Why We Can't "Cure" Cancer

Yes, we can't cure cancer. Not now,

probably not in your lifetime, and the

reason has nothing to do with money or

willpower. The pitch you've heard for 50

years, "Cancer is one breakthrough away.

AI plus CRISPR plus mRNA equals a finish

line." runs on a category error. Cancer

isn't a disease you cure. It is

evolution by natural selection running

inside your body on fast forward. Every

treatment that kills 99% of the cancer

breeds the 1% that survive. In 1971,

Richard Nixon signed the National Cancer

Act and called it a moonshot with

advocates promising meaningful progress

by the American Bicentennial. 55 years

and roughly 200 billion dollars of

public funding later, pancreatic

cancer's 5-year survival rate sits at

13%. The reason isn't that we're behind.

It's that we've been fighting the wrong

shape of the problem. Start with the

word itself, cancer, singular, as if

naming it makes it one thing. It isn't.

The word covers something like 200

distinct diseases that share exactly one

trait, cells multiplying when they

shouldn't. That is the entire common

feature. Beyond that, the cancer in a

smoker's lung and the cancer in a

child's bone marrow have less in common

biologically than a hummingbird and a

shark. Bert Vogelstein at Johns Hopkins

spent the 1990s mapping what actually

goes wrong inside tumors. His lab found

that even a single colorectal tumor, one

tumor in one person, typically carries

between 30 and 70 distinct mutations.

Each tumor is genetically its own

snowflake. Two patients with what the

pathologist labels as stage three colon

cancer are often carrying diseases that

share a name and almost nothing else.

Take breast cancer, the example everyone

knows. The pathologist's label hides at

least four molecularly different

illnesses, luminal A, luminal B, HER2

enriched, and triple negative. Each one

responds to different drugs, has

different prognosis, and is in any

meaningful sense a different disease.

The chemotherapy that saves a HER2

positive patient does very little for

triple negative. Same word on the chart,

different biology underneath. This is

why a treatment that works miraculously

for one woman with breast cancer does

nothing for the woman in the next room.

Not because the medicine failed her,

because she did not have the same

disease. And this isn't a problem clever

oncology can solve. It is a

categorization mistake baked in

centuries ago. We named cancers by where

they started, lung and breast and

pancreas, because that's all the doctors

of the 1800s could see. Modern molecular

biology shows that a lung tumor with a

particular mutation profile responds to

drugs that work on melanoma, while two

lung cancers in adjacent hospital beds

may need entirely different

chemotherapies. The taxonomy was always

a fiction. We just didn't have

microscopes good enough to see through

it. So, the first wall is linguistic

before it is biological. We say cure

cancer the same way someone might say

cure infection. There is no cure for

infection. There are cures for specific

infections, sometimes, when we're lucky

and the pathogen cooperates. Cancer is

the same kind of word hiding the same

kind of plurality. Now, the deeper trap.

Peter Nowell, a pathologist at the

University of Pennsylvania, published a

paper in Science in 1976

with a title most people have never

heard, but that quietly rewrote cancer

biology, The Clonal Evolution of Tumor

Cell Populations. Nowell proposed

something that sounds obvious in

retrospect and was heretical at the

time. A tumor isn't a clump of identical

bad cells. It is a population, a small

ecosystem, and like any ecosystem, it

evolves. Here's the mechanism, step by

step. A tumor starts when one cell picks

up the wrong combination of mutations

and begins dividing without permission.

As that cell divides, daughter cells

accumulate new mutations of their own.

Some daughters carry changes that make

them grow faster and they outcompete

their cousins. Some can slip past immune

surveillance and those win, too. Within

a few months, what looks like a single

clump under the microscope is actually a

forest of slightly different cells, each

holding a slightly different genetic

hand. Now you walk in with chemotherapy.

Chemotherapy is, mechanically, a

selection pressure. A drug that kills

99% of cancer cells leaves the 1% that

happen to carry mutations conferring

resistance. Those survivors aren't

damaged, they're selected. They divide.

Within months, you have a tumor

naturally optimized to ignore the drug

you just used. This isn't a failure of

medicine. It is natural selection doing

exactly what it always does, just inside

a human on the time scale of a single TV

season. Charles Darwin would have

recognized this immediately. Cancer

cells are organisms competing in an

environment, your body, reproducing with

variation, and being selected for traits

that help them survive. The fact that

this evolution is killing the host is

just bad luck for the host. The cancer

doesn't care. It can't care. Evolution

does not have a plan beyond the next

generation. Carlo Maley at Arizona State

University has spent his career arguing

that the dream of curing cancer is

incoherent for this exact reason. You

can slow evolution and sometimes

redirect it. You can occasionally corner

a tumor into a genetic dead end. What

you can't do is cure evolution itself.

Because the same process that produced

humans over 4 billion years is what's

running inside the tumor. Asking

medicine to permanently outsmart natural

selection is asking it to win a chess

game where the opponent gets a new piece

every move. And the move is fast. In a

metastatic tumor, hundreds of millions

of cells divide each day. Each division

is a potential roll of the dice toward

resistance. By the time a doctor sees an

MRI showing the cancer has come back,

the genome of those cells has often

shifted in ways the original biopsy

could not predict. The tumor that

returns is not the tumor that was

treated. It's the tumor's grandchild

with grievances. Reason three sits one

layer deeper where most popular cancer

coverage refuses to go. Cancer isn't an

invader. It isn't a virus you can catch

or a bacterium you can sterilize. Cancer

is your own cells doing what cells do,

going slightly wrong. And the going

slightly wrong isn't a malfunction. It

is statistically guaranteed. Here's the

math. Your body contains roughly 37

trillion cells, an estimate worked out

by the Italian biophysicist Eva Bianconi

in 2013. Most of those cells divide

regularly. Your gut lining replaces

itself every few days. Your skin every

month. Your blood cells constantly.

Every division copies about 3 billion

DNA base pairs, and the copying

machinery is staggeringly accurate with

an error rate somewhere around one

mistake per billion bases. Sounds great

until you multiply 3 billion bases times

trillions of divisions times decades of

life. The number of mutations your body

accumulates by age 60 runs into the

quadrillions. Most do nothing. A few hit

important genes. Eventually, the wrong

combination lands in the wrong cell at

the wrong time, and that cell becomes a

cancer. This is why cancer rates climb

almost exponentially with age. It isn't

bad luck so much as dice rolls finally

hitting the wrong number after enough

rolls, which raises a strange puzzle. If

cancer is a numbers game, large animals

should be drowning in it. A blue whale

has roughly a thousand times the cells

of a human. So, by the math, a whale

should carry a thousand times the cancer

risk. They don't. Whales and elephants

get cancer at roughly the same rate we

do, sometimes even less. This

contradiction has a name, Peto's

paradox, after the Oxford epidemiologist

Richard Peto, who pointed out in 1977.

The answer, worked out over the last 20

years, is that big, long-lived animals

had to evolve better tumor suppression

or they could not exist at scale.

Elephants, in research led by Joshua

Schiffman at the University of Utah,

carry roughly 20 copies of the master

tumor suppressor gene TP53. Humans carry

one. Whales have stacked redundancies in

DNA repair across multiple genes. Naked

mole rats, studied by Vera Gorbunova at

the University of Rochester, produce a

high molecular weight version of

hyaluronic acid that physically prevents

their cells from packing densely enough

to form tumors. They are, in practice,

almost cancer immune. You might think,

"Fine, copy the elephants. Add more

TP53. Make humans naked mole-rat proof."

But these adaptations are tangled into

the entire genome. They evolved over

tens of millions of years with side

effects compensated by other mutations

elsewhere. You can't bolt them onto a

human the way you'd add a feature to a

phone. We're roughly stuck with the

tumor suppression we have, plus whatever

drugs can borrow from the outside. What

strikes me about this, the line that

doesn't get said cleanly, is that humans

are basically halfway up the ladder.

Long-lived enough that mutations

accumulate, not big enough to have

evolved heavy redundancy. Cancer is,

more or less, the bill for hitting 80 in

a body engineered to last 40. There's a

constraint nobody likes to talk about

because it points out where the dream

actually breaks. Roughly 90% of cancer

deaths aren't from the original tumor.

They're from metastasis. Cells that left

the primary site, traveled through blood

or lymph, and seeded colonies in lung,

liver, brain, and bone. A breast tumor

doesn't usually kill you. Breast cells

growing in your liver kill you. This

part of the disease in 2026

is mostly unsolved. Modern oncology is

genuinely good at the primary tumor.

Surgeons can cut it out, radiation can

shrink it, and chemo or targeted drugs

can knock it back. Five-year survival

for early-stage localized breast cancer

in the US now sits at around 99%. That's

a real triumph, and the optimist case

earns it. The collapse comes when these

cells have already left. Stage four

breast cancer five-year survival drops

to about 30%. Pancreatic at the same

stage runs 3%. Lung sits around 8%. The

numbers fall off a cliff because nothing

in the toolkit reliably cleans up cells

distributed across an entire body.

Surgery can't excise 20 organs at once.

Radiation can't dose the whole human,

and the chemotherapy dose required to

kill every micro metastasis would also

kill the patient several times over.

Immunotherapy,

the celebrated breakthrough of the last

decade, helps roughly 20 to 30% of

patients with certain cancer types and

barely touches others. CAR-T cell

therapy works almost magically for some

blood cancers and has spent years

failing to translate to solid tumors.

The headlines are real. They're also

narrow. There's a worse layer

underneath. The one oncologists call

dormancy. Hiding slowly enough to evade

every imaging tool we have. They wake up

when something we don't fully understand

changes. A hormonal shift, a course of

steroids, a drop in immune surveillance

with age. This is why women treated

successfully for breast cancer in their

40s sometimes recur in their 60s. Not

new cancer, old cells finally finishing

the job. There is currently no test that

reliably finds these dormant cells and

no drug that reliably kills them. They

sit below the resolution of our best

imaging and below the threshold of our

best treatments. And yes, the AI and

CRISPR pitch keeps coming. Faster

sequencing, better drug design, models

that predict which mutation responds to

which compound. All real, all useful,

none of it changes the underlying

physics. AI can read a tumor's genome

faster than ever. It can't stop the

tumor's genome from mutating while the

AI reads. The constraint isn't

computational. It is biological. So, the

actual ceiling becomes visible. We can

detect cancer earlier and survive

primary tumors better. We can extend

life with metastatic disease by months,

sometimes years, and keep moving the

line. What we cannot do is abolish it.

The framing of cure was always doing

rhetorical work. Curing implies a state

where the disease is gone, the way polio

is gone from a vaccinated population.

Cancer can't reach that state because

cancer isn't a separate thing. It's what

cells do when their copying machinery

makes the unlucky mistake in a body that

has trillions of cells copying

constantly for decades in a long-lived

animal that wants to keep being a

long-lived animal. Eliminating cancer

means eliminating one of the basic

statistical consequences of being a 37

trillion-cell organism that lives to 80.

The thing oncologists actually pursue

and rarely say out loud is more honest.

Push the average age at cancer death

past the average age of every other

cause of death. Make cancer a manageable

late-life condition like high blood

pressure. Catch it early while the math

still favors you. Treat it long enough

that something else takes the patient

first. That is the actual plan and it

isn't cure. It is containment. 55 years

into the war on cancer, the funding has

done extraordinary things. Childhood

acute lymphoblastic leukemia, once a

death sentence, is now survivable in

over 90% of cases. HPV-driven cervical

cancer can, for the first time in human

history, be functionally prevented in

vaccinated populations. Decades have

been added to lives that would have once

ended in the first round. None of that

is nothing. It just isn't the cure that

was sold. Cancer is the price of being a

multicellular organism with cells that

divide, paid in installments,

accelerated by age, and rewritten by

every drug we throw at it. There is no

breakthrough that closes the bill

because the bill is built into being

alive. Every cell in your body is, right

now, copying itself with errors. And the

only reason you don't have cancer is

that the wrong errors haven't happened

yet. If you'd rather see how the rest of

medicine's promises hold up under

physics, subscribe. I cover what the

press releases skip.