancer Methods Twenty-one primary prostate cancers from 19 Finnish prostate cancer families were analyzed for somatic genetic changes by comparative genomic hybridization (CGH). Results The average number of genetic alterations per tumor was 4.0 ± 1.9, distributed equally among losses and gains. The most common losses were found at chromosomal regions 13q14–q22 (29%), 8p12-pter (24%), and 6q13–q16 (14%), and the most common gains at 19p (25%), 19q (14%) and 7q (14%). Conclusions These results suggest that prostate cancers in genetically predisposed individuals arise for the most part through similar somatic genetic progression pathways as sporadic prostate cancers. This also implies that the biological properties of tumors from the two groups may not be different from one another. Prostate 46:233–239, 2001. © 2001 Wiley-Liss, Inc.

The genetic changes underlying the development and progression of male breast cancer are poorly understood. Germline BRCA2 mutations account for a significant part of male breast cancer, but the majority of patients lack a known inherited predisposition. We recently demonstrated that the progression of breast cancer in female carriers of a germline BRCA1 or BRCA2 mutation follows specific genetic pathways, distinct from each other and from sporadic breast cancer. In the present study, we performed a genome-wide survey by comparative genomic hybridization (CGH) of somatic genetic aberrations in 26 male breast cancers, including five tumors from BRCA2 mutation carriers. BRCA2 tumors exhibited a significantly higher number of chromosomal aberrations than sporadic tumors. The most common alterations in sporadic male breast cancer were +1q (38%), +8q (33%), +17q (33%), -13q (29%), and -8p (24%). In tumors from BRCA2 mutation carriers, the five most common genetic changes were +8q (100%), +20q (100%), +17q (80%), -13q (80%), and -6q (60%). The CGH results in these two groups of male breast cancers are almost identical to those identified in the corresponding sporadic and BRCA2-associated female breast cancers. The results suggest that despite substantial hormonal differences between females and males, similar genetic changes are selected for during tumor progression. Furthermore, the presence of a highly penetrant germline BRCA2 mutation apparently leads to a characteristic somatic tumor progression pathway, again shared between affected male and female mutation carriers.

Comparative genomic hybridization (CGH) reveals DNA sequence copy number changes that are shared among the different cell subpopulations present in a tumor and may help to delineate the average progression pathways of breast cancer. Previous CGH studies of breast cancer have concentrated on selected subgroups of breast cancer. Here, 55 unselected primary breast carcinomas were analyzed using optimized quality-controlled CGH procedures. Gains of 1q (67%) and 8q (49%) were the most frequent aberrations. Other recurrent gains were found at 33 chromosomal regions, with 16p, 5p12-14, 19q, 11q13-14, 17q12, 17q22-24, 19p, and 20q13 being most often (> 18%) involved. Losses found in > 18% of the tumors involved 8p, 16q, 13q, 17p, 9p, Xq, 6q, 11q, and 18q. The total number of aberrations per tumor was highest in poorly differentiated (P = 0.01) and in DNA aneuploid (P = 0.05) tumors. The high frequency of 1q gains and presence of +1q as the sole abnormality suggest that it is an early genetic event. In contrast, gains of 8q were most common in genetically and phenotypically advanced breast cancers. The vast majority of breast cancers (80%) have gains of 1q, 8q, or both, and 3 changes (+1q, +8q, or -13q) account for 91% of the tumors. In conclusion, CGH results indicate that certain chromosomal imbalances are very often selected for, sometimes in a preferential order, during the progression of breast cancer. Further studies of such common changes may form the basis for a molecular cytogenetic classification of breast cancer.

Comparative genomic hybridization (CGH) provides a molecular cytogenetic approach for genome-wide scanning of differences in DNA sequence copy number. The technique is now attracting wide-spread interest, especially among cancer researchers. The rapidly expanding database of CGH publications already covers about 1500 tumors and is beginning to reveal genetic abnormalities that are characteristic of certain tumor types or stages of tumor progression. Six novel gene amplifications, as well as a locus for a cancer-predisposition syndrome, have been discovered based on CGH data. CGH has now been established as a first-line screening technique for cancer researchers and will serve as a basis for ongoing efforts to develop high-resolution next-generation genome scanning, such as the microarray technology.

We describe an algorithm for fully automated flow cytometric DNA histogram classification and analysis that provides rapid, reproducible determination of DNA index and S-phase fraction (SPF). Automated classification agreed with subjective assessment of DNA ploidy in 96–98% of DNA histograms. Automated and conventional analyses of DNA index (r = 0.95) and SPF (r = 0.89) were also highly correlated with one another. In a series of 86 node-negative breast carcinomas, SPF calculated with the fully automated method was a significant predictor of 10 year survival (P = 0.009). Automation greatly increased the speed of DNA histogram analysis, allowing evaluation of the same set of histograms with different methods. In a preliminary study exploring the optimization of DNA histogram analysis, the best association between SPF and prognosis of breast cancer patients was achieved using sliced nuclei debris modeling, reporting only the aneuploid SPF (in aneuploid histograms), while excluding small aneuploid clones (

Fluorescence in situ hybridization (FISH) and specific DNA probes for peri-centromeric repeat regions and unique sequence loci have made it possible to study chromosomal aberrations from interphase tumor nuclei. Large-scale retrospective studies on the prognostic value of interphase cytogenetics would become feasible if these techniques were readily applicable to nuclei from archival formalin-fixed tumor tissues. We describe here an improved technique for interphase FISH analysis of tumors that have been extensively fixed in formalin. The protocol aims at improving probe penetration and hybridization efficiency by inducing chromatin decondensation and swelling of the nuclei with a heat treatment in a 90 degrees C glycerol solution prior to hybridization. Using this cell pretreatment, FISH results on the detection of chromosome copy number aberrations and amplification of the c-erbB-2 oncogene from formalin-fixed, paraffin-embedded tissues were highly concordant with those from fresh tissues. In contrast to previously described methods, separate adjustments of denaturation or proteinase K digestion are not required for each sample. This method facilitates retrospective analyses of large series of tumors and is also useful for applying FISH to routine diagnostic purposes using formalin-fixed material.

Androgens play a critical role in the growth of both androgen dependent and castration-resistant prostate cancer (CRPC). Only a few micro-RNAs (miRNAs) have been suggested to be androgen regulated. We aim to identify androgen regulated miRNAs.We utilized LNCaP derived model, we have established, and which overexpresses the androgen receptor (AR), the VCaP cell line, and 13 intact-castrated prostate cancer (PC) xenograft pairs, as well as clinical specimens of untreated (PC) and CRPC. The expression of miRNAs was analyzed by microarrays and quantitative RT-PCR (Q-RT-PCR). Transfection of pre-miR-141 and anti-miR-141 was also used.Seventeen miRNAs were1.5-fold up- or downregulated upon dihydrotestosterone (DHT) treatment in the cell lines, and 42 after castration in the AR-positive xenografts. Only four miRNAs (miR-10a, miR-141, miR-150*, and miR-1225-5p) showed similar androgen regulation in both cell lines and xenografts. Of those, miR-141 was found to be expressed more in PC and CRPC compared to benign prostate hyperplasia. Additionally, the overexpression of miR-141 enhanced growth of parental LNCaP cells while inhibition of miR-141 by anti-miR-141 suppressed the growth of the LNCaP subline overexpressing AR.Only a few miRNAs were found to be androgen-regulated in both cell lines and xenografts models. Of those, the expression of miR-141 was upregulated in cancer. The ectopic overexpression of miR-141 increased growth of LNCaP cell suggesting it may contribute to the progression of PC.

BACKGROUND Mutations in the p53 gene resulting in the accumulation of altered p53 proteins with prolonged half-life have been found in a large variety of human malignancies. PURPOSE We studied the significance of p53 protein accumulation in prostatic carcinoma. METHODS The material consisted of 137 paraffin-embedded, primary prostatic carcinomas. Accumulation of p53 protein was studied by immunohistochemical staining using a polyclonal p53-specific CM-1 antibody. Proliferation activity was determined by DNA flow cytometry and by immunohistochemical detection of proliferative cell nuclear antigen (PCNA) using a monoclonal PC10 antibody. RESULTS Eight (6%) of the tumors showed intense p53 staining in more than 20% of the tumor cells, 15 (11%) had only lower level immunoreactivity, and 114 (83%) showed no staining. High-level p53 accumulation was associated with high histologic grade (P less than .001), DNA aneuploidy (P less than .05), and high cell proliferation rate as defined by flow cytometric S-phase analysis (P less than .01) or PCNA expression (P less than .01). High-level p53 accumulation predicted short, progression-free interval (P less than .01) and poor survival (P less than .001), with about a 12-fold relative risk of death as compared with p53-negative cases. Low-level p53 accumulation had no prognostic significance. CONCLUSIONS Accumulation of p53 confers proliferative advantage for prostatic carcinoma cells and defines a small subgroup of highly malignant carcinomas.

c-erbB-2 protein over-expression was studied immunohistochemically in 319 paraffin-embedded breast carcinomas representing 89% of all breast-cancer cases operated in the Tampere University Hospital between 1977 and 1981. The immunohistochemical evaluation of c-erbB-2 was optimized using protease pre-treatment and verified using antibodies for both the external and the internal domains of the protein. c-erbB-2 over-expression was found in 72 (23%) of the 319 cases and was associated with high histological and nuclear grade (p less than 0.0001), DNA aneuploidy (p = 0.003), high tumor S-phase fraction (p less than 0.0001), and lack of estrogen (p less than 0.0001) and progesterone (p = 0.03) receptors. Overall, breast-cancer patients with c-erbB-2 over-expression had about 2.2-fold relative risk (RR) of death (p less than 0.001) as compared with those without over-expression. According to a multivariate analysis, c-erbB-2 over-expression was an independent prognostic factor in the whole material as well as in the node-negative sub-set. In node-negative breast-cancer tumor size, S-phase and c-erbB-2 status defined a large patient group with only 4% 5-year and 15% 10-year mortality rate without adjuvant therapy. In comparison with c-erbB-2-negative tumors, those with over-expression of this gene metastasized 3 times more often (p = 0.0002) to the lungs, liver and brain and 3 times less often to the bone. Our findings suggest that the prognostic value of c-erbB-2 over-expression may be related not only to increased cell proliferation rate but also to a distinctive pattern of metastasis.

Purpose: To comprehensively evaluate ephrin receptor B2 (EphB2) expression in normal and neoplastic tissues. EphB2 is a tyrosine kinase recently implicated in the deregulation of cell-to-cell communication in many tumors. Experimental Design: EphB2 protein expression was analyzed by immunohistochemistry on tissue microarrays that included 76 different normal tissues, >4,000 samples from 138 different cancer types, and 1,476 samples of colon cancer with clinical follow-up data. Results: We found most prominent EphB2 expression in the intestinal epithelium (colonic crypts) with cancer of the colorectum displaying the highest EphB2 positivity of all tumors. Positivity was found in 100% of 118 colon adenomas but in 33.3% of 45 colon carcinomas. EphB2 expression was also observed in 75 tumor categories, including serous carcinoma of the endometrium (34.8%), adenocarcinoma of the esophagus (33.3%), intestinal adenocarcinoma of the stomach (30.2%), and adenocarcinoma of the small intestine (70%). The occasional finding of strong EphB2 positivity in tumors without EphB2 positivity in the corresponding normal cells [adenocarcinoma of the lung (4%) and pancreas (2.2%)] suggests that deregulation of EphB2 signaling may involve up-regulation of the protein expression. In colon carcinoma, loss of EphB2 expression was associated with advanced stage (P < 0.0001) and was an indicator of poor overall survival (P = 0.0098). Conclusions: Our results provide an overview on the EphB2 protein expression in normal and neoplastic tissues. Deregulated EphB2 expression may play a role in several cancer types with loss of EphB2 expression serving as an indicator of the possible pathogenetic role of EphB2 signaling in the maintenance of tissue architecture of colon epithelium.

The etiology and pathogenesis of male breast cancer (MBC) are poorly known. This is due to the fact that the disease is rare, and large-scale genetic epidemiologic studies have been difficult to carry out. Here, we studied the frequency of eight recurrent Finnish BRCA2 founder mutations in a large cohort of 154 MBC patients (65% diagnosed in Finland from 1967 to 1996). Founder mutations were detected in 10 patients (6.5%), eight of whom carried the 9346(-2) A>G mutation. Two novel mutations (4075 delGT and 5808 del5) were discovered in a screening of the entire BRCA2 coding region in 34 samples. However, these mutations were not found in the rest of the 120 patients studied. Patients with positive family history of breast and/or ovarian cancer were often BRCA2 mutation carriers (44%), whereas those with no family history showed a low frequency of involvement (3.6%; P < .0001). Finally, we found only one Finnish MBC patient with 999 dell, the most common founder mutation in Finnish female breast cancer (FBC) patients, and one that explains most of the hereditary FBC and MBC cases in Iceland. The variation in BRCA2 mutation spectrum between Finnish MBC patients and FBC patients in Finland and breast cancer patients in Iceland suggests that modifying genetic and environmental factors may significantly influence the penetrance of MBC and FBC in individuals carrying germline BRCA2 mutations in some populations.

Genetic risk factors for male breast cancer (MBC) are poorly understood. High penetrance genes such as BRCA1 or BRCA2 account for only a small proportion of the disease. A 1100delC mutation in CHEK2 (previously known as CHK2), a cell-cycle checkpoint kinase, has been implicated in predisposition of Li-Fraumeni syndrome (LFS) and breast cancer in families suggestive of LFS. This 1100delC mutation has also been shown to confer a 2-fold increase of breast cancer risk in women and a 10-fold increase of risk in men. It was estimated to account for 1% of breast cancers in women and as much as 9% of breast cancers in men at the population level based on analysis of breast cancer families without BRCA1 or BRCA2 mutations. We wanted to evaluate the significance of CHEK2 1100delC in predisposition to MBC by assessing its frequency in a population-based material of 114 Finnish MBC patients. Two patients (1.8%) carried the 1100delC mutation. The mutation frequency among MBC cases was similar to that seen in population controls (26/1885, 1.4%). Our results indicate that CHEK2 1100delC variant does not substantially increase the risk of male breast cancer at the population level. We cannot exclude the fact that a small fraction of hereditary, family-positive male breast cancers could be attributable to CHEK2 mutations.

Inhibitor of apoptosis (IAP) family proteins are suppressors of apoptosis that have been implicated in apoptosis resistance in some cancers. Their expression and relevance to the prognosis of prostate cancer were investigated.The expression of four members of the IAP family (cellular inhibitor of apoptosis protein 1, cellular inhibitor of apoptosis protein 2, X chromosome-linked IAP, and survivin) was examined by immunohistochemistry and immunoblotting in human prostate cancers and in prostate tissues from transgenic mice expressing SV40 large T antigen under control of a probasin promoter.Tumor-associated elevations in the levels of all four IAP family members were common in prostate cancers of both humans and mice, suggesting concomitant up-regulation of multiple IAP family proteins. Compared with normal prostatic epithelium, increased IAP expression was often evident even in prostatic intraepithelial neoplasia lesions (carcinoma in situ), suggesting that deregulation of IAP expression occurs early in the pathogenesis of prostate cancer. IAP expression did not correlate with Gleason grade or prostate-specific antigen levels.The findings demonstrate that tumor- associated elevations in the expression of several IAP family proteins occur as a frequent and early event in the etiology of prostate cancer.

Only a proportion of breast cancer families has germline mutations in the BRCA1 or BRCA2 genes, suggesting the presence of additional susceptibility genes. Finding such genes by linkage analysis has turned out to be difficult due to the genetic heterogeneity of the disease, phenocopies and incomplete penetrance of the mutations. Isolated populations may be helpful in reducing the level of genetic heterogeneity and in providing useful starting points for further genetic analyses. Here, we report results from a genome-wide linkage analysis of 14 high-risk breast cancer families from Finland. These families tested negative for BRCA1 and BRCA2 germline mutations and showed no linkage to the 13q21 region, recently proposed as an additional susceptibility locus. Suggestive linkage was seen at marker D2S364 (2q32) with a parametric two-point LOD score of 1.61 (theta=0), and an LOD score of 2.49 in nonparametric analyses. Additional genotyping of a 40 cM chromosomal region surrounding the region of interest yielded a maximum parametric two-point LOD score of 1.80 (theta=0) at D2S2262 and a nonparametric LOD score of 3.11 at an adjacent novel marker 11291M1 in BAC RP11-67G7. A nonparametric multipoint LOD score of 3.20 was seen at 11291M1 under the assumption of dominant inheritance. While not providing proof of linkage considering the small number of families and large number of laboratory and statistical analyses performed, these results warrant further studies of the 2q32 chromosomal region as a candidate breast cancer susceptibility locus. Both linkage and association studies are likely to be useful, particularly in other isolated populations.

Thymidylate synthase (TS) is the target enzyme for 5-fluorouracil (5-FU), and TS expression may determine clinical response and survival after therapy with 5-FU in colorectal cancer. 5-FU is also widely used in the adjuvant therapy of pancreatic cancer. Therefore, we explored the hypothesis that TS expression was associated with patient prognosis and the response to adjuvant therapy in pancreatic cancer.Cylindrical tissue cores from a large retrospective, nonrandomized series covering 132 resected patients were used to build a pancreatic cancer tissue microarray. TS expression was determined using immunohistochemistry.High intratumoral TS expression and low intratumoral TS expression were present in 83 of 132 (63%) and 49 of 132 (37%) tumors, respectively. Median survival among patients with low intratumoral TS expression (18 months) was longer than that among patients with high TS expression (12 months). In multivariate analysis, more advanced pathological stage [risk ratio (RR) = 1.70; P = 0.015], poorly differentiated histology (RR = 1.71; P = 0.015), management with adjuvant therapy (RR = 0.49; P = 0.011), and high TS expression [RR = 1.66; 95% confidence interval (CI) = 1.05-2.63; P = 0.029] were independent predictors of mortality. The risk of death was significantly reduced by any adjuvant therapy (RR = 0.40; 95% CI = 0.18-0.90; P = 0.001) among patients with high TS expression. This difference in survival among patients with low- and high-TS-expressing tumors became more significant when the analysis was restricted to the 73 patients receiving 5-FU-based adjuvant therapy (RR = 0.37; 95% CI = 0.16-0.86; P = 0.0006). In contrast, 5-FU-based adjuvant therapy did not influence survival among patients with low-TS-expressing pancreatic cancer.High TS expression is a marker of poor prognosis in resected pancreatic cancer. Patients with high intratumoral TS expression benefit from adjuvant therapy.

Mutations in the androgen receptor (AR) gene have been suggested to predispose to male breast cancer (MBC). Studies on MBC patients have not been based on the mutation screening of the entire coding region of the AR and the number of subjects has been small. Therefore, some AR gene alterations may have remained undetected. In the present study, we have comprehensively screened the entire coding region of the AR gene for mutations and also studied the role of AR CAG and GGC repeat lengths as risk factors for MBC in a cohort of 32 Finnish MBC patients. To estimate the possible involvement of the prostate cancer predisposing AR Arg726Leu germ-line mutation in MBC, this mutation was tested in 117 MBC patients. No germ-line mutations were found and the CAG and GGC repeat lengths were similar among MBC cases as among Scandinavian population. Our data indicate that the AR gene does not substantially contribute to MBC predisposition.

Genetic changes underlie tumor progression and may lead to cancer-specific expression of critical genes. Over 1100 publications have described the use of comparative genomic hybridization (CGH) to analyze the pattern of copy number alterations in cancer, but very few of the genes affected are known. Here, we performed high-resolution CGH analysis on cDNA microarrays in breast cancer and directly compared copy number and mRNA expression levels of 13,824 genes to quantitate the impact of genomic changes on gene expression. We identified and mapped the boundaries of 24 independent amplicons, ranging in size from 0.2 to 12 Mb. Throughout the genome, both high- and low-level copy number changes had a substantial impact on gene expression, with 44% of the highly amplified genes showing overexpression and 10.5% of the highly overexpressed genes being amplified. Statistical analysis with random permutation tests identified 270 genes whose expression levels across 14 samples were systematically attributable to gene amplification. These included most previously described amplified genes in breast cancer and many novel targets for genomic alterations, including the HOXB7 gene, the presence of which in a novel amplicon at 17q21.3 was validated in 10.2% of primary breast cancers and associated with poor patient prognosis. In conclusion, CGH on cDNA microarrays revealed hundreds of novel genes whose overexpression is attributable to gene amplification. These genes may provide insights to the clonal evolution and progression of breast cancer and highlight promising therapeutic targets.

In breast cancer, several chromosomal sites frequently undergo amplification, implicating the location of genes important for tumor development and progression. Here we cloned two novel genes, breast carcinoma amplified sequence 3 (BCAS3) and 4 (BCAS4), from the two most common amplification sites in breast cancer, 17q23 and 20q13. The BCAS3 gene at 17q23 spans more than 600 kb at the genomic level and was predicted to encode a 913 amino acid nuclear protein. The BCAS4 gene at 20q13.2 encodes a 211 amino acid cytoplasmic protein. Both BCAS3 and BCAS4 represent novel genes with no homologies to any other known gene or protein. In the MCF7 breast cancer cell line, the BCAS3 and BCAS4 genes were co-amplified, and cloning of a highly overexpressed 1.3-kb transcript revealed a rearrangement fusing the last two exons of BCAS3 with BCAS4. The fusion led to a novel message in which only the first exon of BCAS4 and part of exon 23 of BCAS3 were transcribed. The BCAS4-BCAS3 fusion transcript was detected only in MCF7 cells, but the BCAS4 gene was also overexpressed in nine of 13 breast cancer cell lines. In conclusion, our results indicate that these novel genes, BCAS3 at 17q23 and BCAS4 at 20q13.2, undergo amplification, overexpression, and fusion in breast cancer and therefore may have a role in the frequent chromosomal alterations affecting these two loci.