Similar expression pattern was also observed in Arabidopsis, chickpea , foxtail millet and maize

is of importance in evaluating the utility of targeted therapeutic drugs. To explore how well this panel of 32 PDGCX models represented human GC, we performed further genetic characterization studies using IHC and FISH assays, in order to provide an in situ analysis of the target gene’s copy number and protein expression. The individual PDGCX model’s histological and genetic profiles were compared with the parental human GC tumors and agreement rates determined according to biomarker positivity and by Cohen’s Kappa. Our data revealed either high or perfect agreement in the majority of the biomarkers tested by either method, especially at the DNA level. Although ERBB1 and ERBB3 protein expression were judged to have `slight agreement’ according to kappa value, these biomarkers still showed an agreement rate of 59% and 75% between parental tumors and PDGCX models. A closer look at the data 8 / 13 PDGCX Characterization revealed that parental tumors and PDGCX models agreed well for both protein biomarkers, when the parental tumor was positive for the biomarker. This link was less obvious when the primary tumor was negative for the biomarker. Thus, the disagreement judged by Cohen’s Kappa could be a consequence of statistical bias due to the relatively small number of primary tumor negative samples. Furthermore, intratumoral heterogeneity of different biomarkers has been frequently reported in surgical GC samples, and may represent a more intrinsical reason for the inconsistencies between primary tumors and models. Nevertheless, the profiles of all tested biomarkers across the whole panel of 32 PDGCX models accurately reflects that of their prevalence in human GC samples, which is reported as 30~75% for ERBB1 positivity, 15~22% for ERBB2 positivity, 60~70% for ERBB3 high expression, 39~47% for PTEN loss, 45% for FGFR2 amplification, 20~70% for MET high expression and 0~10% for MET amplification. This high concordance could in part be attributable to the lack of correlation between most of the patient clinicopathological parameters and model success rates. Importantly, the positivity of ERBB2, FGFR2 and MET, either at the DNA or protein level, were stably maintained in all PDGCX models, out to at least passage F12. This considerable fidelity and duration of genetic maintenance underscores another advantage of using PDGCX models for evaluation of pre-clinical drug efficacy. The purpose of generating this panel of PDGCX models was to evaluate the anti-tumor efficacy of targeted therapeutic drugs. Indeed, a number of PDGCX models from this panel with different genetic aberrations have been previously used to successfully explore preclinical drug efficacy. For example, models with positive ERBB2, FGFR2 amplification, PTEN loss and MET amplification were sensitive to Trastuzumab, AZD4547, buy 1702259-66-2 AZD5363 and Volitinib, respectively. Taken together, these data highlight the considerable PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/19756382 utility of PDGCX models in evaluating preclinical drug efficacy, enabling definition of prospective biomarker selection criteria and modeling of tumor architecture and genetic heterogeneity. Conclusions In summary, we have successfully established a panel of 32 PDGCX models and demonstrated that these models faithfully recapitulate the histological characteristics and genetic diversity of the primary human tumors. Furthermore, these PDGCX xenograft models PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/19755711 maintain genetic diversity until at least the 12th model passage. Our data also show that the panel in its entirety ac