Skip to main content

MicroRNA profiling in plasma samples using qPCR arrays: Recommendations for correct analysis and interpretation.

Gevaert AB, Witvrouwen I, Vrints CJ, Heidbuchel H, Van Craenenbroeck EM, Van Laere SJ, Van Craenenbroeck AH

Publication: PLoS One (2018) • Vol 13 • Page e0193173PMID: 29474497View on PubMed

Purpose

The purpose of this paper was to investigate the effects of elution volume, dilution of the preamplified product before analysis, and normalization method on the results of TaqMan low density arrays (TLDA).

Conclusion

The lowest mean quantification cycle (Cq) values were obtained when miRNA was eluted into 100 µl extraction buffer (rather than 50 or 150 µl) and when the preamplified product was diluted 1:1. The authors state that both the geNorm and Norm-Finder strategies resulted in low technical variability and stability coefficients (highest stability) while retaining biological variability, but geNorm was the superior choice for normalization due to lower data dispersion when using local regression plots. Normalization to U6, the global mean, or ath-159a resulted in inferior results and normalization to RNU44 and RNU48 was not possible as they were not found in all specimens.

Studies (1)

Purpose

The purpose of this study was to investigate the effects of elution volume, dilution of the preamplified product before analysis, and normalization method on the results of TaqMan low density arrays. Blood from 19 patients with clinically stable heart failure was collected into EDTA Vacutainer tubes using a 19-gauge needle. Plasma was obtained by centrifugation at 1500 x g for 15 min within 30 min of blood collection. Plasma was immediately aliquoted and frozen at -80°C for 164-3181 days (median 2425 days).  To optimize the extraction protocol, three plasma aliquots from a single patient were thawed on ice and miRNA isolated using a modified mirVana Paris Kit into 50, 100, or 150 µl water. Extraction was then performed using the same method with the inclusion of Arabidopsis thaliana miR-159a for all of the other aliquots and a 100 µl extraction volume. RNA was stored 3-21 days (median 8.5 days) after isolation before real-time PCR was performed. RNA was reverse-transcribed using TaqMan MicroRNA Reverse Transcription Kit and MegaPlex Human Pool A RT primers and then preamplified using MegaPlex Human Pool A PreAmp primers. The preamplified product was diluted in 0.1 x Tris EDTA at 1:1, 1:4, or 1:40 to determine the best dilution before real-time PCR using TaqMan Low Density MicroRNA Array Human Cards A within 72 h of reverse transcription. The data was normalized using the spike-in control ath-miR-159a, endogenous control miRNAs (RNU44, RNU48, and U6), geNorm, Norm-Finder, and the Global mean calculation.

Summary of Findings

The lowest mean quantification cycle (Cq) values were obtained when miRNA was eluted into 100 µl extraction buffer (rather than 50 or 150 µl). The mean Cqs were lower and percentage of wells with amplification higher when the preamplified product was diluted 1:1 rather than 1:4 or 1:40. Of the 377 miRNA investigated, 233 had Cq values of ≤35 in more than 20% of the specimens but 15 of these were excluded due to a CV of >4%. The remaining miRNAs were considered informative and consisted of 181 miRNAs with a CV <4% and 37 miRNAs with unknown CV. RNU44 and RNU48 were not found in all specimens and thus excluded from further analysis. The technical variability was lowest when Norm-Finder was used (2.13%) and highest when U6 was applied (9.96%). The stability coefficient M was lowest when geNorm was used (M=0.618), indicating the most stability; followed by when Norm-Finder was applied (M=0.0633). Similarly, the stability coefficient ρ was lowest when Norm-Finder (ρ=0.0132) was applied, indicating the most stability; followed by geNorm (ρ=0.0197). Much higher stability coefficients were observed for M and ρ when data was normalized to ath-miR-159a (M=0.1101, ρ=0.0598), global mean (M=0.0943, ρ=0.0550), or U6 (M=0.1228, ρ=0.0436). Importantly, the biological variability remained high at 7.87%, 7.28%, and 9.96% after normalization using geNorm, Norm-Finder, and U6; respectively, but was only 4.62% and 2.14% after normalization to ath-miR-159a or the global mean, respectively. The overall CV was lowest using geNorm. Finally, local regression plots showed the least dispersion when geNorm or Norm-Finder were applied with slightly better results with geNorm. The authors stated that both geNorm and Norm-Finder performed well but geNorm was the superior choice for normalization due to lower data dispersion when using local regression plots.

  • Biospecimens

    FluidPlasma
  • Preservation Types

    Frozen
  • Diagnoses

    Other diagnoses
  • Platforms

    RNAReal-time qRT-PCR
    RNALow density array
  • Pre-analytical Factors

    ClassificationPre-Analytical FactorValue(s)
    Analyte Extraction and PurificationAnalyte isolation method
    Eluted into 100 µl
    Eluted into 150 µl
    Eluted into 50 µl
    Real-time qRT-PCR SpecificData handling
    Normalized to U6
    Normalized to ath-miR-159a
    Normalized using Norm- Finder (hsa-miR-106b)
    Normalized using geNorm (geometric mean of hsa-miR-17 and hsa-miR-106a)
    Normalized using the global mean
    Real-time qRT-PCR SpecificTemplate/input amount
    Diluted 1:1
    Diluted 1:4
    Diluted 1:40