What a Blood Test for Colorectal Cancer Can and Cannot Tell You: The Methylated-SEPT9 Assay

Tell a patient there's a blood test for colorectal cancer, and watch their relief. No prep, no scope, no missed day of work — just a blood draw. The phrase "blood test for cancer" carries an implied promise of reassurance that the methylated-SEPT9 (mSEPT9) plasma assay cannot actually deliver, and the gap between that expectation and the diagnostic-accuracy data is the whole story worth telling here. This article is about what the test measures, what the numbers show, and — just as important — what a negative result does not establish.

The Diagnostic Question

Colorectal cancer sheds molecular signatures into the bloodstream. The clinical question the mSEPT9 assay tries to answer is narrow and specific: in a person's plasma, can we detect a tumor-associated DNA methylation signal that reliably flags the presence of colorectal cancer? The surrounding hope — that such a signal might also catch the precancerous lesions we most want to remove before they ever become malignant — is a separate question, and the evidence treats it very differently.

The Analyte and the Assay

The biology rests on epigenetics rather than mutation. The SEPT9 gene encodes a septin protein involved in cell division. In colorectal tumor tissue, the promoter region of SEPT9 frequently becomes aberrantly methylated — cytosine residues within CpG dinucleotides acquire methyl groups in a pattern that differs from normal colonic mucosa. Tumor cells release fragments of this abnormally methylated DNA into the circulation as cell-free DNA (cfDNA), where it can be detected in a plasma sample.

Detecting methylation requires a chemical trick. The plasma cfDNA is treated with sodium bisulfite, which converts unmethylated cytosines to uracil while leaving methylated cytosines intact. That conversion turns an epigenetic mark into a sequence difference that a methylation-specific polymerase chain reaction can amplify and read. A positive result means the assay found methylated-SEPT9 cfDNA above its detection threshold; a negative result means it did not.

Here is the crucial interpretive point. The assay reports the presence or absence of a methylation signal — it does not localize a lesion, size it, or distinguish a cancer from any other source of the same signal. It is a flag, not a map.

The Evidence

The anchor dataset is the PRESEPT study, reported by Church and colleagues in Gut in 2014 [1]. This was a prospective evaluation of the mSEPT9 plasma assay, and its value lies in how it frames performance against the reference standard of colonoscopy.

For detecting colorectal cancer, the assay showed a sensitivity of 48.2% and a specificity of 91.5% [1].

Sit with those two numbers for a moment, because they pull in opposite directions. A specificity above 90% is respectable — most people without colorectal cancer will correctly test negative. But a sensitivity of roughly 48% means the assay misses more than half of the colorectal cancers that are actually present [1]. If you are using a test to find a disease, catching fewer than one in two cases is a serious limitation, and it has to govern how any negative result is read.

The figure that matters most to anyone thinking about prevention, though, is the advanced-adenoma sensitivity: 11.2% [1].

What strikes me about that number is how decisively it answers a question people do not always realize they are asking. The entire rationale for colorectal screening is to find and remove advanced adenomas — the precancerous lesions — before they progress to invasive cancer. An assay that detects roughly one in nine of those lesions is, in practical terms, nearly blind to the very targets that make early intervention worthwhile. Reading these numbers honestly, the mSEPT9 assay is a cancer-detection test with modest sensitivity, not an advanced-adenoma detection test in any meaningful sense. Colonoscopy, by contrast, both visualizes those lesions and removes them in the same procedure — detection and intervention collapsed into one act — which is precisely what the methylation signal cannot do.

Where This Leaves Us

So how should a thoughtful clinician hold this test? I think the honest framing is that mSEPT9 is a real biomarker with genuine, measured accuracy, and that its accuracy is not good enough to make a negative result reassuring. That distinction does most of the clinical work.

When a patient asks whether a negative blood test means they are in the clear, the PRESEPT data say plainly that it does not. With a cancer sensitivity near 48% [1], a negative mSEPT9 result leaves open a substantial possibility that a cancer is present and simply was not detected — and with advanced-adenoma sensitivity at 11.2% [1], it tells you almost nothing about precancerous disease. A false sense of security is itself a clinical harm, because it can delay or displace a more sensitive evaluation. This is the piece of the conversation that matters most, and it is where the "blood test for cancer" shorthand does the most damage if left uncorrected.

The specificity figure deserves a fairer hearing. At 91.5% [1], the assay is reasonably good at not raising false alarms, which means a positive result carries more interpretive weight than a negative one — it is a signal that warrants direct colonic evaluation rather than something to dismiss. But even here, specificity below 100% means some positive results will not correspond to colorectal cancer, and the assay itself cannot tell you which.

Several things remain genuinely open, and it is worth being precise about them. The PRESEPT metrics describe performance in a particular study population against colonoscopy as the reference standard [1]; how an individual assay version performs in a different clinical context, or how these numbers compare against other blood-, stool-, or tissue-based approaches, is not something a single study can settle. I have deliberately kept this discussion to diagnostic accuracy, because questions of population screening policy, program-level benefit, and cost involve considerations well beyond the sensitivity and specificity of one test.

If there is one line I would want a trainee to carry out of this, it is this: the number that should reshape expectations is not the 48% cancer sensitivity but the 11.2% advanced-adenoma sensitivity [1]. A test that cannot see the lesions you most want to remove is, by design, a late detector rather than a preventer — and a negative result from such a test should quiet no one's concern on its own.

References

1. Church TR, et al. Prospective evaluation of methylated SEPT9 in plasma for detection of asymptomatic colorectal cancer (PRESEPT). Gut. 2014. PMID 23408352.