Skip to main content

Optimization and Standardization of Circulating MicroRNA Detection for Clinical Application: The miR-Test Case.

Marzi MJ, Montani F, Carletti RM, Dezi F, Dama E, Bonizzi G, Sandri MT, Rampinelli C, Bellomi M, Maisonneuve P, Spaggiari L, Veronesi G, Bianchi F, Di Fiore PP, Nicassio F

Publication: Clin Chem (2016) • Vol 62 • Page 743-54PMID: 27127244View on PubMed

Purpose

This paper developed a workflow for the analysis of microRNA (miRNA, miR) in serum and investigated the effects of patient fasting status, clot time, hemolysis and serum storage on levels of a lung cancer biomarker and control miRNAs targeted by the real-time PCR based miR-Test.

Conclusion

The quantification values were lowest and least variable when RNA was purified using miRNeasy columns manually from serum volumes of up to 400 µL and when a pre-amplification step was included.  The authors identified a panel of 13 lung cancer biomarkers and 6 housekeeping genes with low Cq values (<30, mean 21.9) and very strong correlations in levels in serial dilution experiments from which they designed the miR-Test. Inter-assay variance for the whole assay was low (<0.2 Cq), although more variance was attributed to extraction than real-time PCR or preamplification. For the most part, variance in real-time PCR results was attenuated by normalizing raw data to either the 6 stable housekeeping genes identified or to the spike-in control. Levels of the 13 biomarkers in the miR-Test changed based on fasting status and serum storage temperature, and to a lesser extent the clot time; however, miR-Test score was unaffected by hemolysis. Nevertheless, a hemolysis sensor based on the ratio of expression of three hemolysis-sensitive miRNAs (miR-15b, miR-19a, and miR-19b) to a housekeeping gene that was unaffected by hemolysis (miR-146a) was very strongly correlated to the extent of sample hemolysis. In the larger cohort of 972 specimens, there was no apparent effect of hemolysis sensor score on the miR-Test score when specimens were analyzed by quartiles of hemolysis sensor value. However, the authors found a difference in extraction efficiency in this larger cohort between the first extraction of the day and subsequent extractions which seems to correspond with a shift in the machine temperature and humidity; however, these differences were attenuated by data normalization to the 6 stable housekeeping genes that were identified by the authors.

Studies (2)

Purpose

This study compared real-time PCR quantified miRNA levels and inter-and intra-assay variability of both raw and normalized data collected from a range of serum input volumes that were extracted using one of two different miRNA purification columns.Blood was collected into S-Monovette tubes with clot activator from 972 healthy and lung cancer patients in the COSMOS lung cancer screening trial after a fast longer than 3 h. The first 3 mL of blood collected was discarded. Blood was allowed to clot for 3 h at room temperature before separation of plasma by centrifugation at 1000 g for 10 min. Serum was then transferred to cryotubes and snap-frozen on dry ice prior to storage at -80°C. To test the effects of purification method, RNA was extracted from two serum specimens using TriZol-LS and then purified with miRVANA columns or miRNeasy columns both manually and using a QIAcube. To test the effect of serum volume, RNA was purified from two 100, 200, 300, 400 and 500 µL serum aliquots using miRNeasy columns with the QIAcube instrument. For all other experiments, RNA was extracted from 300 µL serum using TriZol-LS and purified from 350 µL of the aqueous phase using the miRNeasy Mini Kit and a QIAcube machine. miRNAs were reverse transcribed using the MicroRNA Reverse Transcription Kit and pre-amplified using TaqMan PreAmp before real-time PCR amplification of miR-16, miR-19b and spiked-in miR-34a (positive control) or use of TaqMan low density arrays containing primers for 48 miRNAs of interest (6 housekeeping, 34 of interest and 8 controls). The control miRNA in the TaqMan assays included: three miRNAs with no or extremely low serum levels (miR-34a, miR-159 and cel-miR-39), three miRNAs expressed in serum known to be involved in other diseases (miR-21, miR-122, and miR-155), and two small RNAs (RNU44, RNU48).  The hemolysis index was measured by spectrophotometer.

Summary of Findings

Cq values were lower and less variable when RNA was purified using miRNeasy columns manually or using QIAcube rather than miRVANA columns. Endogenous miRNA levels increased with increasing serum input up to 400 µL; however, a miRNeasy column input volume of 500 µL did not increase endogenous miRNA levels and resulted in lower recovery of spiked in miR-34 compared to volumes between 100-400 µL.  Collectively, data suggests that volumes >400 µL overload the column. Pre-amplification prior to real-time PCR lowered mean CT from 31 to 23 while remaining linear. A panel of 13 biomarkers and 6 housekeeping genes were found to have low Cq (<30, mean 21.9) and very strong correlations in serial dilution experiments (R2>0.90). The remaining miRNA were considered to have failed validation and were excluded from further analysis. Inter-assay variance for the whole assay was low (<0.2 Cq), with more variance attributed to extraction than real-time PCR or pre-amplification (P<0.01). Nevertheless, when extraction was performed on different days the variance remained low. Normalization of the raw data to either the 6 housekeeping genes or to the spike-in control reduced the variability in levels of the 13-diagnostic miRNA chosen for the miR-Test. The low inter-assay variability was confirmed by repeating the extraction step with 12 serum specimens from healthy individuals and with 12 serum specimens from patients with asymptomatic lung cancer. In specimens from 972 patients in the COSMOS study, the authors found a difference in extraction efficiency between the first extraction of the day and subsequent extractions which seemed to correspond with a shift in the machine temperature and humidity.  However, this effect was attenuated when real-time PCR miRNA levels were normalized to the 6 stable housekeeping genes that were identified.

  • Biospecimens

    FluidSerum
  • Preservation Types

    Frozen
  • Diagnoses

    NeoplasticCarcinoma
    Normal
  • Platforms

    RNAReal-time qRT-PCR
  • Pre-analytical Factors

    ClassificationPre-Analytical FactorValue(s)
    Biospecimen Aliquots and ComponentsAliquot size/volume
    100 µL,
    200 µL,
    300 µL,
    400 µL,
    500 µL,
    Analyte Extraction and PurificationAnalyte isolation method
    miRNeasy automated on QIAcube
    miRNeasy manual
    miRVANA
    Real-time qRT-PCR SpecificNucleic acid amplification
    Not pre-amplified
    Pre-amplified