Molecular sampling limits of ctDNA detection in clinical plasma samples

Area of Science:

• Molecular Biology
• Genomics
• Clinical Diagnostics

Background:

• Circulating tumor DNA (ctDNA) assays are crucial for cancer monitoring.
• Current detection limits are often based on fixed variant allele fraction (VAF), potentially overestimating sensitivity.
• The finite number of DNA molecules in plasma samples is a critical, often overlooked, factor.

Purpose of the Study:

• To evaluate the impact of plasma sample input volume on the limit of detection for ctDNA.
• To determine the achievable sensitivity of ctDNA assays considering the physical constraints of DNA molecules.
• To propose a more accurate interpretation of ctDNA results based on sample molecular input.

Main Methods:

• Integrated genome-equivalent (GE) distributions from 5238 plasma samples across three cohorts.
• Employed a Poisson-based sampling model to estimate input-limited detectability for single-locus variants.
• Defined the lower limit of detection as the minimum VAF for 95% probability of detecting 1, 3, 5, or 10 mutant molecules.

Main Results:

• Genome-equivalent input varied significantly (median 5531 GE).
• While 1% VAF was detectable in most samples, sensitivity decreased sharply at lower VAFs.
• At 0.1% VAF, only 72% of samples could detect at least one mutant molecule; this dropped to <10% at 0.01% VAF.

Conclusions:

• Many clinical plasma samples have insufficient molecular input for reliable low-VAF ctDNA detection, irrespective of assay design.
• Sample-aware interpretation accounting for molecular input is essential for accurate ctDNA analysis.
• This highlights the need to consider physical DNA quantity alongside VAF for robust liquid biopsy results.