Droplet digital PCR outperforms qPCR by revealing clinically relevant low-frequency KRAS mutations in colorectal cancer

Background : KRAS (Kirsten rat sarcoma viral oncogene homolog) mutation testing is essential for therapeutic decision-making in colorectal cancer (CRC), as the presence of any activating KRAS variant predicts resistance to anti-epidermal growth factor receptor (anti-EGFR) therapy. Routine real-time PCR (qPCR) assays widely used in diagnostic laboratories typically exhibit limits of detection (LOD) around 1–5% variant allele frequency (VAF), making them prone to missing low-frequency or subclonal mutations. Droplet digital PCR (ddPCR) offers absolute quantification and significantly higher analytical sensitivity, enabling detection of variants below 1% VAF. Aim : To compare the analytical performance of ddPCR with qPCR for detecting KRAS hotspot mutations in FFPECRC tissue and determine whether ddPCR identifies additional clinically relevant low-frequency variants missed by qPCR. Methods : Fifty FFPECRC samples were genotyped using routine qPCR (easyPGX KRAS ) and subsequently re-tested using ddPCR (Droplex KRAS Mutation Detection Kit, Gencurix) on the Bio-Rad QX200 platform. Hotspot mutations across exons 2, 3, and 4 were assessed. ddPCR outputs (mutant/wild-type droplet counts and VAF) were analyzed using QuantaSoft Analysis Pro. Concordance, sensitivity, specificity, predictive values, Cohen’s κ, VAF distributions, and ROC/AUC analyses were performed. Discordant cases were evaluated for VAF relative to qPCRLOD. Results : qPCR detected KRAS mutations in 25/50 samples (50.0%), whereas ddPCR detected mutations in 31/50 samples (62.0%). ddPCR confirmed all 25 qPCR-positive cases and identified six additional mutations in qPCR wild-type samples. These ddPCR-only events exhibited VAF values between 0.9 and 2.3% (median 1.55%), consistently below the empirical qPCR detection threshold. Concordant qPCR+/ddPCR+ cases showed substantially higher VAF (18–55%, median 41.5%). ddPCR demonstrated 100% sensitivity, 76% specificity, 80.6% PPV, 100% NPV, and 88% overall accuracy versus qPCR, with Cohen’s κ = 0.76, indicating substantial agreement. ROC analysis based on ddPCR-derived VAF yielded AUC = 1.00, with an optimal discriminative threshold of ~18% VAF for qPCR positivity. Detected mutations reflected typical CRC epidemiology, with a predominance of exon 2 variants (G12x, G13D) and occasional exon 4 mutations (A146x). ddPCR also identified multiple subclonal co-existing mutations in several cases, not detectable by qPCR. Conclusion : ddPCR demonstrates superior analytical sensitivity for KRAS mutation detection in FFPECRC tissue and reliably identifies clinically meaningful low-frequency mutations missed by qPCR. Because any detectable KRAS mutation predicts lack of response to anti-EGFR therapy, failure to detect low-VAF subclones may adversely impact therapeutic selection. These results support incorporating ddPCR as a complementary reflex assay within RAS testing workflows, particularly for samples with low tumor purity, borderline amplification, or suspected intratumoral heterogeneity.