Searchable abstracts of presentations at key conferences in endocrinology
Endocrine Abstracts (2015) 37 GP18.06 | DOI: 10.1530/endoabs.37.GP.18.06

ECE2015 Guided Posters Pituitary–Basic and IGF-1 (9 abstracts)

Comparison of high-throughput platforms in evaluation of whole genome miRNA expression profiles in pituitary tissues

Henriett Butz 1 , Otto Darvasi 2 , Peter M Szabo 1 , Istvan Liko 2 , Lorinc Pongor 2 , Sandor Czirjak 3 , Karoly Racz 1, & Attila Patocs 2


1Molecular Medicine Research Group, Hungarian Academy of Sciences, Semmelweis University, Budapest, Hungary; 2‘Lendület’ Hereditary Endocrine Tumors Research Group, Hungarian Academy of Sciences, Semmelweis University, Budapest, Hungary; 3National Institute of Neurosurgery, Budapest, Hungary; 42nd Department of Medicine, Faculty of Medicine, Semmelweis University, Budapest, Hungary.


Introduction: There are three principal high-throughput methods that have been widely used to determine whole genome miRNAs expression profiles: i) microarrays, ii) qPCR based arrays, and iii) next generation sequencing (NGS). Our aim was to compare the results obtained from these platforms in normal and adenomatous pituitary samples and to validate the results by an independent sample set using qPCR.

Material and methods: Using four normal pituitary (NP) and eight non-functioning pituitary adenoma (NFPA) samples we determined miRNA expression profile by GeneChip microRNA Galaxy Array v1, SOLiD NGS and qPCR based TaqMan Low Density Array Card (TLDA). For biological confirmation expression of 22 miRNAs by individual TaqMan assays on additional 24 NFPAs and ten NPs was used as validation cohort.

Results: Totally 848 miRNAs of which only 162 were detected by all three approaches. Significant but not strong correlations between microarray and NGS (R2: 0.471 and 0.220; P<0.01), microarray and TLDA (R2: 0.462 and 0.339; P<0.01), and NGS and TLDA (R2: 0.353 and 0.290; P<0.01) were detected in NFPAs and NPs respectively. Various bioinformatics, including commercially available software CLCbio, free, web based (Bowtie) and one our own algorithms were used in order to test whether the different evaluation of NGS data would affect the correlations. Correlations between different bioinformatical approaches were strong (R2: 0.96–0.99) suggesting that these algorithms are useful and give similar results. Biological validation showed that 81% of TLDA results, and 72% of both microarray and NGS results could be validated regarding the direction of expression changes.

Conclusion: miRNA expression profiles measured by different platforms showed poor correlation and they were hardly comparable. Consequently, selection of screening method can influence experimental results obtained by analyses using high-throughput data (e.g. pathway analysis). However, individual miRNA expression from microarrays and NGS results were replicable in an acceptable percentage by qPCR.

Disclosure: KTIA-201200010.

Article tools

My recent searches

No recent searches.