# ctDNA units, fractions and concentrations

Source: https://onco.cc/terms/ctdna-units/  
OnCo record `ctdna-units` (Term). Data CC BY-NC 4.0, attribute "Data from OnCo (onco.cc)"; commercial use needs a licence.

## TL;DR

A ctDNA result is a number with a unit. Molecules per millilitre, parts per million and allele fraction answer different questions, so a printed number means nothing until its unit and denominator are known.

## Summary

Circulating tumour DNA (ctDNA) can be reported as a concentration, as a fraction, or as a yes or no call, and the three are not interchangeable. A concentration counts tumour-attributed molecules in a stated volume of plasma. One example is mean tumour molecules per millilitre (MTM/mL), which the I-SPY 2 breast-cancer study of Magbanua and colleagues used for ctDNA across tracked variants. A fraction divides tumour-attributed molecules by a stated total, such as all evaluated alleles at one position (variant allele fraction, VAF) or all evaluated DNA (an estimated tumour fraction). Parts per million (PPM) is a fraction multiplied by one million, so the number is only as meaningful as the fraction it scales: 0.01% is 100 PPM, and 0.00024% is 2.4 PPM, the lowest allele fraction Santonja and colleagues reported for ctDNA detection in breast cancer.

Concentration and fraction can move apart. If two equal one-millilitre aliquots each hold the same number of tumour molecules but one holds more background cell-free DNA, the concentration is unchanged and the fraction falls. A detection call and a reliable amount are also different claims, because the detection limit and the quantitation range are validated separately. Nothing here gives a conversion between assays: MTM/mL cannot be turned into PPM from the printed result, and a VAF of 100 PPM at one variant is not a tumour fraction of 100 PPM.

## Fields

- Kind: Term
- Last checked: 2026-11-12
- Also known as: MTM/mL; mean tumour molecules per millilitre; ctDNA reporting units; ctDNA PPM
- Tags: biomarker; measurement

## Notes

- Method card, what it measures: the quantity an assay states, either a count of tumour-attributed molecules per volume of plasma (a concentration), a share of a named total (a fraction), or a call. How: sequencing or PCR on one finite plasma aliquot, then a calculation the laboratory defines. Keep the unit, the denominator and the assay version attached to every value.
- Table, MTM/mL: mean tumour molecules per millilitre of plasma under the assay's own calculation across its tracked features. Boundary: a concentration estimated from tracked features, not the number of cancer cells per millilitre and not a share of anything.
- Table, PPM: parts per million, a fraction multiplied by one million, so 1 PPM is 0.0001% and 100 PPM is 0.01%. Boundary: name the fraction being scaled, because 100 PPM of variant alleles, of evaluated DNA and of tumour-attributed molecules are three different claims.
- Table, variant allele fraction: the share of evaluated alleles at one position that carry a stated variant. Boundary: it depends on copy number, on which clones carry the variant and on the measurement, and a variant present in a blood-cell clone counts the same way as one from a tumour.
- Table, tumour fraction: an estimated share of evaluated DNA attributable to tumour. Boundary: it is a model output that rests on the method and its calibration, and it is not automatically any single variant's VAF.
- Fictional arithmetic, concentration against fraction: two equal 1 mL aliquots each hold 4 tumour-attributed molecules. Aliquot A also holds 16 background molecules, so 4 of 20 is 20% and the concentration is 4 molecules per mL. Aliquot B holds 36 background molecules, so 4 of 40 is 10% and the concentration is still 4 per mL. More background lowered the fraction and left the concentration alone. These toy shares are not a VAF, an estimated tumour fraction or MTM/mL.
- Fictional arithmetic, rescaling: a VAF of 0.01% at Variant R equals 100 PPM of variant-supporting alleles. That does not make a tumour fraction of 100 PPM, and a second report of 0.2 MTM/mL is a concentration, so 100 and 0.2 cannot be compared or converted without the methods and calibration of both assays.
- Method limit, detection against quantitation: a qualifying detection call and a reliable numerical amount have separate validation criteria. A detected value below the validated quantitative range does not support an exact fold change, and extra decimal places do not create precision.
- Method limit, finite input: the plasma volume drawn is not the volume or the number of molecules recovered and evaluated. Low molecular counts, changes in background DNA, poor recovery or incompatible calibration all limit comparison across assays, and a printed zero is not a count of zero cancer cells anywhere in the body.
- Evidence read: in Magbanua's 84-patient pilot of neoadjuvant treatment, ctDNA in MTM/mL was correlated with MRI functional tumour volume at every time point (Spearman rho 0.42 to 0.49). Santonja's comparison reported allele fractions down to 2.4 PPM for hybrid-capture sequencing of 1,347 to 7,491 tumour-identified mutations and 4.7 PPM for multiplex PCR of 21 to 47 structural variants. Each figure belongs to its assay and cohort.
- A trend also needs the collection time, treatment context, assay version and quantitative range. In its November 2024 drug-development guidance, FDA asks for precision, accuracy and detection-limit evaluation across relevant inputs and cut-offs, using representative clinical specimens and appropriately comparable reference material. A printed number below another assay's nominal detection limit does not establish that the other assay would miss that sample: the measured quantity, specimen, input, recovery and calling rules may differ. A reference-material result is not automatically a clinical sensitivity estimate.
- Consider a fictional assay estimating tumour fraction at 6 parts per million (PPM) and then 2 PPM. Both samples are called detected, but the second is below the assay's reliable quantitation range. The printed estimates fall by two-thirds; they do not establish that the cancer volume fell by two-thirds. Retain the qualifying calls and uncertainty, because shedding and background DNA can change the signal. A second method reporting molecules per millilitre cannot simply be placed on this fraction scale. Even a matching unit label does not establish matching calibration. Keeping serial methods and input comparable helps interpretation without removing biological variation.
- The blank, detection and quantitation limits answer different questions. The blank limit is the highest apparent analyte concentration expected from replicate samples with no analyte; the detection limit is the lowest concentration reliably distinguished from that background; the quantitation limit is the lowest concentration meeting predefined bias and imprecision goals (Armbruster and Pry, 2008). Quantitation and detection limits may coincide or differ. A qualifying detection below a validated quantitative range supports presence more strongly than an exact fold change. For a read-based variant allele fraction, the denominator is informative reads covering that locus, not cancer cells or plasma volume; keep a molecule-based method's own definition instead.

## Sources

- Wikipedia: https://en.wikipedia.org/wiki/Circulating_tumor_DNA
- Magbanua et al., npj Breast Cancer 2021: ctDNA in mean tumour molecules per mL plasma against MRI functional tumour volume: https://doi.org/10.1038/s41523-021-00239-3
- Santonja et al., EMBO Mol Med 2023: tumour-informed and tumour-naive ctDNA assays compared, allele fractions to 2.4 parts per million: https://doi.org/10.15252/emmm.202216505
- FDA, Use of Circulating Tumor DNA for Curative-Intent Solid Tumor Drug Development, November 2024: https://www.fda.gov/media/183874/download
- Armbruster and Pry, 2008: limit of blank, detection and quantitation: https://pubmed.ncbi.nlm.nih.gov/18852857/

## Connected records

- terms: [Cell-free DNA (cfDNA)](https://onco.cc/terms/cfdna/), [Circulating tumour DNA (ctDNA)](https://onco.cc/terms/ctdna/), [Minimal / molecular residual disease (MRD)](https://onco.cc/terms/mrd/), [Variant allele frequency (VAF)](https://onco.cc/terms/vaf/)
- biomarkers: [Circulating tumour DNA fraction (and what a negative plasma result means)](https://onco.cc/biomarkers/ctdna-tumour-fraction/)
- technologies: [Liquid biopsy (ctDNA)](https://onco.cc/technologies/liquid-biopsy/), [MRD / molecular residual disease testing](https://onco.cc/technologies/mrd-testing/)
- key papers: [Comparison of tumor-informed and tumor-naïve sequencing assays for ctDNA detection in breast cancer](https://onco.cc/key-papers/paper-santonja-tumour-informed-vs-naive-breast-embo-mol-med-2023/)

---
JSON: https://onco.cc/api/v1/entities/ctdna-units.json