FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2726355231> ?p ?o ?g. }

• W2726355231 endingPage "851" @default.
• W2726355231 startingPage "847" @default.
• W2726355231 abstract "The use of cell lines has been, and continues to be, an invaluable tool for understanding the cell and molecular biology of malignant haematopoiesis. Good culture practice should involve regularly confirming authenticity to identify possible cross-contamination from other cells (Capes-Davis et al, 2013). However, a further issue exists: how closely does a cell line represent the parental disease process? A gene expression profile represents an integration of all of the molecular and mutational processes driven by the disease and has been used to characterise, classify and predict response and to provide prognostic information. For example, does an acute myeloid leukaemia (AML) cell line actually represent the molecular processes associated with the phenotype? Using the power of gene expression profiling of a disease reference data we have robustly allocated a panel of commonly used leukaemia cell lines to disease sub-groups. GECA, Gene Expression Compositional Assignment (Blayney et al, 2016), is an in silico method using compositional statistics to transfer prior knowledge from a reference data set to an unclassified query set. The approach enables the comparison of similarities in gene expression profiles across independent datasets, platforms and technologies, thus removing cross platform normalisation. GECA had been applied to a library of epithelial ovarian cell lines with respect to a reference set of solid tumours (Blayney et al, 2016). In this study, the GCEA pipeline used 10 000 random gene-sets (of 500 in length) to compare the similarity of three cell line query datasets to the disease reference data set. These similarity assignments were tested for robustness, by random permutation of ranks to produce an estimated p-value, adjusted for multiple comparisons. A false discovery rate (FDR) of 0·05 was then applied. The disease reference data set used was the Microarray Innovations in LEukaemia (MILE) data set (Haferlach et al, 2010) (GEO: GSE13204): 2096 samples from 17 leukaemia sub-groups. Query dataset 1 (Q1) comprised 39 haematological cell lines as part of an Astra Zeneca project of 627 cell lines (GEO: GSE57083), profiled on the Affymetrix (Affy) HG133Plus2 platform (Affymetrix, Santa Clara, CA, USA). Query dataset 2 (Q2) was 20 profiles from 11 different cell lines, also on Affy HG133Plus2 platform; whilst Query data 3 (Q3) included six profiles representing five cell lines on the Affy HG133A platform. For Q1 and Q2, GCEA analysis was undertaken at the probe-set level (n = 54 630) with NA values removed; whilst for Q3, intersecting platform probe-set levels were used (n = 22 232) also with NA values removed. Composite aggregate assignments for each of the 18 MILE subgroups were then calculated, converted to percentages (Table SI) and displayed as heat maps for the query data sets (Fig 1). Several cell lines were represented in two or three of the query data sets, which enabled the development of a consensus similarity across platforms; summarised in Table 1. HL60, a commonly used cell line, showed the largest transcriptional similarity with AML with normal karyotype + other abnormalities (AML-NKO) reference group (average 46·5% similarity over Q1-Q3). Similarities were found with AML with 11q23/KMT2A (~15%); AML with t(8;21) (~11·9%) and AML with t(15;17) (8·1%) groups. HL60 was derived from a patient with AML French-American-British (FAB) type M2 (Collins, 1987), is responsive to all trans retinoic acid differentiation therapy, similar to acute promyelocytic leukaemia (APL) and is used as a model for that disease. K562 cells were established from a chronic myeloid leukaemia (CML) patient in blast crisis with a Philadelphia chromosome (Lozzio & Lozzio, 1975). Across Q1–3, K562 demonstrated high levels of similarity with AML-NKO (29-64%), with variable similarity (6·4–47·5%) with the MDS group. Surprising, and across all three query sets, no, or very little, similarity with the CML or acute lymphocytic leukaemia (ALL) with t(9;22) group was seen. NB4 (Duprez et al, 1992) and Kasumi-1 (Asou et al, 1991), represent AML with t(15;17) and AML with t(8;21) respectively. For NB4 cells, the dominant similarity was with the AML-NKO group (Q1-Q3 average of 36·7%) followed by AML with t(15;17) (22·4%). Kasumi-1 had the greatest similarity with AML with t(8;21) (37·4% in Q1, 60·6% in Q2). U937 was most similar to AML-NKO (47·6%) with 24·4% similarity to AML with t(11q23)/KMT2A; U937 cells have the t(10;11)(p14;q23) translocation (Sundstrom & Nilsson, 1976). SKM1 and UT7 cell lines were shown to have similarity with AML-NKO, but in addition, GECA predicted assignments to MDS: SKM-1 with an average of 15% and UT-7 at 22·3%. SKM-1 cells were established from an AML patient in transformation following MDS (Nakagawa & Matozaki, 1995) whilst the UT7 was from an AML FAB type M7 patient (Komatsu et al, 1991) but with no description of a preceding MDS phase as suggested by the GCEA results. Interestingly REH cells, derived from an ALL at first relapse with the t(12;21) translocation, had GECA similarity of 35·6% (Q2) and 48·4% (Q3) to ALL with t(12;21) samples, yet only 5·5% using the Q1 REH profile. Surprisingly the Q1 REH showed 26·8% similarity to ALL with t(1;19). Overall, GCEA identified both similarities and conflicts in the query cell line expression profiles compared to the patient derived reference dataset. The majority of cell lines mapped to a core disease of AML-NKO, probably reflecting the stage of maturation arrest occurring during myeloid differentiation. Specific and relevant expression contributions to sub-diseases were seen for t(11q23/KMT2A) U937, t(15;17) for NB4, t(8;21) for Kasumi-1 and MDS for SKM1. Perhaps, the most surprising result was the lack of similarity of K562, across the query data sets, for CML or ALL with t(9;22) profiles despite the confirmed presence of the Philadelphia chromosome. Some dissimilarity could be explained by different sources of cell lines or variations in laboratory protocol, whilst others could be explained by the reduced number of probesets in Q3, which used the smaller 133A array. However, this study has demonstrated that GECA can provide insight into the value, relevance and validity of using cell lines as models of sub-types of acute leukaemia for molecular and cellular research. The results presented here should be considered when interpreting experimental data. JB and KIM designed the study, performed the analysis and wrote the paper. The authors have no competing interests. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article." @default.
• W2726355231 created "2017-07-14" @default.
• W2726355231 creator A5033442365 @default.
• W2726355231 creator A5075911816 @default.
• W2726355231 date "2017-06-27" @default.
• W2726355231 modified "2023-09-27" @default.
• W2726355231 title "Compositional analysis gives insight into leukaemia cell lines expression profiles compared to those within patient sub-groups" @default.
• W2726355231 cites W1540357669 @default.
• W2726355231 cites W1773256867 @default.
• W2726355231 cites W1994041743 @default.
• W2726355231 cites W2103771397 @default.
• W2726355231 cites W2133764756 @default.
• W2726355231 cites W2467820808 @default.
• W2726355231 cites W4239139991 @default.
• W2726355231 doi "https://doi.org/10.1111/bjh.14711" @default.
• W2726355231 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/28653436" @default.
• W2726355231 hasPublicationYear "2017" @default.
• W2726355231 type Work @default.
• W2726355231 sameAs 2726355231 @default.
• W2726355231 citedByCount "0" @default.
• W2726355231 crossrefType "journal-article" @default.
• W2726355231 hasAuthorship W2726355231A5033442365 @default.
• W2726355231 hasAuthorship W2726355231A5075911816 @default.
• W2726355231 hasBestOaLocation W27263552311 @default.
• W2726355231 hasConcept C199360897 @default.
• W2726355231 hasConcept C41008148 @default.
• W2726355231 hasConcept C70721500 @default.
• W2726355231 hasConcept C71924100 @default.
• W2726355231 hasConcept C86803240 @default.
• W2726355231 hasConcept C90559484 @default.
• W2726355231 hasConceptScore W2726355231C199360897 @default.
• W2726355231 hasConceptScore W2726355231C41008148 @default.
• W2726355231 hasConceptScore W2726355231C70721500 @default.
• W2726355231 hasConceptScore W2726355231C71924100 @default.
• W2726355231 hasConceptScore W2726355231C86803240 @default.
• W2726355231 hasConceptScore W2726355231C90559484 @default.
• W2726355231 hasIssue "6" @default.
• W2726355231 hasLocation W27263552311 @default.
• W2726355231 hasLocation W27263552312 @default.
• W2726355231 hasLocation W27263552313 @default.
• W2726355231 hasOpenAccess W2726355231 @default.
• W2726355231 hasPrimaryLocation W27263552311 @default.
• W2726355231 hasRelatedWork W1506200166 @default.
• W2726355231 hasRelatedWork W1995515455 @default.
• W2726355231 hasRelatedWork W2048182022 @default.
• W2726355231 hasRelatedWork W2080531066 @default.
• W2726355231 hasRelatedWork W2469483056 @default.
• W2726355231 hasRelatedWork W2748952813 @default.
• W2726355231 hasRelatedWork W2899084033 @default.
• W2726355231 hasRelatedWork W3031052312 @default.
• W2726355231 hasRelatedWork W3032375762 @default.
• W2726355231 hasRelatedWork W3108674512 @default.
• W2726355231 hasVolume "181" @default.
• W2726355231 isParatext "false" @default.
• W2726355231 isRetracted "false" @default.
• W2726355231 magId "2726355231" @default.
• W2726355231 workType "article" @default.