KRAS G12C describes a specific change in the KRAS protein: glycine, abbreviated G, is replaced by cysteine, C, at position 12. It is a biomarker that can matter for cancer treatment. It is not the name of a cancer type, an automatic treatment recommendation, or proof that a cancer was inherited.

The useful question is not simply “Is KRAS positive?” It is: which KRAS variant, in which sample and cancer, measured by which test, and relevant to which treatment setting?

Why the exact letters matter

KRAS is part of a cellular signalling system that helps regulate growth. Certain variants disrupt that regulation. G12C, G12D and G12V are different protein changes; a medicine developed to bind the G12C form cannot automatically be assumed to work against another form. The cysteine substitution helped make a previously difficult target accessible to a particular class of drugs. NCI account of the first sotorasib approval.

Read a report in five steps

  1. Identify the specimen. Was the analysis performed on tumour tissue or circulating DNA in plasma? What was the collection date?
  2. Copy the exact variant. “KRAS mutation detected” is less informative than a clearly specified protein and DNA change with a reference sequence.
  3. Check scope and quality. Did the method include G12C? Was there sufficient tumour material? What sensitivity and limitations were reported?
  4. Connect the finding to the cancer. Histology, stage, previous treatments and other biomarkers affect its clinical significance.
  5. Check the relevant evidence and current local indication. A trial result, a regulatory approval and an individual treatment decision are different things.

Biomarker testing can identify a possible treatment or study, but a match does not guarantee benefit or access. Some results have no established therapeutic implication. NCI: biomarker testing for cancer treatment.

Two schematic plasma samples: one contains detectable tumour-derived DNA and the other contains little or none. A negative result in the second sample may be non-informative; a clinician decides whether tissue or other testing is needed.
Two schematic plasma samples: one contains detectable tumour-derived DNA and the other contains little or none. A negative result in the second sample may be non-informative; a clinician decides whether tissue or other testing is needed.

Open full-size illustration (new tab)

Why a negative blood test may not settle the question

A plasma test depends on tumour-derived DNA reaching the sample in a detectable amount. A negative result may therefore be non-informative rather than proof that the tumour lacks G12C. The Canadian consensus for KRAS G12C-mutated non-small-cell lung cancer recommends considering tissue testing when a negative liquid biopsy has not demonstrated detectable tumour-derived DNA. The oncology team must weigh tissue availability, safety, urgency and the exact assay. Canadian consensus recommendations.

Fictional example: two reports appear to disagree. A patient’s earlier tumour sample contains G12C, but a later plasma report detects no tumour variants. The clinician checks whether the plasma sample had evidence of tumour DNA at all. The negative result is not automatically interpreted as disappearance of the original variant. Disease course, treatments, sampling and assay sensitivity all enter the discussion.

One variant does not mean one treatment across cancers

A useful dated example is the FDA’s January 2025 approval of sotorasib with panitumumab for adults with KRAS G12C-mutated metastatic colorectal cancer after specified prior chemotherapy. The combination, cancer type and prior treatment requirements were part of the indication. This is not equivalent to saying “sotorasib treats every G12C cancer.” Check the current prescribing information and local guidance before applying any historical approval summary. FDA approval notice.

Fictional example: a search result is for a different cancer. Marta finds a promising lung-cancer trial after receiving a colorectal-cancer report. The shared variant makes the article worth discussing, but the study population and treatment combination still matter. Her oncology team checks evidence in colorectal cancer rather than importing the lung-cancer result unchanged.

A practical trial-reading example

In the FDA summary of CodeBreaK 300, median progression-free survival was 5.6 months in the evaluated sotorasib/panitumumab group and 2.0 months with the specified standard-care options; the hazard ratio was 0.48. The study was not powered for overall survival, and its final overall-survival analysis was not statistically significant. These are different endpoints. FDA efficacy assessment.

A hazard ratio is not the percentage of patients cured, and a difference between group medians is not extra time promised to every participant. To interpret a headline, write down the outcome, comparison treatment, uncertainty interval, follow-up and adverse effects. Our risk guide and clinical-trial reading guide explain these distinctions in more detail.

Is KRAS G12C hereditary?

A tumour biomarker result does not by itself establish a germline finding or an inherited cancer syndrome. Most treatment-oriented KRAS G12C findings are evaluated as tumour alterations. If the clinical history raises an inherited-risk question, that deserves the appropriate germline assessment; it should not be inferred from tumour testing alone. NCI explanation of tumour biomarkers and inherited-risk testing.

For a research or clinical discussion

  • Distinguish a research assay or product catalogue from a clinically validated test.
  • Check that a trial actually includes the variant and tumour type of interest.
  • Record prior treatment requirements and the combination being studied.
  • Separate response, progression-free survival, overall survival and quality of life.
  • Ask how resistance, adverse effects and alternative treatments affect the decision.

Use our tumour-versus-germline guide for discordant reports and the paper-comparison workspace to compare studies. This guide explains evidence; treatment selection remains an oncology decision.