Simple Polydisperse Droplet Emulsion Polymerase Chain Reaction with Statistical Volumetric Correction Compared with Microfluidic Droplet Digital Polymerase Chain Reaction

Abstract

Nucleic acid amplification technology, such as polymerase chain reaction (PCR), has enabled highly sensitive and specific disease detection and quantification, leading to more accurate diagnosis and treatment regimens. Lab-on-a-chip applications have developed methods to partition single biomolecules, such as DNA and RNA, into picoliter-sized droplets. These individual reaction vessels lead to digitization of PCR enabling improved time to detection and direct quantification of nucleic acids without a standard curve, therefore simplifying assay analysis. Though impactful, these improvements have generally been restricted to centralized laboratories with trained personnel and expensive equipment. To address these limitations and make this technology more applicable for a variety of settings, we have developed a statistical framework to apply to droplet PCR performed in polydisperse droplets prepared without any specialized equipment. The polydisperse droplet system allows for accurate quantification of droplet digital PCR (ddPCR) and reverse transcriptase droplet digital PCR (RT-ddPCR) that is comparable to commercially available systems such as BioRad's ddPCR. Additionally, this approach is compatible with a range of input sample volumes, extending the assay dynamic range beyond that of commercial ddPCR systems. In this work, we show that these ddPCR assays can reduce overall assay time while still providing quantitative results. We also report a multiplexed ddPCR assay and demonstrate proof-of-concept methods for rapid droplet preparation in multiple samples simultaneously. Our simple polydisperse droplet preparation and statistical framework can be extended to a variety of settings for the quantification of nucleic acids in complex samples.

General methods for quantitative interpretation of results of digital variable-volume assays

“…Digital PCR experiments were performed as previously described. 14 In each digital PCR experiment, polydisperse droplets were prepared using 100 μL of the PCR reagents (see details below) and 200 μL of BioRad oil/surfactant mixture (BioRad, Hercules, CA, USA). These fluids were vortexed at maximum speed for 30 seconds to create a population of polydisperse droplets with diameters in the range of 1.5 to 13 000 μm.…”

“…We characterized the truness of the methods by applying them to simulated assay results and experimental digital PCR results. 14 In particular, the metric is the bias in the natural logarithm of the concentration, which is the difference between the estimators (Λ) obtained by the specified methods and the corresponding known quantities (Λ) set in the simulation or measured by a reference method (in the cases of experimental results).…”

“…7,8 These systems require the compartments to have uniform volumes, 9 posing a challenge in precise microfabrication and limiting total addressable sample volumes due to the volume limitations of most microfluidic systems. Simple, non-microfluidic, methods to make compartments typically result in widely distributed compartment volumes, [10][11][12][13][14] making quantification inaccurate using traditional statistics. 15 Herein, we report general statistical methods to quantify results of digital assays with any volume distributions, termed Digital Variable Volume (dvv), and Digital Variable Volume Approximation (dvva).…”

“…17 These methods provide accurate results when the volumes follow the specified distributions, or when the standard deviations are small. However, in systems where compartments are made by a simple method, such as vortexing, 13,14 the volume distribution does not have a known analytical form and spans orders of magnitude. A method to account for general volume variation was previously published 18 but has not been shown to work with very wide volume distributions (e.g.…”

“…A method to account for general volume variation was previously published 18 but has not been shown to work with very wide volume distributions (e.g. those of manually made droplets 13,14 with standard deviations exceeding a few times the mean). Another method was built upon results for multivolume digital PCR (MVdPCR) 19 but with the assumption that the compartment volumes are non-identical, and without the confidence intervals of the estimated concentrations.…”