Problems with early cancer detection

Like many breakthroughs, this discovery begins and the observations are meaningless. In 1948, two French researchers Paul Mandel and Pierre Métais published a well-known paper in the journal Science. They worked in a laboratory in Strasbourg and have been sorting the chemical content of plasma, the river of life, containing proteins, sugars, wastes, wastes, nutrients and cellular debris. In this familiar inventory, they discovered an unexpected existence: fragments of DNA drift freely.

The discovery violates bioorthodox. It is believed that DNA remains locked in the nucleus of the cell rather than floating on its own. Strangers are still not the entire genome, but the broken pieces – the float of genes emits from an unknown source to wander.

Mandel and Métais weren’t sure what to do. The scientific community has been equally confusing for more than a decade, largely overlooking the paper. But biological mysteries are rarely buried. Ultimately, the researchers returned to the question with a simple explanation. Every day, as billions of cells die, they break, spilling their contents, including the blood. These fragments circulate briefly before being metabolized or cleared by the kidneys. The researchers concluded that this “cell-free DNA” is a residue of the ongoing cycle of death and renewal in the body.

DNA seems to be stand out from debris such as debris in sunken blood vessels by dying cells. What seems to be the waste may be witnesses – socks, spoons, necklaces drifting from the flooded compartment, and everyone implies the life they once lived. Do these fragments in our blood bring information in the cells that release them? Can scientists assemble these molecular fragments and reconstruct the identity of the cells they come from?

In the nineteen and sixties, Aaron Bendich, a cancer researcher in New York, suggested that like healthy cells, it might drip DNA into the bloodstream. By 1989, the discovery of Mandel and Métais over the decades, researchers found concrete evidence of tumor-derived cell-free DNA in cancer patients’ blood.

The meaning is profound. For generations, scientists have looked for ways to detect cancer early: mammograms, colonoscopy, pap smears – all aimed at capturing malignant tumors before spreading. The idea that cancer cells may be secretly leaking into the bloodstream hints a fundamental new possibility: We may have detected malignant tumors through simple blood, not through imaging or physical examination. Scientists eventually call it a “liquid biopsy” and for many, it adds to a transformative leap in cancer screening.

Hope for early detection – The promise of capturing cancer before it is announced through symptoms can drive research and investment in the field. However, this hope may obscur a more complex reality.

“If we are going to beat cancer, it can be said that early detection and diagnosis are the most effective means we can use,” a group of cancer research in the UK announced in a 2020 review. this Lancet Oncology. The case of cancer screening assumes the shape of a simple narrative: a lump formed in a woman’s breast; it was detected by mammograms; a biopsy confirmed the malignant tumor; the surgeon deleted it before it spreads. Her life was saved.

But now imagine two women going to a mammogram clinic. Both were found to have the same lump. Both have been diagnosed with early stage breast cancer and are scheduled for surgery. Every time I go home, I feel relaxed and believe that modern medicine has intervened in time. As one woman told me, recalling for a moment: “Once I know it’s inside me, I want to call as soon as possible. I call the surgeon’s office every hour until they make an appointment for me next week.”

The problem is that mammograms reveal only the shadow of the tumor – it cannot sacred the nature of the tumor. It shows the cancer’s body, not its mind: i.e., mammograms cannot tell us whether the tumor is aggressive, whether it has spread or will remain inert. The image has no clues of intention and future tendencies.

Assuming the first woman had surgery, this was reassuring due to the idea of ​​”early” testing, it turns out that cancer has sent metastasized cells out of the scope of the scalpel. The process, although strict, has no benefits. She endured no gains, which was contrary to the old medical ban: First, don’t hurt.

The second woman faces the opposite. Her tumor appears ominous, but essentially, her tumor is lazy-emitting, non-invasive, never destined to threaten her life. However, she also underwent surgery, anesthesia, and recovery. This process eliminates no dangerous tumors. Again: There is no harm in profit.

This paradox reveals a core flaw in our current cancer screening model. We are already good at locating the physical presence of cancer – its body form – but remain largely blind to its traits, behaviors and future. We used genomic assays and histopathological grading, but many early tumors remain biologically ambiguous. They may be early cancers that can be cured by surgery. They grow slowly and are unlikely to cause harm. Or, most of the concerns are that they may have shifted so that local interventions do not. Three possibilities – but we often cannot tell which one we are facing.

More complex, false positives abound: tests that indicate the presence of cancer, leading to unnecessary procedures, anxiety and harm. To begin to get involved in this dangerous terrain, we may turn to a curious figure – the clergy and mathematicians of the Enlightenment, whose minds now guide us through the darkness.

Thomas Bayes is not a doctor. Born in the early 18th century, he was the president of the elders and performed in formal logic, an interpreter of uncertainty in an age of certainty. In a portrait of Bayes traditionally (although the babysitter may have been misidentified), he looks like a massive, confident man with a Wall Street hairstyle: Alec Baldwin, a clergy coat. Bayes published only two papers in his life: one defending God’s kindness and the other defending Newton’s calculus. His lasting contribution was in the Royal Society’s paper on conditional probability. Its arguments still inform the way we evaluate information.

Imagine a group of thousands of smokers from the 1960s. One of them has lung cancer. The Bayesians call it “previous probability”, which is a thousand chances-the possibility of suffering from this disease before we know anything else. Now assume that the test we used correctly detected lung cancer was ninety-nine times. That is the “sensitivity” of testing. It also correctly gives a negative result of ninety-nine percent of the absence of cancer, i.e. the “specificity” of the test.

So, if someone in the group tests positive, what is the chance that the person actually has cancer? Bayesian arithmetic gives a surprising answer: it can be expected that the test can identify a person who actually has cancer, but it will also mistakenly flag ten people without cancer. This means there are about 11 positive results, but only one of them is accurate. Well, the chances of a positive person having cancer are just over 9%. In other words, 11 people will be sent for follow-up procedures, such as biopsy. Ten of them will go through a dangerous and invasive process that may involve piercing the lungs, bleeding or other complications – without any benefit.

In short, if you plan to find a needle in a haystack, you will mostly have hay even with the best detector. Choose a haystack with thousands of needles scattered in a large bag and you will start to discover more needles than hay. The posterior probability (the chance you find the needle) depends on the previous probability (how many needles were there at the beginning).

In Bayesian models, knowledge is always temporary, a process of updating beliefs based on new evidence. In a fifty-eight-year-old breast cancer survivor, the disease has a strong family history and a new mass near the original site may recover – intervention is necessary. In a twenty-year-old with no relevant history, the same finding may be benign – the waiting to watch may be enough.

The consequences of ignoring these principles are shocking. According to an estimate, in 2021, the United States spent $4 billion on cancer screening. On average, a year of screening value produces 9 million positive results, of which 8.8 million are wrong. Millions of continuous scans, biopsies and anxiety disorders, so only two hundred thousand true positives can be found, and even smaller parts can be cured with topical treatments such as resection. The rest is mistakenly considered the noise of the signal, and the harm is mistakenly considered to be helpful.

The problems discovered early on are not over. I sometimes ask the interns a question on the morning round: “How do we tell if a screening cancer test works?” The answer usually comes quickly: “If the test detects malignant tumors at a high rate or early stage.”

But, as the mammogram story shows, it is just that the tumor is found to know nothing about its effect. So I go further. Their next answer also quickly emerged: “By dividing the population into screened and unscreened groups, and then measuring which group grows longer without cancer.” But this approach raises another fallacy.

Suppose two identical twins develop breast cancer simultaneously in 2025. One was screened regularly. Her tumor was very early. She started treatment-surgery, chemotherapy. The process is frustrating: blood clots after surgery, infections during chemotherapy, recovery for months. Four years have passed. She endured everything and hoped to heal.

Her sister was shaken by the torture of an old friend and avoided screening. She moved to upstate New York, leaning towards apple trees, reading books and avoiding medical interventions. By 2029, symptoms of breast cancer appear, but she will reduce treatment.

In 2030, the first sister learned that her cancer had recovered. She was admitted to a hospital in New York City. The same month, her sister (now obviously ill) was admitted to the same facility. They lay on the adjacent bed, reflecting on their choices. They died the same week.

Now it’s a fantasy. The first twin’s survival record was five years after diagnosis. Second, only one year. Doctors who review their cases may conclude that screening has extended survival by five times. But both women were born and died at the same time. Screening has no effect on life span. The obvious benefit is the statistical phantom, which is the time when we start the clock. This is a “delivery time bias” that expands survival time without improving results.

Delivery time bias is not the only fantasy that distorts screening for cancer results. Consider a village where cancer occurs in two forms – one that is fast and deadly and the other is slow and harmless. With annual screening, slow-growing tumors are more likely to be labeled: they last longer in detectable asymptomatic phases. In contrast, aggressive diseases often develop symptoms between screenings and are diagnosed clinically. (Patients with them may even die between annual tests.) Ten years later, the data looks promising: more early cancers were found and longer survival after diagnosis. But the obvious benefit is misleading. Screening disproportionately detects lazy tumors, and first, tumors with smaller tumors may be smaller. This is called the length of time bias.

These twin hallucinations—always biased and prolonged biased—present annoying lights to screening efforts. One expands our measurement of survival by moving the starting line. Other claims to succeed by favoring tumors that are already prone to harm. Together, they misled cancer researchers for decades.