Exploring the nature of prediagnostic blood transcriptome markers of chronic lymphocytic leukemia by assessing their overlap with the transcriptome at the clinical stage

Abstract

BackgroundWe recently identified 700 genes whose expression levels were predictive of chronic lymphocytic leukemia (CLL) in a genome-wide gene expression analysis of prediagnostic blood from future cases and matched controls. We hypothesized that a large fraction of these markers were likely related to early disease manifestations. Here we aim to gain a better understanding of the natural history of the identified markers by comparing results from our prediagnostic analysis, the only prediagnostic analysis to date, to results obtained from a meta-analysis of a series of publically available transcriptomics profiles obtained in incident CLL cases and controls.ResultsWe observed considerable overlap between the results from our prediagnostic study and the clinical CLL signals (p-value for overlap Bonferroni significant markers 0.01; p-value for overlap nominal significant markers < 2.20e-16). We observed similar patterns with time to diagnosis and similar functional annotations for the markers that were identified in both settings compared to the markers that were only identified in the prediagnostic study. These results suggest that both gene sets operate in similar pathways.ConclusionAn overlap exists between expression levels of genes predictive of CLL identified in prediagnostic blood and expression levels of genes associated to CLL at the clinical stage. Our analysis provides insight in a set of genes for which expression levels can be used to follow the time-course of the disease; providing an opportunity to study CLL progression in more detail in future studies.Electronic supplementary materialThe online version of this article (doi:10.1186/s12864-017-3627-4) contains supplementary material, which is available to authorized users.

Evolving DNA methylation and gene expression markers of B-cell chronic lymphocytic leukemia are present in pre-diagnostic blood samples more than 10 years prior to diagnosis

“…Eleven of the Bonferroni- and 34 of the FDR-significant genes have been reported [ 32 , 33 ] to be differentially expressed in B-cells isolated from CLL patients. Furthermore, 22 FDR-significant DE genes are among those found in CLL, with the same direction of change (Additional file 2 : Table S5), in a meta-analysis of transcriptomic profiles of CLL patients [ 18 ]. These observations further support the suggestion that cells with changes characteristic of clinical CLL were present in our prediagnostic samples.…”

“…Of the 238 DM CpGs observed in the long TtD sub-group, 168 (70.5%) have been reported to be differentially modified in clinically diagnosed CLL [ 24 ]. Furthermore, of the 21 DE genes significant in the same sub-group, 12 are among 2095 genes (hypergeometric distribution test p = 0.013) reported to be differentially expressed in clinical CLL [ 32 , 33 ] and 9 are among those found significant in a corresponding meta-analysis included in the report by Vlaanderen et al [ 18 ]. Finally turning to miRNA, examination of the limited number of subjects for which data were available indicates that, while no significant changes in miRNA expression could be detected in the long TtD group, 3 miRNAs (miR-155-5p, miR-150-5p, both overexpressed, and miR-4486, underexpressed) were significant (FDR < 0.05) in the short TtD group.…”

“…Although the numbers of differential signals detected after adjustment for WBC composition are dramatically decreased relative to those found without such adjustment (Additional file 2 : Table S2) [ 17 ], the majority of these signals (approx. 70% for both methylation and gene expression, both miRNAs), are known to be similarly modified in clinical CLL [ 18 , 24 , 32 – 34 , 37 ]. The same holds for the signals detected in the sub-groups of cases with B-cell fraction <10% (Additional file 2 : Table S6) and TtD > 7.4 years (Additional file 2 : Table S7), supporting the idea that the differential risk profiles identified are unlikely to represent false findings resulting from residual confounding by variations in WBC composition but are in fact associated with the early phase of the pathogenesis of CLL.…”

“…On the other hand, we recently reported that a gene expression profile measured in peripheral blood leukocytes could identify with high accuracy individuals who were diagnosed with CLL 2–17 years later [ 17 ] and highlighted genes whose expression was deregulated long before disease diagnosis. In extended analyses we showed that these gene expression profiles showed marked overlap with expression profiles of clinical CLL samples [ 18 ] suggesting that circulating cells long before diagnosis harbor CLL-traits. A common limitation of these studies lies in the lack of information regarding the hematological status of the study subjects at recruitment, including the presence or not of undiagnosed CLL.…”

Chronic lymphocytic leukaemia clones are detectable decades before diagnosis

“…Only a few studies of genomic, transcriptome, and methylation profiles have been published. 8,10,11,12 However, plasma markers would be preferable from a logistic perspective; in the CCHS and CGPS a high plasma total immunoglobulin E (IgE) level has been associated with decreased risk of CLL. 13 Thus, differences in immunoglobulin levels among other plasma markers could serve as a potential marker for CLL and hypothetically MBL.…”