Article — "Screening Beyond the Standard Playbook: Guidelines That Miss At-Risk Patients": Article — "Screening Beyond the Standard Playbook: Guidelines That Miss At-Risk Patients": At least 9% of women meet the criteria for adding breast MRI to their standard screening. The actual utilization rate is 0.4%.
Per Attia, fewer than one in twenty women who should be getting the imaging their guidelines recommend are actually getting it. That is not a scientific gap or a technology gap. Both exist and both work. It is a routing gap.
Start with the counterintuitive scoreboard. We have never had more or better cancer-detection tools — MRI, PSA density, PSA velocity, low-dose CT, transperineal biopsy, spatial genomics. And yet advanced and metastatic prostate cancer diagnoses are showing up at higher rates today than they were fifteen years ago.
The tools work. So the thing failing is the rulebook that decides who gets pointed at them, and when.
Consider prostate cancer first. Per Attia, early detection reportedly carries a 15-year survival rate approaching 97% from first diagnosis. About one in eight men will be diagnosed; roughly 36,000 American men will die this year.
The math should be triumphant. It isn't, because PSA is read as a single value against a threshold rather than tracked as a trajectory over time.
The math should be triumphant. It isn't, because PSA is read as a single value against a threshold rather than tracked as a trajectory over time.
A trajectory tells you something a snapshot never can. A PSA of 2.5 is unremarkable in isolation and alarming if it was 1.0 two years ago. The single-number habit throws away the derivative and keeps only the level.
Then there is finasteride, taken by millions of men for hair loss. It suppresses PSA. If a clinician doesn't account for that suppression, a man can sit on a rising true signal that reads as reassuringly flat — a ticking clock nobody is watching.
This is the same measurement failure that dogs any population-normal reference range: the number looks fine, the individual is already in trouble. It is the logic the CPET people apply to resting labs. We will leave that argument to them and stay on screening.
Now the second contested group: high-risk women who are told to screen but never matched to the right imaging.
The guidelines are not ambiguous here. For high-risk women, MRI is supposed to start early — often by 25 or 30 — because the benefits of early detection outweigh the false-positive and overdiagnosis harms. At least 9% of women meet the criteria for MRI as part of standard screening.
The utilization is 0.4%.
And the failure is worst precisely where the stakes are highest. In one analysis of over 6,000 women at a Detroit cancer center, reportedly 97.7% of high-risk patients were incongruently screened — meaning they lacked the supplemental imaging they qualified for. Among average-risk patients, incongruent screening ran at 0.9%.
Read those two numbers together. The women with the most to gain from MRI were the ones almost never getting it, while low-risk women were sometimes getting supplemental imaging they didn't need. The system is inverted.
This is not the guideline being wrong. Several guidelines explicitly stratify by risk. It is an implementation failure — knowing how to identify high-risk women and then failing to route them.
The third group is the one written out of the criteria entirely: never-smokers.
Roughly 10 to 20% of lung cancer — about one in seven cases — occurs in people who never voluntarily inhaled. Per the reported data, lung cancer in never-smokers is the fifth-leading cause of cancer death worldwide. Among Asian-American women with lung cancer, an astonishing 57.4% never smoked; among Hispanic-American women, 32.6%.
The USPSTF recommends annual low-dose CT for adults 50 to 80 with a 20-pack-year history who currently smoke or quit within 15 years. Never-smokers are not recommended for screening.
Here is the crucial distinction, and it is easy to miss. The Task Force did not examine trials of never-smokers and conclude screening wouldn't help them. When it analyzed the evidence, it included trials that enrolled never-smokers — but it only evaluated the evidence in smokers. It also excluded many never-smoker studies from East Asia by its selection criteria.
So never-smokers weren't judged unfit for screening. They were never judged at all.
That reflects a mandate — improving health at the population level, which pushes attention toward the highest-risk group. It is a defensible choice for a population body. It is a terrible default for the never-smoking Asian-American woman sitting in an exam room, whose personal risk profile the guideline simply never modeled.
This is the load-bearing distinction of the whole piece. The guideline's job is population defensibility: minimize net harm across millions, survive scrutiny, inform insurance coverage. The patient's job is different — not dying of the specific cancer they specifically carry.
Those two jobs are not the same job, and the structural mismatch systematically abandons whoever falls between the population's high-risk bucket and its average-risk default.
The good news is that risk-based frameworks already beat the trigger-based ones on their own turf. In a 2026 comparison, risk-based strategies reportedly outperformed USPSTF age-and-pack-year criteria for lung cancer screening. A real-world study likewise found current lung screening guidelines miss a high-risk population the criteria weren't built to see.
The pattern generalizes beyond cancer. Per the study, analytics-driven individualized diabetes screening, modeled on more than 60,000 patients, added a quality-adjusted life year per patient at 35% lower cost than existing guidelines. Glaucoma screening tells the same story from the other side: per the reported analysis, the USPSTF calls the evidence insufficient, while 50% of people with glaucoma remain undiagnosed and targeted high-risk screening reads as highly sensitive.
None of this requires new science. It requires matching the right person to the right test — the thing the guidelines already know how to specify and routinely fail to deliver.
Which brings us to the frontier, and to the part where the confidence stops.
The tumor microenvironment — the immune cells, fibroblasts, and blood vessels surrounding a tumor — strongly shapes whether a cancer spreads and whether it responds to therapy. Today the only clinical window into it is a biopsy, and treatment decisions aren't even based on the TME; they lean on biomarkers like tumor mutation burden and PD-L1.
Aaron Newman's group at Stanford built something remarkable: reportedly, from 10 million single cells across 10 cancer types, they defined 9 "spatial ecotypes" — neighborhoods within the tumor — then trained a blood test, the Liquid Ecotyper, to read them from cell-free DNA.
The neighborhoods carry real prognostic weight. Per the reported data, SE4 describes an immune system asleep in a hypoxic tumor.
Because it's a blood test, it can be run serially — repeatedly, over time — instead of the single biopsy snapshot that can't track a dynamic environment. In other words, it does for the TME exactly what a PSA trajectory does for prostate signal: it turns a snapshot into a trend.
Now the arm's-length part. The Liquid Ecotyper was reportedly validated in 78 pre-treatment melanoma patients from Yale. That is the whole clinical foundation.
There is, in the words of the researchers describing the field, "still no way of assessing it in the clinic by a blood test," and this test "isn't used for clinical decision making." Work in lung, bladder, and mesothelioma is ongoing but unpublished. A company, Liquid Cell Dx, exists; approvals do not.
So hold it honestly. It is genuinely exciting and categorically not ready — a personalization frontier, not a tool you can ask for.
The point of the three worked cases is that we don't need the frontier to close most of the gap. The man whose PSA is rising under finasteride, the woman who meets MRI criteria at 9% and gets it at 0.4%, the never-smoker the criteria never modeled — none of them are waiting on a Nature paper.
They are waiting on a rulebook that remembers its own job is not their job.