While cancer's microRNA signature continues to yield to scientific scrutiny, tricky sample prep and the limitations of qPCR have hampered use of miRNA expression patterns as non-invasive biomarkers.
A pair of recently published studies from the University of Maryland School of Medicine, however, now show droplet digital PCR to be an effective tool for quantifying low-abundance circulating miRNAs from both plasma and sputum of lung cancer patients.
Feng Jiang and colleagues undertook the two efficacy studies to determine whether ddPCR could be used to quantify miRNA in lung cancer. Although previous studies have shown ddPCR to be a sensitive and precise way to detect low copy number nucleic acids, relative to qPCR, Jiang noted a paucity of data in the literature applying ddPCR technology specifically to clinical samples.
The group previously published a substantial body of work correlating different miRNAs with lung cancer, using standard qPCR on both sputum and plasma samples. They also showed a panel of miRNA biomarkers could be used to enhance computed tomography in lung cancer diagnosis, and help distinguish the small fraction of solitary pulmonary nodules that are in fact tumors.
His group is now one of a handful pursuing droplet digital PCR to quantify low-abundance, or diluted, nucleic acid signatures in clinical samples.
In one of the recent studies, published in The Journal of Cancer Research and Clinical Oncology, the group showed that miR-31 and miR-210 could be amplified by ddPCR from sputum, and that these miRNAs were differentially expressed in lung cancer patients. Two other miRNAs, miR-21-5p and miR335-3p, were detectable in plasma and showed expression differences correlated with lung cancer. The later study was published in Biomarker Insights.
Determining the best endogenous miRNA controls for a given sample source has been problematic, according to the studies. There exist no widely accepted standards, and "spike-in" controls don't account for template-specific effects or bias due to primer design, the researchers claimed. Droplet digital PCR could get around these issues, and additionally has the advantages of not relying on Ct values or calibration curves.
In the plasma study, the researchers ran a head-to-head comparison of ddPCR and qPCR. The qPCR was run on an Applied Biosystems 7900HT thermocycler, while the ddPCR was carried out using Bio-Rad's QX100 droplet generator, with PCR amplification in Applied Biosystems' T100 thermocycler, and droplet reading on the Bio-Rad droplet reader.
Both techniques detected the higher concentrations of artificially seeded miR-21-5p and miR335-3p. But the study reported qPCR had no significant amplification signal in samples with fewer than 100 copies of miRNA per microliter input. It turned out the dynamic range of the ddPCR was narrower than qPCR, with ddPCR able to pick up between 1 and 10,000 copies of miRNA, but qPCR detecting levels between 100 and 10 million copies.
The researchers next isolated the same two miRNAs from a lung cancer cell line and ran a series of dilutions. Again, qPCR was unable to produce amplification signals in highly-diluted samples.
Clinical samples showed similar results. The researchers isolated miRNAs from 36 lung cancer patients (all stage I non-small-cell lung cancer, including both adenocarcinoma and squamous cell carcinoma), versus 38 cancer-free controls. In these cases, clinical samples were from blood draws into EDTA tubes, with plasma isolated within two hours and subsequently stored at -80°C. For the miRNA known to be more highly abundant, miR-21-5p, there was no significant difference in copy number determined by the two techniques. For the low-abundance miRNA, mi-335-3p, ddPCR showed a copy number of about 100 per microliter of plasma, which proved to be too low for qPCR to amplify and generate a Ct curve.
Effectively detecting miRNAs may be moot, however, if they have no diagnostic potential. The study ultimately revealed a statistically significant difference between lung cancer patients and cancer-free controls in terms of copy numbers of these two miRNAs in blood plasma. However, the authors determined that the sensitivity and specificity of using only these two miRNAs as a diagnostic was likely too low to be useful in a clinical setting. The researchers also pointed out that miRNAs are dysregulated with other types of cancers, and, because of these and other confounds, expanding their results to a larger cohort and a broader panel of miRNAs would be necessary.
Importantly, this study also emphasized a potential weakness of ddPCR in diagnostic applications. While qPCR could not amplify miRNA signal below a threshold of 100 copies per microliter of plasma, concentrations above 10,000 in ddPCR could lead to droplets with two or more templates, defeating the digital nature of the assay and making it ineffective. Ultimately the authors proposed a synergistic approach, using qPCR to assay miRNAs suspected to be abundant, and ddPCR for those likely to be in lower concentrations.
In the second study, analyzing clinical samples of sputum, the group looked at miR-31 and miR-210. Here, the head-to-head portion of the study showed digital PCR and qRT-PCR to agree in terms of copy numbers and expression levels. The researchers also concluded that the digital nature of ddPCR meant data handling was more straightforward, and the platform was robust and able to obtain absolute quantification without external references.
This recent work is in harmony with the conclusions of a study published in September in Nature Methods, in which the authors ran nested analyses of ddPCR versus qPCR on cDNA from a dilution series of six synthetic miRNAs in both water and plasma on three different days. That study found a lower coefficient of variation in ddPCR. Then, in a side-by-side comparison of miR-141 in serum from patients with prostate cancer, they found ddPCR improved day-to-day reproducibility seven-fold, and improved the confidence over qPCR in differentiating cases from controls.