An important axis for classifying ctDNA assays is whether they are tumor-informed or tumor-naïve. Tumor-informed assays are tailor-made designs based on the genetic information provided by a tumor biopsy. Tumor-naïve assays are typically fixed panels containing primers or probes targeting highly recurrent mutations present in specific tumor types, for example, breast or colon cancer (Figure 1). Thus, tumor-informed assays can only be used post-diagnosis, and often post-surgery. Additionally, they normally require at least targeted sequencing, and ideally, exome sequencing, of the primary tumor to identify a sufficiently large number of high-quality, somatic, tumor-specific mutations. In return, tumor-informed assays have orders-of-magnitude greater sensitivity than tumor-naive multigene panels, as the latter typically detect only an average of 2 - 5 variants per patient despite using large panels (e.g., >100 genes). Several properties of both types of tests are summarized in the below table.
Although multigene panels provide improved sensitivity relative to a single marker (e.g., a qPCR assay), a tumor-naïve panel may not cover variants found in some patients, especially for cancer types with a highly heterogeneous tumor mutational landscape and few recurrent mutations between tumors. By contrast, prior knowledge from assessment of variants in tumor tissue allows tracking of a greater number of high-quality variants, enhancing sensitivity, whereas tumor-naïve approaches assay many regions unlikely to contain a relevant variant, increasing the chance of false-positive results. False positives are a major clinical concern and a barrier to adoption if a therapeutic intervention is to follow a positive test.
Figure 1. Sensitivity of ctDNA detection methods.
A significant challenge to maintaining the specificity of ctDNA testing is confounding by clonal hematopoiesis of indeterminate potential (CHIP)1. CHIP mutations originate from clonally expanded hematopoietic progenitor cells carrying cancer-associated genetic variants, such as those found in the tumor suppressor gene TP53. Recent studies have reported that 14% of patients with early-stage lung cancer and 25% of patients with late-stage solid tumors harbor CHIP mutations 2. Because of the difference in the methodology used to detect and define CHIP variants between these studies, comparing results does not permit inferences about CHIP mutation frequency by cancer type and stage.However, the high frequency of CHIP variants observed in both studies underlines how misclassifying CHIP variants as ctDNA variants may reduce specificity for MRD detection. Approaches to address this misclassification include sequencing paired peripheral blood cells for in silico filtering of variants common to peripheral blood cells and ctDNA and using tumor-informed methods to identify clonal tumor variants.
| Comparison | Tumor-informed | Tumor-naïve |
| Sensitivity | < 0.001% VAF | > 0.5% VAF |
| Specificity | High | Medium |
| CHIP confounding | Highly unlikely | Suitable |
| Tumor DNA limited, unavailable, or low quality | Not suitable | Suitable |
| Acquired resistance mutations | If present in the tumor or if added to the panel | Cannot be added to the panel |
| Cost | Additional cost for genotyping and personalized design | Plasma cost and single design only |
| Turnaround time | Longer time for initial test due to design and genotyping | Depends on turnaround time for the plasma test only |
References
*1 Marnell, C. S., Bick, A. & Natarajan, P. Clonal hematopoiesis of indeterminate potential (CHIP): Linking somatic mutations, hematopoiesis, chronic inflammation and cardiovascular disease. J Mol Cell Cardiol 161, 98–105 (2021).
*2 Coombs, C. C. et al. Therapy-Related Clonal Hematopoiesis in Patients with Non-hematologic Cancers Is Common and Associated with Adverse Clinical Outcomes. Cell Stem Cell 21, 374-382.e4 (2017).