Optimization and Expansion of Digital PCR Multiplexing for Rare Mutation Detection

Introduction:: Circulating tumor DNA (ctDNA) is a promising blood-based biomarker that is increasingly used to monitor cancer progression. ctDNA is heavily fragmented cell-free DNA released by cancer cells during apoptosis and reflects the mutational landscape of the originating tumor. ctDNA can be used to predict cancer relapse, evaluate treatment efficacy, and detect emergent chemoresistance^1,2,3. We have previously developed a digital PCR platform for ultrasensitive ctDNA detection in early-stage pancreatic cancer ⁴. Multiplex Enrichment using Droplet Pre-amplification (MED-Amp) is a two-step digital PCR amplification strategy targeting the four most common KRAS mutations in pancreatic cancer (Figure 1A). MED-Amp assay sensitivity is comparable to gold-standard next-generation sequencing approaches but requires significantly less input DNA (≥ 5 ng)^4,5. The original version of the MED-Amp assay traded mutation specificity for increased resolution between positive droplets (Figure 1B). We now present an experimental and computational framework for optimizing a higher-order multiplex panel. Our aim is to achieve complete discrimination of mutation clusters while maintaining the sensitivity and specificity of the MED-Amp assay.


Materials and Methods:: Samples & Assay Design: We designed 125 base pair gblocks for KRAS G12C/D/R/V and wild-type to simulate the fragment size of ctDNA.⁶ Assays were engineered to ensure consistent amplification performance in a multiplex reaction. DNA preamplification: Preamplification was performed as previously described, briefly DNA was partitioned into nanoliter sized droplets with a high-fidelity polymerase master mix and underwent 9 rounds of amplification. Droplets were de-emulsified and amplicons were recovered using a PCR cleanup kit before re-partitioning with TaqMan probes followed by downstream detection using dPCR. Experimental optimization and modeling: Preamplification experiments used a Box-Behnken design to reduce total cost and time compared to factorial experiments. Annealing temperature, annealing time, and primer concentration were varied deterministically across thirteen independent reactions. The data was used to fit a second-order response surface model. The optimal conditions predicted by the model were validated across a range of mutant KRAS allelic frequencies (0.05%-5%). Ten replicates were performed at each allelic frequency, with wild-type only and water as negative controls. Assay performance was compared using dropletized DNA preamplification (MED-Amp) or without (Bulk Preamp). Multiplex digital PCR: Reactions were performed using either the Bio-Rad QX200 or QX600 digital droplet platform. Reactions were prepared using Probe Supermix (Bio-Rad) with custom probes (TaqMan) and primers (IDT) as previously described⁴. Results were analyzed using a combination of QX Manager and R.


Results, Conclusions, and Discussions:: To establish a robust pipeline for higher-order dPCR multiplexing, we required a quantitative metric to evaluate not only the number of mutant positive droplets, but also the quality of each mutation cluster's signal. Resolution measures the degree of separation between two adjacent signals and integrates critical information on the signal dispersion and intensity (Figure 1C). Resolution is strongly correlated with signal quality and is sensitive to PCR reaction conditions, as shown in an annealing temperature gradient experiment (Figure 1D). While the total number of positive droplets and relative ratios are similar across the gradient range, the uniformity of droplet fluorescence was highest at the optimal temperature of 60°C (p< 0.05). These results suggest resolution is an ideal metric for optimizing multiplexed dPCR reactions.



We then used response surface methodology to build a model elucidating the relationship between key amplification parameters and signal quality (Figure 1E). We first addressed whether optimization could eliminate the need for the first dropletized preamplification step in MED-Amp. We ran experiments using conventional preamplification without dropletization (Bulk Preamp) and used the resulting model to predict ideal preamplification conditions. We then evaluated assay performance using both conventional preamplification and MED-Amp. MED-Amp still outperformed conventional preamplification, particularly at low mutant allelic frequencies (< 1%). MED-Amp also had a broader linear dynamic range (Figure 1E) and increased signal-to-noise ratio (Figure 1F). Our data confirmed that partitioning is essential for uniform and robust amplification of rare alleles.



Our approach addresses key limitations in dPCR assay design. By pairing experimental design with statistical modeling, we can better understand the role of interacting factors in digital PCR and build a framework for rational optimization of new multiplexed assays. Second, we use higher-order multiplexing to increase both assay capacity and resolution. Our preliminary data shows this technique can be used to design a discriminating multiplex reaction for five unique targets using a 2-color dPCR system (Figure 1B). With advanced 6-color dPCR platforms, we have the potential to track up to fifteen unique mutational targets. These advances make MED-Amp a versatile platform that can be easily generalized and integrated into multiple clinical studies




Acknowledgements (Optional): :

References (Optional): : 1 Pietrasz, Clin Can Res (2017). 2 Lin, Cancer Discov. (2019). 3 Bettegowda, Sci Transl Med. (2014). 4 Pratt, Anal Chem. (2019). 5 Pratt, BioRxiv (2021). 6 Elazezy, Comp. Struct. Biotechnol. J. (2018).