[Genomics and progress in oncology]. / Place des analyses génomiques centralisées dans la découverte et la validation d'innovations diagnostiques en oncologie.

Genomics and genetics are becoming pillars of clinical research, allowing certain diseases to be understood at the molecular level and leading to improved treatments. Successful genetic analyses in the clinical oncological setting depend on the quality of the samples collected. At clinical sites, attention must be paid to the steps taken prior to analysis, and to the fragility of the molecules to be tested. The development of a standard procedure specifically adapted to the identification and validation of biomarkers through genomics requires an understanding of the different technologies used at each time point, from the analysis to the data processing, and the implication of personnel at all levels: the health care centers, the biotech companies, and/or the pharmaceutical industries.

Identification of new classes among acute myelogenous leukaemias with normal karyotype using gene expression profiling.

Conventional cytogenetic analysis currently stratifies acute myelogenous leukaemia (AML) into prognostically relevant groups. However, approximately 50% of adult AMLs have normal cytogenetics (NC-AMLs), and represent a heterogeneous and poorly understood group. We analysed gene expression in 55 AML samples including 53 cases from adult patients with NC-AML (n = 36), trisomy 8, t(15;17), t(8;21), t(11;19), 7q deletion, and two cell lines using 9000-gene DNA microarrays. Global hierarchical clustering showed that NC-AMLs are a heterogeneous group. Supervised analysis distinguished two subgroups of NC-AML: one subgroup constituted a homogeneous NC cluster ('pure NC-AML'), and the other NC-AMLs were close to the AML cases with translocations ('translocation like'). Gene expression signatures were also derived for patients with trisomy 8, as well as FLT3 and MLL gene duplications. Importantly, samples from 24 NC-AML patients who could be evaluated for clinical outcome were analysed. In all, 43 genes that discriminated two classes of patients with significantly different prognosis were identified. The poor prognosis class contained a majority of 'pure NC-AMLs', whereas the 'translocation-like' AMLs were in the good prognosis class. Discriminator genes included genes involved in drug resistance (TOP2B), protein transport (MTX2, SLC35A2), and cell signalling (MAPK1, PRKAB2). Our results demonstrate the transcriptional heterogeneity of NC-AMLs, and suggest the existence of 'translocation-like' NC-AMLs and of a gene expression signature that may predict response to chemotherapy.

Identification of TCL1A as an immunohistochemical marker of adverse outcome in diffuse large B-cell lymphomas.

We used a combination of DNA-microarray and tissue-microarray (TMA) analyses to identify markers that could be routinely used to predict the outcome of diffuse large-B-cell lymphoma (DLCL) patients. Gene expression profiling was performed using DNA-microarrays on 52 tumour biopsy samples [31 DLCL and 21 follicular lymphomas (FL)] from 48 patients (28 DLCL and 20 FL). T-cell leukemia/lymphoma-1A (TCL1A) mRNA overexpression was correlated with relapse in DLCL patients. TMA analysis was applied on a distinct series of 36 formalin-fixed, paraffin-embedded DLCL samples and showed that TCL1A immunoexpression was correlated with either higher relapse (p=0.02) or lower 5-year overall survival (p=0.009) rates. Moreover, the prognostic value of TCL1A was independent from IPI in our series. Our data suggest that TCL1A immunodetection is an independent marker of adverse outcome that could be used in routine settings for the management of DLCL patients.