Best Practices for Unbiased and Reproducible dPCR Runs

Digital PCR (dPCR) has enjoyed considerable growth in its usage over the past decade. Yet despite its binary output, ease-of-use, forgiving nature, as well as other unique advantages and its copious applications, it is not foolproof. To maximize its performance and attain all its benefits, certain guidelines should be followed. This article will look at some of the factors that should be considered in order to ensure unbiased and reproducible dPCR runs.

How dPCR yields its benefits

“Unlike traditional methods, dPCR compartmentalizes the nucleic acid sample into thousands of individual reactions, enabling the detection and quantification of even a single target molecule without the need for a standard curve,” explains Lama Kdouh, Director of Global Market Development, Genetic Analysis Division at Thermo Fisher Scientific.

Each of these reactions—whether taking place in an oil droplet or a physical microwell—will contain on average a single DNA template molecule. If the sequence of interest is present on the template it will amplify, causing the partition to fluoresce. If the sequence is not on the template, the partition will not amplify and the partition will not fluoresce. Upon analysis the proportion of fluorescence-positive partitions indicates the concentration of the target sequence in the original sample. A sequence with a frequency of 1/1000 will yield 20 positives in a run with 20,000 partitions.

Its high sensitivity and ability to provide absolute quantification make dPCR ideally suited for detection of single nucleotide polymorphisms (SNPs) and copy number variations (CNVs), and allows for such applications as detecting rare alleles and liquid biopsies to monitor residual disease in cancer patients.

Pre-run

Even before the actual PCR run, many things from lab basics to assay planning can affect whether the end result is free of bias and reproducible.

For example, pipettes should be calibrated, and it’s best not to pipette less than 5 µL, advises Marwan Alsarraj, Assistant Director of Marketing for Digital Biology with Bio-Rad. It’s also a good idea to centrifuge your sample to eliminate debris, adds Kdouh. And Jo Vandesompele, Director of Scientific Initiatives at CellCarta, reminds users that reagent stability and batch or lot variability can affect reproducibility and are often ignored, as are numerous other pre-analytical variables that can impact accuracy as well.

Errors can be introduced during the extraction process, the template modification process, and during reverse transcription. Large pre-analytical errors can compromise precision and introduce bias. All procedures including, but certainly not limited to, sampling, extraction, qualitative and quantitative assessment, and storage, must be documented and reported.

Stilla Technologies’ co-founder and CTO Rémi Dangla advises to try and have the simplest workflow possible—from the raw sample to the result—to minimize user intervention. That includes the dPCR consumable and system as well: “having a single consumable, single pipetting approach is, I think, quite important for reproducibility.” Just load the PCR mix into the cassette and the system will process the sample inside the microfluidic chip, doing the partitioning, PCR amplification, and the readout all inside the cassette. Automation also helps with reproducibility, he adds.

The dynamic range of dPCR is less than that of qPCR, and so it may be necessary to have prior knowledge of template concentration in order to avoid saturation (which would underestimate the number of positive partitions). Note that spectrophotometric and fluorometric procedures estimate the mass of nucleic acid present rather than molar quantity. On a similar note, Kdouh notes that “having primer and probe concentrations that are not limiting in dPCR experiments is crucial for obtaining reliable and accurate results.”

You might need to do a restriction digestion on the sample for copy number studies. Otherwise, linked targets “might co-localize in the same partition and be counted as one, even though there are two copies,” cautions Alsarraj.

Controls

It is difficult to put too much emphasis on the use of controls to ferret out accuracy, bias, and reproducibility issues in PCR-based studies. You want to know if the assay amplifies what, and only what, it is supposed to, whether there is cross- or carryover contamination, as well as what the range of the assay is and whether it is reproducible.

Negative controls—master mix, primers, and probe, but with water in place of template DNA—will identify contamination by amplifiable DNA from another sample or a previous experiment, points out Alsarraj. A related control uses “negative DNA,” instead of the water, to assess the specificity of the reaction—especially important to have confidence in low false positives while trying to pick out rare variants from a sea of wild-type sequences.

It’s really important to establish the limit of blank (LOB), emphasizes Dangla. By running perhaps 20 or 30 samples without any template, “you’ll see experimentally that you don’t get pure negative result on negative controls all the time. There is noise out there.”

“It’s also very important to assess the sensitivity of the dPCR assay and system,” he continues. This is done by spiking in diminishing amounts of positive control templates to establish the (lower) limit of detection (LOD).

Quality control, too, helps to obtain unbiased, reproducible results. This means not only assuring that the samples are a known quantity and quality, and assays work as advertised (running positive and negative controls help to establish the latter). It also applies to “using systems that allow you to go deep into data—to look at the raw data—to make sure that rare calls are true positive calls,” Dangla says, pointing out that a spurious fluorescent signal conceivably could come from a dust particle.

dMIQE

Much of this article points out what may to many be carryovers of lessons from other PCR experiments, or just common sense. Yet we hope that it also provides some insights into avoiding bias and assuring reproducibility in dPCR, which may not be so obvious. To delve deeper, see dMIQE2000—what amounts to the industry’s dPCR user’s manual.

The objective of the dMIQE2000 [Minimum Information for Publication of Quantitative Digital PCR Experiments for 2020] guidelines “is to create awareness for the user which workflow elements are important and may go wrong, and to transparently report what was done, so others can replicate and critically appraise,” notes Vandesompele, a member of the dMIQE Group.

The guidelines are “very comprehensive. They’re very good at highlighting everything that can impact the quality of the data and digital PCR,” Dangle elaborates. “If you follow them carefully then you’re very unlikely to make mistakes.”

Key Takeaways

• There are differences between qPCR and dPCR to take note of
• Pay attention to lab basics and assay planning
• Try to have the simplest workflow possible
• Don’t forget to document and report all procedures
• The use of controls is essential
• Establishing the limit of blank (LOB) and limit of detection (LOD) are important
• Review the dMIQE2000 guidelines