Risks II: Cancer

Part II of the Risks series. Part I covered cardiovascular disease.

Cancer is the second leading cause of death in this country, behind heart disease. For the age group I take care of — call it 38 to 58 — it is the one people are actually afraid of. Nobody lies awake worried about their LDL. Plenty of people lie awake after a friend their age gets a diagnosis.

That fear is not irrational. Roughly one in two men and one in three women will get a cancer diagnosis at some point. About 2 million new cancers were diagnosed in 2022, and the projection for 2026 is around 2.1 million. Breast cancer is the most common diagnosis, followed by prostate, lung, and colorectal.

But fear is a bad organizing principle. So here is a better one. There are exactly three places you can intervene on cancer: lifestyle on the front end, screening in the middle, and treatment on the back end. Everything in this article fits in one of those three buckets, and the buckets are not equally powerful — which turns out to be the most interesting thing about the whole subject.

What actually drives the risk

The number one risk factor for cancer is age. Not diet, not stress, not your water filter. Age. The average age at diagnosis across all cancers is 67. A handful skew young — testicular around 33, Hodgkin lymphoma around 39, cervical around 50, thyroid around 51 — but the overwhelming pattern is that cancer is a disease of accumulated cellular time.

I say that up front because everything else in this section is modification at the margins, and I would rather you know that than think you can outrun this with a supplement.

That said, the margins are worth a lot. Roughly 40% of cancer cases trace back to modifiable factors:

• Tobacco, in any form

• Excess weight and metabolic dysfunction

• Alcohol

• UV exposure

• Physical inactivity

• A handful of infections — HPV, hepatitis B and C, H. pylori

Two of these deserve more than a bullet.

Alcohol. It is classified as a Group 1 carcinogen, the same category as tobacco and asbestos. The dose-response relationship starts lower than most people expect, and it is causally linked to cancers of the breast, colon, liver, esophagus, and head and neck. Almost nobody I talk to knows this. It is the single most underappreciated item on the list, and I will give it a full article in November.

Excess weight and insulin resistance. Thirteen cancers are formally classified as obesity-associated. Insulin resistance frequently travels with obesity, and there is a plausible, well-studied mechanism — chronic hyperinsulinemia is mitogenic, and insulin-like growth factor signaling promotes cell proliferation and suppresses apoptosis. I want to be careful with the verb here: the association is strong and the mechanism is credible, but "insulin resistance causes these cancers" is a stronger claim than the evidence supports. What I will say confidently is that fixing metabolic health is one of the few interventions that lowers risk across cardiovascular disease, cognitive decline, and cancer simultaneously. That is a rare deal.

The things people worry about that don't deserve it

Stress, sugar, plastics, and "toxins" occupy an enormous share of the public conversation about cancer and a small share of the actual risk. I want to be precise about why, because the honest answer is different for each one.

Stress and sugar are not established carcinogens. There is no good evidence that psychological stress causes cancer, and no evidence that dietary sugar "feeds" tumors in any way that matters clinically. Where both matter is indirectly — chronic stress and a high-sugar diet both drive the metabolic dysfunction that is on the list. That is a real pathway, and it is not the same as the direct one people imagine.

Industrial and environmental carcinogens are real and dose-dependent. This is where the honest answer gets uncomfortable, because it cuts both ways. Chronic occupational exposure — industrial settings, firefighting and first response, certain manufacturing — meaningfully raises cancer risk, and those populations deserve more screening than they get. Trace exposures in ordinary consumer life are a different order of magnitude. Minimize what is easy to minimize. Do not reorganize your life around it while you are still drinking four nights a week.

Family history

About 5% to 10% of cancers are hereditary. That number is worth sitting with, because it says two things at once: the vast majority of cancer is not written in your genes, and the minority that is matters enormously for the people who have it.

Know your family history. Specifically: who, which cancer, and at what age. Early-onset disease, multiple cancers in one relative, the same cancer across generations, or a known pattern like breast-plus-ovarian should trigger a real conversation about genetic testing. "My grandfather had cancer in his eighties" should not. The difference between those two is the difference between a screening plan and unnecessary anxiety.

Where the money goes, and what it buys

Here is the part that reframed this topic for me.

Global spending on cancer medicines reached $252 billion in 2024, growing at about 12% per year. US cancer care overall — drugs, surgery, radiation, imaging, hospitalization — was roughly $183 billion in 2015 and is projected to exceed $245 billion by 2030. Oncology now accounts for roughly a third of the pharmaceutical development pipeline. Public research funding runs about $7.2 billion per year through the National Cancer Institute.

And it has worked. The US cancer death rate has fallen about 34% since 1991, which the American Cancer Society translates to roughly 4.8 million deaths averted through 2023. Five-year survival across all cancers is now about 70%, up from 49% in the mid-1970s. Those are real numbers and real people.

Now the part that should change how you think.

A 2024 modeling study in JAMA Oncology looked at five common cancers over 45 years and asked where the averted deaths actually came from. The answer: about 80% came from prevention and screening. Only 20% came from treatment. For lung cancer, 98% of averted deaths traced to smoking cessation. For cervical cancer, essentially all of it was screening.

So we spend overwhelmingly on the back end and we get our results overwhelmingly from the front end.

It gets sharper. A landmark analysis of 58 cancer drugs found the benefit-adjusted launch price per year of life gained rose from about $54,000 in 1995 to $207,000 in 2013 — roughly 10% per year. Average survival benefit across those drugs: 0.46 years. And newer drugs showed no greater survival gain than older ones. Price went up. Benefit did not.

I want to be fair to the treatment side, because the 80/20 finding is easy to over-read. Treatment gains are concentrated in cancers where prevention and screening have nothing to offer — metastatic melanoma, myeloma, certain leukemias, targeted therapy in lung cancer. Those gains are genuine and in some cases extraordinary. The study also modeled five cancers, not all of them. But the direction of the finding is not in dispute, and the mismatch between where the dollars go and where the life-years come from is the single most important fact in this article.

Screening

Only four cancers have a screening test recommended by the US Preventive Services Task Force. Those four represent about 29% of all cancers diagnosed in this country. Fifty-seven percent of diagnosed cancers have no recommended screening test at all.

Which is why the number that surprises people most is this one: only about 14% of cancers in the US are detected by a recommended screening test. The variation underneath that average tells the story — roughly 61% of breast cancers are screen-detected, versus about 3% of lung cancers.

The current menu:

• Colorectal — everyone, starting at 45. Annual FIT, Cologuard every 3 years, flexible sigmoidoscopy every 5 years, or colonoscopy every 10. Colonoscopy is the only one on this entire list that prevents cancer rather than finding it, because you can remove the polyp before it becomes anything. Note the starting age: it moved from 50 to 45, because colorectal cancer is rising in people under 50.

• Breast — mammography, with the starting age and interval depending on risk and which guideline you follow.

• Cervical — Pap and HPV testing.

• Lung — annual low-dose CT, but only for people with a significant smoking history.

Skin exams are cheap, low-tech, and skipped by most of the people who need them.

Symptoms that override any screening schedule

Screening is for people without symptoms. If you have any of these, you do not wait for your next scheduled test:

• Unexplained weight loss

• Rectal bleeding

• Blood in the urine

• A lump persisting more than two weeks

• Postmenopausal bleeding

• A cough that won't resolve

• Hoarseness lasting more than a couple of weeks

Why earlier is better — and where that logic breaks

The case for screening is straightforward when you look at survival by stage. Colorectal cancer caught while localized has a 91% five-year survival. Caught after it has spread to distant sites, 16%. Same disease. Six-fold difference based on when it was found.

That comparison is real, and it is also partly an illusion. Three biases inflate it:

• Lead-time bias. If a cancer is going to kill you in 2032 either way, finding it in 2026 instead of 2029 adds three years to "survival" and zero days to your life.

• Length-time bias. Screening preferentially catches slow-growing tumors, because aggressive ones show up symptomatically between tests. The screen-detected group is enriched for indolent disease.

• Overdiagnosis. Some cancers found by screening would never have caused symptoms. Every one is counted as a success. Every one gets treated.

The cleanest proof that these are not academic concerns comes from ovarian cancer. The UKCTOCS trial followed over 200,000 women for a median of 16 years. Screening found 39% more early-stage cancers and 10% fewer late-stage cancers. It shifted the stage distribution exactly as designed.

It did not reduce ovarian cancer deaths. Not at all.

Finding cancer earlier only helps if there is a treatment at that earlier stage that changes the outcome. Colorectal screening works because you can remove a polyp. Ovarian screening failed because finding stage II ovarian cancer doesn't give you a better answer than finding stage III.

I will spend a full article on this in December. For now: earlier is usually better, and "usually" is doing real work in that sentence.

The new tools

Two categories are being marketed hard to people like my members, and they deserve an honest read.

Personalized risk tools. Platforms that combine family history and genetic data to generate an individualized screening protocol. This is the least controversial thing in this section — it is essentially systematizing the family-history conversation, and it tends to produce more appropriate screening rather than just more screening.

Multi-cancer early detection blood tests. The best known is Galleri, from GRAIL — a blood test that reads DNA methylation patterns to screen for many cancers at once. The premise is genuinely compelling: throw a wider net and catch the 57% of cancers that have no screening test.

The evidence, as of this writing, is mixed and recent. In February 2026, the NHS-Galleri trial — 142,000 asymptomatic adults, three years — missed its primary endpoint. It did not produce a statistically significant reduction in stage III and IV diagnoses. It did show a meaningful reduction in stage IV diagnoses and a higher overall detection rate, and GRAIL has submitted an FDA premarket approval application.

So: a real reduction in the worst-stage diagnoses, an unmet primary endpoint, and still no mortality data. That is not nothing, and it is not yet the thing it is being sold as. If you want one of these tests, I will order it and talk you through what a positive result actually means — which, right now, most often means a diagnostic workup and a period of significant uncertainty. What I will not do is tell you it is proven to help you live longer, because that study has not read out.

What's in your control

If there is one idea running through this entire series, it is this: control what you can control, and stop spending energy on what you can't.

Not in your control

• Age. The number one risk factor, and it only goes one direction.

• Genetics. You can know your family history. You cannot change it.

In your control

• Your metabolic health, your alcohol intake, whether you use tobacco, whether you move, whether you burn.

• Whether you actually complete the four screenings that have evidence behind them.

• Whether you get the workup when something is wrong instead of waiting nine months.

That second list is shorter than the wellness industry would like you to believe and longer than the fatalists would. It is also, notably, the same list that shows up in the cardiovascular article, and it will show up again when I write about cognitive decline. The upstream work is not cancer-specific. That is the good news hiding in an article about cancer.

3 Bullet Point Summary 

• Age is the number one risk factor for cancer, and nothing else is close. Roughly 40% of cases trace to things you control. That leaves 60% that you don't — and the honest version of this article says both halves out loud.

• We spend on treatment and we get results from prevention and screening. Roughly four out of five cancer deaths averted over the past 45 years came from prevention and screening. Almost all the money goes the other direction.

• Only 14% of cancers in this country are found by a recommended screening test. Not because screening doesn't work, but because most cancers have no screening test at all.

Three Point Take Away

1. Write down your family history this week. Not "cancer runs in my family." Who, which cancer, what age at diagnosis, on which side. One page. Bring it to your next visit. This single document does more to individualize your screening plan than any test you can buy, and it costs you twenty minutes and a couple of phone calls.

2. Pick the one modifiable factor you're actually avoiding and address it. You already know which one it is. For most people in my practice it is alcohol, and the reason it is alcohol is that nobody told them it was a Group 1 carcinogen. Metabolic health is the other big one, and it pays out across three different disease categories at once.

3. Get current on the four screenings with real evidence, before you buy a fifth thing. Colorectal starting at 45, breast, cervical, and lung (if you have a smoking history). If you are behind on any of those, being behind is a bigger problem than not having a whole-body MRI. Fix the proven thing first.

Next in this series: Risks III — Cognitive Decline.