Early Cancer Detection: What Works and What Is Oversold

A woman of fifty-two comes in for a preventive assessment. She has had a whole-body scan at a private clinic, two rounds of an expensive blood panel marketed on early cancer detection, and a supplement regimen she researched carefully. She also mentions, almost in passing, that she has been meaning to arrange a colonoscopy for about six years, and that her mother was diagnosed with colorectal cancer at sixty-one. The single test most likely to change her life expectancy is the one she has not done.

This is the shape of the problem. Early detection of cancer is not one subject but three, and they are constantly confused with one another: a small number of screening programmes with randomised evidence that they reduce deaths, a set of genuinely promising technologies whose outcome data are still maturing, and a much larger commercial market selling reassurance. The distinctions are not subtle, and they matter more than any individual test.

Stage at diagnosis dominates almost everything

For most solid tumours, the strongest determinant of outcome is not which drug the patient eventually receives. It is whether the disease was still confined to its organ of origin when it was found. Breast and colorectal cancers detected while localised carry five-year survival above ninety percent. Once distant metastases are present, the same diseases fall to a small fraction of that. Lung cancer shows the same gradient, from a majority surviving five years when caught as a small localised nodule to a small minority once it has spread.

Those figures come with a caveat worth stating up front, because it is the caveat that separates honest screening from marketing: survival measured from the moment of diagnosis is a biased yardstick when you change the moment of diagnosis. We will come back to that. The underlying biology is not in doubt, though. Cancers that have not yet seeded elsewhere can often be removed. Cancers that have, usually cannot.

The screening that already works — and that people skip

A handful of screening programmes have been tested in randomised trials and shown to reduce deaths from the cancer they target. They are unglamorous, they are mostly decades old, and in every health system a large fraction of the eligible population is not up to date with them.

Colorectal screening is the most compelling case in medicine for screening of any kind, because colonoscopy is not only diagnostic but preventive: removing adenomatous polyps takes away the lesions from which most colorectal cancers arise. Faecal immunochemical testing (FIT) is a legitimate alternative — non-invasive, done annually, and effective at a population level, provided a positive result is followed by colonoscopy without delay. Mammography reduces breast cancer mortality; the argument among expert bodies is about starting age and interval, not about whether it works. Cervical screening has moved from cytology to HPV testing as the primary test, which detects the causative infection rather than the cellular changes it has already produced, and allows longer intervals with better sensitivity. Low-dose CT of the chest reduces lung cancer deaths in people with a substantial smoking history — and uptake among precisely those people remains the worst of any programme on this list.

The most valuable cancer test available to most adults is not a new one. It is the colonoscopy, the mammogram or the cervical screen they are already entitled to, already overdue for, and have not booked.

Infection is the most preventable cause of cancer

A meaningful share of the global cancer burden is caused by treatable or preventable infections, and this is the part of oncology where prevention is closest to arithmetic. Chronic hepatitis B and C drive hepatocellular carcinoma, and both are detected by an inexpensive blood test that many adults have never had. Hepatitis C is curable with a short course of oral therapy; hepatitis B is controllable, and those with chronic infection or established cirrhosis warrant structured liver surveillance rather than ad hoc scanning.

Helicobacter pylori is the principal cause of gastric cancer, still one of the leading cancer killers worldwide and disproportionately common in people originating from East Asia, Eastern Europe, the Middle East and Latin America. Testing is simple — breath, stool or biopsy — and eradication reduces the subsequent incidence of gastric cancer, with the largest benefit in high-incidence populations and in people with an affected first-degree relative. HPV vaccination and HPV-based cervical screening address the same logic from two directions, and HPV is also responsible for a rising share of oropharyngeal and anal cancers.

A real family history changes when you start and how you look

Screening guidelines are written for people at average risk. A genuine family history moves you out of that category, and it changes two things at once: the age at which screening begins and the modality used. A first-degree relative with colorectal cancer typically brings colonoscopy forward to age forty, or to ten years before the age at which that relative was diagnosed, whichever comes first — and the interval shortens. A strong family history of breast cancer moves mammography earlier and often adds MRI, which performs far better than mammography in dense breast tissue and in younger women.

A subset of these families carry a recognised hereditary syndrome. BRCA1 and BRCA2 variants confer high lifetime risks of breast and ovarian cancer, and also raise prostate and pancreatic risk. Lynch syndrome, caused by variants in the DNA mismatch repair genes, drives colorectal and endometrial cancer at young ages and calls for colonoscopy beginning in the twenties and repeated every one to two years. These are not situations for a generic screening schedule. They are managed with dedicated surveillance protocols, formal genetic counselling, and consideration of risk-reducing surgery — decisions that belong in a proper consultation, not on a results portal.

Certain patterns in a family should prompt a formal risk assessment rather than reassurance:

• Cancer diagnosed unusually young — breast or colorectal disease before fifty, for instance
• Several relatives on the same side of the family with related cancers
• Breast and ovarian cancer in the same family, or in the same person
• Male breast cancer, or ovarian cancer at any age
• Ashkenazi Jewish ancestry, in which specific founder BRCA variants are markedly more common
• A relative already known to carry a pathogenic variant

Multi-cancer early detection blood tests

The most interesting development in this field is the multi-cancer early detection test: a blood sample analysed for cell-free tumour DNA, typically using methylation patterns, which both flags the presence of cancer and predicts the likely tissue of origin. The concept is sound and the engineering is impressive. A single tube of blood that could triage across dozens of cancers, most of which have no screening programme at all, would be a genuine advance.

The honest limitations are these. Specificity is high, so false positives are uncommon — but sensitivity depends heavily on stage, and it is weakest exactly where early detection would matter most. Stage I disease sheds little DNA into the circulation, and detection rates for the earliest-stage cancers remain modest. A negative result therefore does not mean you do not have cancer, and it does not replace a single item in the table above. Nor has it yet been shown, in the way that matters, that using these tests reduces cancer mortality; large randomised trials, including one embedded in the UK health service, are under way, and mortality results are not yet available. These tests are reasonable to discuss as an addition for informed individuals who understand what a positive and a negative actually mean. They are not a substitute for anything.

Whole-body MRI: real capability, real costs

Whole-body MRI uses no ionising radiation and can identify serious disease before it declares itself — renal, ovarian, pancreatic and hepatic tumours among them, several of which have no screening pathway of any kind. In the right person, that is not a trivial argument.

The costs are not financial. Across published series of asymptomatic adults, the yield of a previously unsuspected, clinically important cancer sits in the low single digits of a percent, while the proportion of scans generating an incidental finding that requires some form of follow-up is very much higher. Most of those findings are benign — cysts, haemangiomas, small adrenal and thyroid nodules — but they are not free. Each one buys further imaging, sometimes a biopsy with its own complication rate, an interval of real anxiety, and occasionally a procedure the person would have been better off without. It is also important to know what whole-body MRI does not do: it does not reliably detect early lung cancer, for which CT is the appropriate test; it does not detect the microcalcifications that mammography finds; and it does not see the flat mucosal lesions that colonoscopy identifies and removes. It is an adjunct for selected, properly informed people. It is not a general-population screening test, and any programme that presents it as one is overselling it.

Overdiagnosis and lead-time bias, plainly

Two ideas explain most of the confusion in this field, and neither is complicated.

Lead-time bias. Suppose a cancer would have become symptomatic at sixty-five and caused death at sixty-eight. Find it on a scan at sixty-two, change nothing about the course of the disease, and five-year survival goes from zero to one hundred percent. The patient has not gained a day. Survival measured from diagnosis has simply been given a longer run-up. This is why survival statistics are almost useless for judging a screening test, and why the only endpoints that count are mortality and, ideally, all-cause mortality in a randomised comparison.

Overdiagnosis. Some cancers detected by screening would never have caused symptoms or shortened life. They grow too slowly, or not at all, or the person dies of something else first. These are real cancers under the microscope, but they are not real problems — and once found they are almost always treated, because we cannot yet reliably tell which is which. The clearest demonstration came from South Korea, where intensive ultrasound screening produced an enormous rise in thyroid cancer diagnoses and thousands of thyroid operations, with essentially no change in thyroid cancer mortality. Similar dynamics operate, to varying degrees, in prostate and breast screening, which is why PSA testing is properly a shared decision rather than a routine.

This is also why screening for some cancers is not recommended despite their severity. A large UK randomised trial of annual CA-125 and ultrasound in postmenopausal women shifted the stage at which ovarian cancer was found but did not produce a significant reduction in ovarian cancer deaths. Wanting a test to work is not evidence that it does.

What a rational plan looks like

A sensible personal screening plan is boring, specific and, above all, executed. It has four layers, in strict order of priority:

• Complete the guideline screening you are due for. Not intend to, not consider — complete it, and know the date of the next one. This single step outweighs everything else on this page.
• Get your family history documented properly. Which relative, which cancer, at what age, on which side. Then have it assessed, and pursue genetic testing where the pattern warrants it.
• Test once for the treatable infectious contributors — hepatitis B and C, and H. pylori where your symptoms, family history or country of origin make it relevant.
• Consider the emerging tools deliberately, not reflexively. Multi-cancer blood tests and whole-body MRI have a place for informed individuals who understand the false-positive burden and accept it. They sit on top of the first three layers, never instead of them.

The questions worth asking your physician are correspondingly plain. Which screening am I actually due for, and when was the last one? Does my family history change the age I should start or the test I should have? Is there anything in my background — ancestry, chronic infection, occupational exposure, prior radiotherapy — that puts me outside the average-risk pathway? And if you are being offered a new test: what happens if it is positive, what happens if it is negative, and has it been shown to reduce deaths, or only to find things?

Early detection genuinely works. It works through a small number of well-validated programmes, applied to the right people, and finished rather than postponed. The most common failure in cancer screening is not choosing the wrong test. It is not doing the right one.