Molecular Comparison of Imatinib-Naïve and Resistant Gastrointestinal Stromal Tumors: Differentially Expressed microRNAs and mRNAs

Abstract

Despite the success of imatinib in advanced gastrointestinal stromal tumor (GIST) patients, 50% of the patients experience resistance within two years of treatment underscoring the need to get better insight into the mechanisms conferring imatinib resistance. Here the microRNA and mRNA expression profiles in primary (imatinib-naïve) and imatinib-resistant GIST were examined. Fifty-three GIST samples harboring primary KIT mutations (exon 9; n = 11/exon 11; n = 41/exon 17; n = 1) and comprising imatinib-naïve (IM-n) (n = 33) and imatinib-resistant (IM-r) (n = 20) tumors, were analyzed. The microRNA expression profiles were determined and from a subset (IM-n, n = 14; IM-r, n = 15) the mRNA expression profile was established. Ingenuity pathway analyses were used to unravel biochemical pathways and gene networks in IM-r GIST. Thirty-five differentially expressed miRNAs between IM-n and IM-r GIST samples were identified. Additionally, miRNAs distinguished IM-r samples with and without secondary KIT mutations. Furthermore 352 aberrantly expressed genes were found in IM-r samples. Pathway and network analyses revealed an association of differentially expressed genes with cell cycle progression and cellular proliferation, thereby implicating genes and pathways involved in imatinib resistance in GIST. Differentially expressed miRNAs and mRNAs between IM-n and IM-r GIST were identified. Bioinformatic analyses provided insight into the genes and biochemical pathways involved in imatinib-resistance and highlighted key genes that may be putative treatment targets.

Keywords: GIST; IPA pathway analysis; drug resistance; gene networks; imatinib; mRNA expression; miRNA.

Figure 1

MicroRNA expression distinguishes imatinib-naïve (IM-n) from imatinib-resistant (IM-r) gastrointestinal stromal tumors (GIST). Fresh frozen tumor samples of IM-n and IM-r GIST patients were subjected to miRNA expression profiling. Depicted is the heat map of a supervised hierarchical clustering based on the 35 most significant (p < 0.01 and False Discovery Rate (FDR) < 20%) differentially expressed miRNAs. In the heat map red indicates relative high expression and green indicates relative low expression. The colored squares beneath the graph designate IM-n and IM-r samples, the malignancy risk and location of the tumors. Note that the sample codes below also indicate which KIT exon is mutated.

Figure 2

MicroRNAs differentially expressed between imatinib-resistant gastrointestinal stromal tumor (GIST) samples with and without secondary mutations in KIT. Depicted is a heat map of a supervised hierarchical clustering based on the 22 most significant (p < 0.01) differentially expressed miRNAs in fresh frozen GIST samples with (green squares) and without (orange squares) secondary imatinib resistance causing mutations in KIT. In the heat map red indicates relative high expression and green indicates relative low expression. Note that the sample codes below also indicate which KIT exon is mutated.

Figure 3

Supervised hierarchical clustering based on differential gene expression discriminates imatinib-naïve (IM-n) and imatinib-resistant (IM-r) gastrointestinal stromal tumor (GIST) samples. Transcript expression profiles were determined using the Affymetrix platform (U133 plus 2) of 29 fresh frozen samples derived of IM-n (n = 14) and IM-r (n = 15) GISTs. Depicted is the cluster tree of a supervised hierarchical clustering based on 352 significant (p < 0.008, False Discovery Rate (FDR) < 10%), differentially expressed transcripts. Note that 100% of the IM-r samples are clustered together with a single IM-n GIST sample. The colored squares beneath the graph designate imatinib-naïve and imatinib-resistant samples, the malignancy risk and location of the tumors (see Figure 1). Note that the sample codes below also indicate which KIT exon is mutated.

Figure 4

Ingenuity Pathway Analysis indicates the relation between genes differentially expressed between imatinib-naïve (IM-n) and imatinib-resistant (IM-r) gastrointestinal stromal tumor (GIST) samples. The 352 significant differentially expressed genes between IM-r and IM-n GIST samples were used as input for an Ingenuity Pathway Analysis (IPA). The depicted IPA networks illustrate and visualize associations between the genes. (A) IPA network highlighting cell cycle related, differentially expressed genes. (B) IPA network highlighting CDK1, AURKB, and FOXM1 interactions. Green and red shading indicates relatively low and high expression in the IM-r samples. See Table S5 for an extended symbol legend.

Figure 4

Ingenuity Pathway Analysis indicates the relation between genes differentially expressed between imatinib-naïve (IM-n) and imatinib-resistant (IM-r) gastrointestinal stromal tumor (GIST) samples. The 352 significant differentially expressed genes between IM-r and IM-n GIST samples were used as input for an Ingenuity Pathway Analysis (IPA). The depicted IPA networks illustrate and visualize associations between the genes. (A) IPA network highlighting cell cycle related, differentially expressed genes. (B) IPA network highlighting CDK1, AURKB, and FOXM1 interactions. Green and red shading indicates relatively low and high expression in the IM-r samples. See Table S5 for an extended symbol legend.

Figure 5

Ingenuity Pathway Analysis integrating differentially expressed genes and microRNAs between imatinib-naïve (IM-n) and imatinib-resistant (IM-r) gastrointestinal stromal tumor (GIST) samples. As input for an Ingenuity Pathway Analysis (IPA) the significantly differentially expressed transcripts (352 genes, p < 0.008, False Discovery Rate (FDR) < 10%) and miRNAs (88 miRNAs, p < 0.03, FDR < 30%) from the same set of IM-r (n = 15) and IM-n (n = 14) GIST samples were used. The depicted network indicates miRNA–gene interactions relevant in context of the cell cycle. Green and red shading indicates relatively low and high expression in the IM-r samples. See Table S5 for an extended symbol legend.