Patient-derived xenograft models of non-small cell lung cancer for evaluating targeted drug sensitivity and resistance.

Patient-derived xenograft (PDX) models are a useful tool in cancer biology research. However, the number of lung cancer PDX is limited. In the present study, we successfully established 10 PDX, including three adenocarcinoma (AD), six squamous cell carcinoma (SQ) and one large cell carcinoma (LA), from 30 patients with non-small cell lung cancer (NSCLC) (18 AD, 10 SQ, and 2 LA), mainly in SCID hairless outbred (SHO) mice (Crlj:SHO-Prkdcscid Hrhr ). Histology of SQ, advanced clinical stage (III-IV), status of lymph node metastasis (N2-3), and maximum standardized uptake value ≥10 when evaluated using a delayed 18 F-fluoro-2-deoxy-d-glucose positron emission tomography (FDG-PET) scan was associated with successful PDX establishment. Histological analyses showed that PDX had histology similar to that of patients' surgically resected tumors (SRT), whereas components of the microenvironment were replaced with murine cells after several passages. Next-generation sequencing analyses showed that after two to six passages, PDX preserved the majority of the somatic mutations and mRNA expressions of the corresponding SRT. Two out of three PDX with AD histology had epidermal growth factor receptor (EGFR) mutations (L858R or exon 19 deletion) and were sensitive to EGFR tyrosine kinase inhibitors (EGFR-TKI), such as gefitinib and osimertinib. Furthermore, in one of the two PDX with an EGFR mutation, osimertinib resistance was induced that was associated with epithelial-to-mesenchymal transition. This study presented 10 serially transplantable PDX of NSCLC in SHO mice and showed the use of PDX with an EGFR mutation for analyses of EGFR-TKI resistance.

TBP-like Protein (TLP) Disrupts the p53-MDM2 Interaction and Induces Long-lasting p53 Activation.

Stress-induced activation of p53 is an essential cellular response to prevent aberrant cell proliferation and cancer development. The ubiquitin ligase MDM2 promotes p53 degradation and limits the duration of p53 activation. It remains unclear, however, how p53 persistently escapes MDM2-mediated negative control for making appropriate cell fate decisions. Here we report that TBP-like protein (TLP), a member of the TBP family, is a new regulatory factor for the p53-MDM2 interplay and thus for p53 activation. We found that TLP acts to stabilize p53 protein to ensure long-lasting p53 activation, leading to potentiation of p53-induced apoptosis and senescence after genotoxic stress. Mechanistically, TLP interferes with MDM2 binding and ubiquitination of p53. Moreover, single cell imaging analysis shows that TLP depletion accelerates MDM2-mediated nuclear export of p53. We further show that a cervical cancer-derived TLP mutant has less p53 binding ability and lacks a proliferation-repressive function. Our findings uncover a role of TLP as a competitive MDM2 blocker, proposing a novel mechanism by which p53 escapes the p53-MDM2 negative feedback loop to modulate cell fate decisions.

Construction of possible integrated predictive index based on EGFR and ANXA3 polymorphisms for chemotherapy response in fluoropyrimidine-treated Japanese gastric cancer patients using a bioinformatic method.

BACKGROUND: Variability in drug response between individual patients is a serious concern in medicine. To identify single-nucleotide polymorphisms (SNPs) related to drug response variability, many genome-wide association studies have been conducted. METHODS: We previously applied a knowledge-based bioinformatic approach to a pharmacogenomics study in which 119 fluoropyrimidine-treated gastric cancer patients were genotyped at 109,365 SNPs using the Illumina Human-1 BeadChip. We identified the SNP rs2293347 in the human epidermal growth factor receptor (EGFR) gene as a novel genetic factor related to chemotherapeutic response. In the present study, we reanalyzed these hypothesis-free genomic data using extended knowledge. RESULTS: We identified rs2867461 in annexin A3 (ANXA3) gene as another candidate. Using logistic regression, we confirmed that the performance of the rs2867461 + rs2293347 model was superior to those of the single factor models. Furthermore, we propose a novel integrated predictive index (iEA) based on these two polymorphisms in EGFR and ANXA3. The p value for iEA was 1.47 × 10(-8) by Fisher's exact test. Recent studies showed that the mutations in EGFR is associated with high expression of dihydropyrimidine dehydrogenase, which is an inactivating and rate-limiting enzyme for fluoropyrimidine, and suggested that the combination of chemotherapy with fluoropyrimidine and EGFR-targeting agents is effective against EGFR-overexpressing gastric tumors, while ANXA3 overexpression confers resistance to tyrosine kinase inhibitors targeting the EGFR pathway. CONCLUSIONS: These results suggest that the iEA index or a combination of polymorphisms in EGFR and ANXA3 may serve as predictive factors of drug response, and therefore could be useful for optimal selection of chemotherapy regimens.

Correlated expression levels of endothelin receptor B and Plexin C1 in melanoma.

Discussion concerning the effect of endothelin receptor B (Ednrb) on melanoma continues because Ednrb has been reported to have both tumor promoting and suppressive effects for melanoma. In order to examine Ednrb-related signaling in melanomagenesis, DNA microarray analysis for a melanoma from a RFP/RET-transgenic mouse (RET-mouse) and a melanoma from an Ednrb-heterozygously deleted RET-mouse [Ednrb(+/-);RET-mouse], in both of which melanoma spontaneously develops, was performed in this study. We found that the expression level of Plexin C1 (PlxnC1), a suppressor for melanoma, in a melanoma from an Ednrb(+/-);RET-mouse was drastically decreased compared to that in a melanoma from a RET-mouse. Therefore, we further examined the correlation between Ednrb and PlxnC1 expression levels in melanomas. PlxnC1 transcript expression levels in melanomas from Ednrb(+/-);RET-mice were lower than those in melanomas from RET-mice. A strong correlation between Ednrb and PlxnC1 transcript expression levels (R = 0.78, p < 0.01) was also found in melanomas from both RET-mice and Ednrb(+/-);RET-mice. Correspondingly, there was a significant correlation between transcript (R = 0.80; p < 0.01) and protein (R = 0.60; p < 0.01) expression levels of EDNRB and PLXNC1 in human primary melanomas. Together with our results showing that the expression level of PLXNC1 transcript was reduced in EDNRB-depleted human melanoma cells, our results showing positively correlated expression levels of Ednrb/EDNRB and PlxnC1/PLXNC1 in melanoma suggest that PlxnC1/PLXNC1 is involved in the Ednrb/EDNRB-mediated suppressive effect on melanoma.

Identification of novel Arabidopsis thaliana upstream open reading frames that control expression of the main coding sequences in a peptide sequence-dependent manner.

Upstream open reading frames (uORFs) are often found in the 5'-leader regions of eukaryotic mRNAs and can negatively modulate the translational efficiency of the downstream main ORF. Although the effects of most uORFs are thought to be independent of their encoded peptide sequences, certain uORFs control translation of the main ORF in a peptide sequence-dependent manner. For genome-wide identification of such peptide sequence-dependent regulatory uORFs, exhaustive searches for uORFs with conserved amino acid sequences have been conducted using bioinformatic analyses. However, whether the conserved uORFs identified by these bioinformatic approaches encode regulatory peptides has not been experimentally determined. Here we analyzed 16 recently identified Arabidopsis thaliana conserved uORFs for the effects of their amino acid sequences on the expression of the main ORF using a transient expression assay. We identified five novel uORFs that repress main ORF expression in a peptide sequence-dependent manner. Mutational analysis revealed that, in four of them, the C-terminal region of the uORF-encoded peptide is critical for the repression of main ORF expression. Intriguingly, we also identified one exceptional sequence-dependent regulatory uORF, in which the stop codon position is not conserved and the C-terminal region is not important for the repression of main ORF expression.

Analysis of gene expression profiles of soft tissue sarcoma using a combination of knowledge-based filtering with integration of multiple statistics.

The diagnosis and treatment of soft tissue sarcomas (STS) have been difficult. Of the diverse histological subtypes, undifferentiated pleomorphic sarcoma (UPS) is particularly difficult to diagnose accurately, and its classification per se is still controversial. Recent advances in genomic technologies provide an excellent way to address such problems. However, it is often difficult, if not impossible, to identify definitive disease-associated genes using genome-wide analysis alone, primarily because of multiple testing problems. In the present study, we analyzed microarray data from 88 STS patients using a combination method that used knowledge-based filtering and a simulation based on the integration of multiple statistics to reduce multiple testing problems. We identified 25 genes, including hypoxia-related genes (e.g., MIF, SCD1, P4HA1, ENO1, and STAT1) and cell cycle- and DNA repair-related genes (e.g., TACC3, PRDX1, PRKDC, and H2AFY). These genes showed significant differential expression among histological subtypes, including UPS, and showed associations with overall survival. STAT1 showed a strong association with overall survival in UPS patients (logrank p = 1.84 × 10(-6) and adjusted p value 2.99 × 10(-3) after the permutation test). According to the literature, the 25 genes selected are useful not only as markers of differential diagnosis but also as prognostic/predictive markers and/or therapeutic targets for STS. Our combination method can identify genes that are potential prognostic/predictive factors and/or therapeutic targets in STS and possibly in other cancers. These disease-associated genes deserve further preclinical and clinical validation.

Application of a combination of a knowledge-based algorithm and 2-stage screening to hypothesis-free genomic data on irinotecan-treated patients for identification of a candidate single nucleotide polymorphism related to an adverse effect.

Interindividual variation in a drug response among patients is known to cause serious problems in medicine. Genomic information has been proposed as the basis for "personalized" health care. The genome-wide association study (GWAS) is a powerful technique for examining single nucleotide polymorphisms (SNPs) and their relationship with drug response variation; however, when using only GWAS, it often happens that no useful SNPs are identified due to multiple testing problems. Therefore, in a previous study, we proposed a combined method consisting of a knowledge-based algorithm, 2 stages of screening, and a permutation test for identifying SNPs. In the present study, we applied this method to a pharmacogenomics study where 109,365 SNPs were genotyped using Illumina Human-1 BeadChip in 168 cancer patients treated with irinotecan chemotherapy. We identified the SNP rs9351963 in potassium voltage-gated channel subfamily KQT member 5 (KCNQ5) as a candidate factor related to incidence of irinotecan-induced diarrhea. The p value for rs9351963 was 3.31×10-5 in Fisher's exact test and 0.0289 in the permutation test (when multiple testing problems were corrected). Additionally, rs9351963 was clearly superior to the clinical parameters and the model involving rs9351963 showed sensitivity of 77.8% and specificity of 57.6% in the evaluation by means of logistic regression. Recent studies showed that KCNQ4 and KCNQ5 genes encode members of the M channel expressed in gastrointestinal smooth muscle and suggested that these genes are associated with irritable bowel syndrome and similar peristalsis diseases. These results suggest that rs9351963 in KCNQ5 is a possible predictive factor of incidence of diarrhea in cancer patients treated with irinotecan chemotherapy and for selecting chemotherapy regimens, such as irinotecan alone or a combination of irinotecan with a KCNQ5 opener. Nonetheless, clinical importance of rs9351963 should be further elucidated.

Macrophage migration inhibitory factor and stearoyl-CoA desaturase 1: potential prognostic markers for soft tissue sarcomas based on bioinformatics analyses.

The diagnosis and treatment of soft tissue sarcomas (STSs) has been particularly difficult, because STSs are a group of highly heterogeneous tumors in terms of histopathology, histological grade, and primary site. Recent advances in genome technologies have provided an excellent opportunity to determine the complete biological characteristics of neoplastic tissues, resulting in improved diagnosis, treatment selection, and investigation of therapeutic targets. We had previously developed a novel bioinformatics method for marker gene selection and applied this method to gene expression data from STS patients. This previous analysis revealed that the extracted gene combination of macrophage migration inhibitory factor (MIF) and stearoyl-CoA desaturase 1 (SCD1) is an effective diagnostic marker to discriminate between subtypes of STSs with highly different outcomes. In the present study, we hypothesize that the combination of MIF and SCD1 is also a prognostic marker for the overall outcome of STSs. To prove this hypothesis, we first analyzed microarray data from 88 STS patients and their outcomes. Our results show that the survival rates for MIF- and SCD1-positive groups were lower than those for negative groups, and the p values of the log-rank test are 0.0146 and 0.00606, respectively. In addition, survival rates are more significantly different (p = 0.000116) between groups that are double-positive and double-negative for MIF and SCD1. Furthermore, in vitro cell growth inhibition experiments by MIF and SCD1 inhibitors support the hypothesis. These results suggest that the gene set is useful as a prognostic marker associated with tumor progression.

Identification of a candidate single-nucleotide polymorphism related to chemotherapeutic response through a combination of knowledge-based algorithm and hypothesis-free genomic data.

Inter-individual variations in drug responses among patients are known to cause serious problems in medicine. Genome-wide association study (GWAS) is powerful for examining single-nucleotide polymorphisms (SNPs) and their relationships with drug response variations. However, no significant SNP has been identified using GWAS due to multiple testing problems. Therefore, we propose a combination method consisting of knowledge-based algorithm, two stages of screening, and permutation test for identifying SNPs in the present study. We applied this method to a genome-wide pharmacogenomics study for which 109,365 SNPs had been genotyped using Illumina Human-1 BeadChip for 119 gastric cancer patients treated with fluoropyrimidine. We identified rs2293347 in epidermal growth factor receptor (EGFR) is as a candidate SNP related to chemotherapeutic response. The p value for the rs2293347 was 2.19 × 10(-5) for Fisher's exact test, and the p value was 0.00360 for the permutation test (multiple testing problems are corrected). Additionally, rs2293347 was clearly superior to clinical parameters and showed a sensitivity value of 55.0% and specificity value of 94.4% in the evaluation by using multiple regression models. Recent studies have shown that combination chemotherapy of fluoropyrimidine and EGFR-targeting agents is effective for gastric cancer patients highly expressing EGFR. These results suggest that rs2293347 is a potential predictive factor for selecting chemotherapies, such as fluoropyrimidine alone or combination chemotherapies.

Reduced GNG2 expression levels in mouse malignant melanomas and human melanoma cell lines.

Heterotrimeric G protein is composed of a Gα-subunit and a Gßγ-dimer. Previous studies have revealed that Gßγ-dimers including the Gγ2 subunit (Gng2/GNG2) are associated with cell proliferation, differentiation, invasion and angiogenesis. At present, however, there is no information on the expression level of Gng2/GNG2 alone in any kind of tumor. In this study, we performed DNA microarray analysis in a benign melanocytic tumor and a malignant melanoma from RET-transgenic mice (RET-mice). Gng2 transcript expression levels in a malignant melanoma were less than 1/10 of the level in a benign tumor. The difference in Gng2 transcript expression levels between benign tumors and malignant melanomas was greatest among all of the G protein γ subunits examined in this study. Moreover, protein expression levels of Gng2 were decreased in malignant melanomas compared with those in benign melanocytic tumors in RET-mice. Analysis of human malignant melanomas also showed reduced GNG2 protein expression levels in five human malignant melanoma cell lines compared with the expression levels in normal human epithelial melanocytes (NHEM). Thus, we demonstrated for the first time that Gng2/GNG2 expression levels are reduced in malignant melanoma, suggesting that GNG2 could be a novel biomarker for malignant melanoma.

Forkhead box A1 transcriptional pathway in KRT7-expressing esophageal squamous cell carcinomas with extensive lymph node metastasis.

Prognosis of cancers with lymph node metastasis is known to be very poor; however, it is still controversial whether metastatic potential can be evaluated by expression profiles of primary tumors. Therefore, to address this issue, we compared gene expression profiles of 24 esophageal squamous cell carcinomas (ESCCs) with extensive lymph node metastasis and 11 ESCCs with no metastatic lymph node. However, there was no gene cluster distinguishing these two groups, suggesting that lymph node metastasis-associated genes are varied depending on cases or subgroups. Therefore, we applied a recently developed filtering method (S2N') to identify such genes, and successfully extracted 209 genes associated with node status. Among them, over-expression of CALB1, KRT7/CK7, MUC1 and CEA/CEACAM5 in poor prognostic cases with metastatic lymph nodes was confirmed in two sets of ESCCs by RT-PCR. Each often seemed to have glandular cell type-characteristics in both the gene expression and morphology. It was also revealed that FOXA1 siRNA treatment of esophageal cancer cells reduced the mRNA level of both KRT7 and a stabilizer of epithelial-mesenchymal transition (EMT) regulator LOXL2, and that both FOXA1 and LOXL2 siRNAs reduced invasion and migration of ESCC cells. In 15 KRT7-expressing ESCCs with metastatic lymph nodes, 60% expressed FOXA1 and 33% expressed both FOXA1 and LOXL2. These results suggest that FOXA1 induces not only KRT7 but also LOXL2 in a subset of poor prognostic ESCCs with metastatic lymph nodes, and it is also plausible, that other FOXA1 downstream genes could be therapeutic targets of poor prognostic ESCCs.

Proposal of new gene filtering method, BagPART, for gene expression analysis with small sample.

A significant problem in gene expression analysis is that the sample size is substantially lower than the number of genes. Bagging is an effective method of solving this problem in the case of small sample datasets. We have devised a combination method, called the BagPART filtering method, that uses the projective adaptive resonance theory (PART) to select important genes and achieve a binary classification more accurately (p<10(-10)) than conventional methods, particularly when the sample size is small.

Gene expression analysis of soft tissue sarcomas: characterization and reclassification of malignant fibrous histiocytoma.

In soft tissue sarcomas, the diagnosis of malignant fibrous histiocytoma (MFH) has been a very controversial issue, and MFH is now considered to be reclassified into pleomorphic subtypes of other sarcomas. To characterize MFH genetically, we used an oligonucleotide microarray to analyze gene expression in 105 samples from 10 types of soft tissue tumors. Spindle cell and pleomorphic sarcomas, such as dedifferentiated liposarcoma, myxofibrosarcoma, leiomyosarcoma, malignant peripheral nerve sheath tumor (MPNST), fibrosarcoma and MFH, showed similar gene expression patterns compared to other tumors. Samples from those five sarcoma types could be classified into respective clusters based on gene expression by excluding MFH samples. We calculated distances between MFH samples and other five sarcoma types (dedifferentiated liposarcoma, myxofibrosarcoma, leiomyosarcoma, MPNST and fibrosarcoma) based on differentially expressed genes and evaluated similarities. Three of the 21 MFH samples showed marked similarities to one of the five sarcoma types, which were supported by histological findings. Although most of the remaining 18 MFH samples showed little or no histological resemblance to one of the five sarcoma types, 12 of them showed moderate similarities in terms of gene expression. These results explain the heterogeneity of MFH and show that the majority of MFHs could be reclassified into pleomorphic subtypes of other sarcomas. Taken together, gene expression profiling could be a useful tool to unveil the difference in the underlying molecular backgrounds, which leads to a rational taxonomy and diagnosis of a diverse group of soft tissue sarcomas.

Classification of intramural metastases and lymph node metastases of esophageal cancer from gene expression based on boosting and projective adaptive resonance theory.

Esophageal cancer is a well-known cancer with poorer prognosis than other cancers. An optimal and individualized treatment protocol based on accurate diagnosis is urgently needed to improve the treatment of cancer patients. For this purpose, it is important to develop a sophisticated algorithm that can manage a large amount of data, such as gene expression data from DNA microarrays, for optimal and individualized diagnosis. Marker gene selection is essential in the analysis of gene expression data. We have already developed a combination method of the use of the projective adaptive resonance theory and that of a boosted fuzzy classifier with the SWEEP operator denoted PART-BFCS. This method is superior to other methods, and has four features, namely fast calculation, accurate prediction, reliable prediction, and rule extraction. In this study, we applied this method to analyze microarray data obtained from esophageal cancer patients. A combination method of PART-BFCS and the U-test was also investigated. It was necessary to use a specific type of BFCS, namely, BFCS-1,2, because the esophageal cancer data were very complexity. PART-BFCS and PART-BFCS with the U-test models showed higher performances than two conventional methods, namely, k-nearest neighbor (kNN) and weighted voting (WV). The genes including CDK6 could be found by our methods and excellent IF-THEN rules could be extracted. The genes selected in this study have a high potential as new diagnosis markers for esophageal cancer. These results indicate that the new methods can be used in marker gene selection for the diagnosis of cancer patients.

Cancer diagnosis marker extraction for soft tissue sarcomas based on gene expression profiling data by using projective adaptive resonance theory (PART) filtering method.

BACKGROUND: Recent advances in genome technologies have provided an excellent opportunity to determine the complete biological characteristics of neoplastic tissues, resulting in improved diagnosis and selection of treatment. To accomplish this objective, it is important to establish a sophisticated algorithm that can deal with large quantities of data such as gene expression profiles obtained by DNA microarray analysis. RESULTS: Previously, we developed the projective adaptive resonance theory (PART) filtering method as a gene filtering method. This is one of the clustering methods that can select specific genes for each subtype. In this study, we applied the PART filtering method to analyze microarray data that were obtained from soft tissue sarcoma (STS) patients for the extraction of subtype-specific genes. The performance of the filtering method was evaluated by comparison with other widely used methods, such as signal-to-noise, significance analysis of microarrays, and nearest shrunken centroids. In addition, various combinations of filtering and modeling methods were used to extract essential subtype-specific genes. The combination of the PART filtering method and boosting--the PART-BFCS method--showed the highest accuracy. Seven genes among the 15 genes that are frequently selected by this method--MIF, CYFIP2, HSPCB, TIMP3, LDHA, ABR, and RGS3--are known prognostic marker genes for other tumors. These genes are candidate marker genes for the diagnosis of STS. Correlation analysis was performed to extract marker genes that were not selected by PART-BFCS. Sixteen genes among those extracted are also known prognostic marker genes for other tumors, and they could be candidate marker genes for the diagnosis of STS. CONCLUSION: The procedure that consisted of two steps, such as the PART-BFCS and the correlation analysis, was proposed. The results suggest that novel diagnostic and therapeutic targets for STS can be extracted by a procedure that includes the PART filtering method.

Modified signal-to-noise: a new simple and practical gene filtering approach based on the concept of projective adaptive resonance theory (PART) filtering method.

UNLABELLED: Considering the recent advances in and the benefits of DNA microarray technologies, many gene filtering approaches have been employed for the diagnosis and prognosis of diseases. In our previous study, we developed a new filtering method, namely, the projective adaptive resonance theory (PART) filtering method. This method was effective in subclass discrimination. In the PART algorithm, the genes with a low variance in gene expression in either class, not both classes, were selected as important genes for modeling. Based on this concept, we developed novel simple filtering methods such as modified signal-to-noise (S2N') in the present study. The discrimination model constructed using these methods showed higher accuracy with higher reproducibility as compared with many conventional filtering methods, including the t-test, S2N, NSC and SAM. The reproducibility of prediction was evaluated based on the correlation between the sets of U-test p-values on randomly divided datasets. With respect to leukemia, lymphoma and breast cancer, the correlation was high; a difference of >0.13 was obtained by the constructed model by using <50 genes selected by S2N'. Improvement was higher in the smaller genes and such higher correlation was observed when t-test, NSC and SAM were used. These results suggest that these modified methods, such as S2N', have high potential to function as new methods for marker gene selection in cancer diagnosis using DNA microarray data. AVAILABILITY: Software is available upon request.

Prediction of peptide binding to major histocompatibility complex class II molecules through use of boosted fuzzy classifier with SWEEP operator method.

To treat autoimmune diseases, it is important to identify which peptides bind to major histocompatibility complex (MHC) class II molecules (HLA-DRs). Predicting the peptides that bind to MHC class II molecules can effectively reduce the number of experiments required for identifying helper T cell epitopes. In our previous study, we applied fuzzy neural networks (FNNs) to solve this problem. However, an FNN requires a long calculation time and a large number of peptides; this means performing several experiments. In this study, we applied a boosted fuzzy classifier with the SWEEP operator method (BFCS) to solve this problem. For comparison, two other conventional modeling methods, namely, support vector machine and FNN combined with the SWEEP operator method (FNN-SWEEP) instead of using solely an FNN, were employed. Compared with FNN, FNN-SWEEP is extremely fast and has an almost identical prediction accuracy. The model constructed by BFCS showed an accuracy approximately 5%-10% higher than that constructed by FNN-SWEEP. In addition, BFCS was 30,000-120,000 times faster than FNN-SWEEP. This result suggests that BFCS has the potential to function as a new method of predicting peptides that bind to various protein receptors.

Construction of robust prognostic predictors by using projective adaptive resonance theory as a gene filtering method.

MOTIVATION: For establishing prognostic predictors of various diseases using DNA microarray analysis technology, it is desired to find selectively significant genes for constructing the prognostic model and it is also necessary to eliminate non-specific genes or genes with error before constructing the model. RESULTS: We applied projective adaptive resonance theory (PART) to gene screening for DNA microarray data. Genes selected by PART were subjected to our FNN-SWEEP modeling method for the construction of a cancer class prediction model. The model performance was evaluated through comparison with a conventional screening signal-to-noise (S2N) method or nearest shrunken centroids (NSC) method. The FNN-SWEEP predictor with PART screening could discriminate classes of acute leukemia in blinded data with 97.1% accuracy and classes of lung cancer with 90.0% accuracy, while the predictor with S2N was only 85.3 and 70.0% or the predictor with NSC was 88.2 and 90.0%, respectively. The results have proven that PART was superior for gene screening. AVAILABILITY: The software is available upon request from the authors. CONTACT: honda@nubio.nagoya-u.ac.jp

Prognostic predictor with multiple fuzzy neural models using expression profiles from DNA microarray for metastases of breast cancer.

Gene expression profiling data from DNA microarray were analyzed using the fuzzy neural network (FNN) modeling method for predicting the distant metastases of breast cancer. The best model consisting of five genes was able to predict metastases of breast cancer with 94% accuracy. Furthermore, 100% accuracy was achieved by majoritarian decision using only 25 genes from five noninferior models which were constructed independently. From the constructed model, gene expression rules, which may cause distant metastases, were explicitly extracted and 60% of the metastases cases could be explained by this rule. The FNN modeling method described in this paper enables precise extraction of significant biological markers affecting prognosis without prior knowledge.