Digital PCR (dPCR) is a refinement of PCR where DNA is subject to PCR for amplification of the template, then the sample undergoes fluorescent detection and sequence-specific alleles can be directly counted. This method offers a highly sensitive quantification technique where only a very small sample is needed for starting material. dPCR is independent of amplification efficiency, and results are extremely repeatable.

Many times, dPCR is used even to retest the results of next generation sequencing (NGS). dPCR may be used to detect rare mutations in a bulk of wild type (WT) sequences, notably for mutational tumor analysis or for other neoplastic diseases and cancers. dPCR may also be used to assess allelic imbalances in samples, including tissues and plasma, and can detect chromosomal aneuploidy or even autosomal recessive disorders.

dPCR is especially useful for non-invasive prenatal testing that is based on the characterization of circulating cell-free fetal DNA in maternal plasma used for the detection and identification of genetic abnormalities.  

First a genomic DNA (gDNA) sample is diluted in a multi-well plate with one template molecule per two wells, on average. Though any well plate may be used, it is important to remember that the high throughput capabilities of this technique allow many samples to be analyzed simultaneously. Optimal dilution of gDNA can also be readily achieved by commercially available DNA quantification kits.

Next, PCR is performed on the sample. Optimization of the primers and experimental conditions is sample-specific, and key for success at this step. After PCR, amplicons in the sample are hybridized with fluorescent probes that allow for the detection of sequence-specific products using various fluorophores.

Typically, molecular beacons may be used as the fluorescent probes. These molecular beacons are a single strand oligonucleotide with a fluorescent dye on the 5' end and a quencher on the 3' end. Structurally, molecular beacons are hairpin shaped, which allows the fluorophore to be close to the quencher, though makes sure that the molecule does not emit fluorescence when not hybridized to a PCR product. Two molecular beacons are used that produce two separate fluorescent signals, where one hybridizes to the sequence with the mutation and the other hybridizes to the WT sequence only. The fluorescent signals produced then report the mutational status of a specific allele, and the fluorescence intensity of the two beacons can be determined for each well, giving the ratio of the two sequences.

Molecular beacons are especially useful for mutational analysis detection as they are highly sensitive, even if relatively low levels of mutations are present. dPCR software can then directly count the number of each of the two alleles in the sample.

Upon hybridization between the molecular beacons and their complementary sequences, the quencher is distanced from the fluorophore resulting in an increased fluorescent signal. As two beacons are used, upon visualization each well will present with a specific color: one color to indicate WT only, another color to indicate the mutant sequence. Depending on the visualization software used, a third color may indicate the presence of the dually active sample, and it is also possible that some wells will not present with any colors, meaning the well does not contain any PCR product. In some instances, it may also only be necessary to use one fluorescent probe, meaning that during visualization only one color, or none, may be present.

Note: Failed wells, as well as those that emit extremely low or high fluorescence, may be placed in boundaries to not sway analytical data.

Various analysis software suites are available, though the statistical foundation of each is based on principles of binomial probability and the Poisson approximation. Analysis methods may then be used to evaluate the strength of evidence for the loss of heterozygosity (LOH) in each sample. Analysis may also provide copy number variations (CNV) which are the gains or losses of genomic regions > 500 bases in size. CNV may be particularly useful for research related to whole genome studies that focus on the role of CNVs in human genetic disorders.

Two general approaches for dPCR exist: chip-based (cdPCR) and droplet-based methods (ddPCR).

For cdPCR, the chip is composed of physically isolated, compartmentalized, nanoliter size chambers. cdPCR is based on integrated fluidic circuits, an arrayed lipid bilayer chamber system, and a self-priming chip. The chip itself is made from PDMS from soft lithography, and may contain fluid lines, valves, or even reaction chambers within the single device.

Adversely, ddPCR uses a water-in-oil microfluidic technology and flow cytometry to count positive PCR reactions. In ddPCR, the reactions are prepared in a tube, partitioned into individual droplets using a droplet generator, then transferred to a plate for PCR amplification. A two-color optical detection system will then read signals in each droplet. It is important to note that the oil chosen must be nonreactive, and must be capable of forming stable microreactors to prevent the diffusion of the reaction reagent. Additional appropriate surfactants may also be required to stabilize the water-oil interface and prevent coalescence of the emulsion.

dPCR, the third generation of the technique, has many benefits over its predecessor, RT-qPCR. dPCR is not as vulnerable to contamination as fluorescence detection is a simplified process. dPCR is also capable of providing absolute, versus relative, quantification, and can be used even with low copy number genes. dPCR may be experimentally faster than RT-qPCR since it does not require a standard curve calibration, that also may require expensive reagents. dPCR also offers increased resistance to PCR inhibitors over other techniques.

It is important to note that the sensitivity of mutational detection depends on a number of factors, including the number of wells included for analysis and the intrinsic mutation rate of the polymerase used in amplification. False positive and false negative events must also be taken into consideration before statistical analysis of the sample is performed. dPCR provides endpoint results, as opposed to those in real-time, so this must be a consideration based on the information needed.

Currently, there are few new methods that allow further optimization of dPCR, and though an incredibly useful technology, it might be unrealistic to invest in the costly dPCR system if a high-throughput RT-qPCR system is already in place.