A news item crossed my feed over the weekend: one in eight cancer cases worldwide is caused by infection. 2.3 million new diagnoses in 2024 — HPV, hepatitis B and C, the bacterium Helicobacter pylori, Epstein–Barr virus. Outlets spread the conclusion: a substantial share of these tumors is preventable through vaccines, screening, and antibiotics, while access to those tools is unevenly distributed.
The comments beneath the post went in the usual direction. Some were recalculating that diet, obesity, and tobacco matter more than infections; others launched into a long argument over what is preferable — a full life to forty-five or a terrible death at eighty. A good conversation. More interesting, though, is the question that never came up — and it weighs more than the number itself.
In the sixties and seventies, the viral theory of cancer was American oncology’s main bet: it received a budget comparable to a major space program, and was officially declared a failure. Today, 12% of cancer cases are infectious. Failure and outcome describe the same phenomenon forty years apart, and that discrepancy is worth untangling.
The WHO’s International Agency for Research on Cancer (IARC) published its analysis in The Lancet Oncology on September 28. It counted twelve agents recognized as carcinogenic to humans: viruses, one bacterium, and several parasites. The numbers line up like this: Helicobacter pylori — 760,000 cases; HPV — 750,000; hepatitis B virus — 360,000; Epstein–Barr virus — 260,000; hepatitis C — 160,000. Those five agents account for almost the entire total.
Two shifts from the previous estimate. Epstein–Barr virus overtook hepatitis C for the first time because its share of stomach cancer was recognized — previously it had been firmly confined to nasopharyngeal cancer and lymphomas. And sixteen new infection–tumor links were added to the list, absent from global estimates before: HIV, Merkel cell polyomavirus, and an expanded set of tumors caused by hepatitis viruses.
Geography pulls everything toward one point. East Asia accounts for 42% of infection-related cases — largely on the strength of the H. pylori–hepatitis B pair. Next comes sub-Saharan Africa, where HPV, Kaposi sarcoma herpesvirus, and HIV are at work; HIV breaks down immune surveillance and multiplies their contribution. Three quarters of cases occur in low- and middle-income countries.
The phrase “caused by infection” in studies like these is a statistical construct. The attributable fraction is obtained by multiplying the prevalence of infection by the relative risk for a specific tumor. The result links infection to hundreds of thousands of cases that no one diagnoses individually as “infectious cancer”: for any particular patient, no one measures what share of a tumor grew out of chronic inflammation and what share came from tobacco.
Then comes the most interesting part, because the same research group has been publishing these estimates for thirty years. 1990 — 15.6% and 1.45 million cases. 2002 — 17.8% and 1.9 million. 2008 — 16.1% and 2 million out of 12.7 million. 2012 — 15.4% and 2.2 million out of 14 million. 2018 — 13% and the same 2.2 million. Now: 12% and 2.3 million.
The share falls; the absolute number rises. The denominator — worldwide cancer incidence as a whole — is growing faster than the numerator, and that is the only stable trend across the entire series. The authors state outright: the percentages cannot be compared across years, because the sources and quality of the data changed. “Cannot” here sounds like an answer to the public’s habit of subtracting one from another and declaring that infections are “retreating.”
Where did the bet on a virus come from in the first place? In 1961, in Niles, a Chicago suburb, leukemia was found in a dozen children over a short span of time, and the word “epidemic” began creeping through the town. As late as the mid-sixties, public health authorities were receiving letters asking whether cancer could be caught from someone in whose house it had lived. Fear was so widespread that in 1964 the National Cancer Institute opened the Leukemia Virus Program — later the Special Virus Cancer Program, SVCP.
The program was organized in a way unusual for science: instead of grants to independent groups, large contracts under centralized management, all subordinate to one goal — find a human cancer virus and make a vaccine. Over fifteen years, more than $6.5 billion in 2017 dollars went into it, more than the later Human Genome Project cost. The political groundwork came in 1971: in May, Nixon requested an additional 100 million and promised a program to conquer cancer; in December, he signed the National Cancer Act, which gave the institute the right to submit its budget directly to the president, bypassing the NIH.
They did not find the main thing. The search was derailed by false leads that people inside the field called “rumor viruses”: most often, they turned out to be laboratory contamination of human cells by animal viruses. From outside, the program was attacked for its very design — contractual bureaucracy versus peer-reviewed science carried out by individual researchers. By the late seventies, the program was shut down.
And this is where the fork in the road begins — the reason it was worth digging into. Oncogenes and reverse transcriptase came out of that program. Howard Temin and Satoshi Mizutani, together with David Baltimore, showed in 1970 that viral RNA is transcribed back into DNA — a reversal in the flow of genetic information that earned a Nobel Prize five years later. One sardonic virologist put the story better than I could: “One sick chicken, two Nobel Prizes” — referring to Rous sarcoma virus, for which Rous received his in 1966.
At the same time, the political compromise that historian Robin Scheffler calls biomedical took shape: under Truman, activist Mary Lasker pushed for universal health insurance, failed, and switched to advocating federal research funding. The state did not enter medicine — the state entered medical science, and the “war on cancer” became, among other things, compensation for an unfinished healthcare system. This is an American peculiarity: Britain and France invested in treatment, Germany in the environment.
The hypothesis the big program failed to prove was eventually confirmed piecemeal, agent by agent. The first example is Taiwan. Universal hepatitis B vaccination began there in July 1984, and liver cancer incidence among children aged 6–14 started moving down: 0.70 per 100,000 in 1981–1986, 0.57 in 1986–1990, 0.36 in 1990–1994. Among children aged 6–9, the gap was sharper still: 0.52 for those born in 1974–1984, versus 0.13 for those born in 1984–1986. Twenty years of follow-up showed the effect did not remain confined to childhood: the relative risk for those vaccinated at birth was 0.31.
The second example is England and HPV. The program began in 2008 with the bivalent vaccine. Among girls vaccinated at 12–13, cervical cancer incidence was 87% lower than among unvaccinated peers from earlier generations; among those vaccinated at 14–16, it was 62% lower; among those offered the vaccine at 16–18, 34% lower. By June 2019, the country had counted roughly 450 prevented cancer cases and 17,200 fewer cases of precancer than expected. In June 2026 came mortality data: in 2020–2024, not a single death from cervical cancer was recorded among women aged 20–24 in England — the first time this had happened for five consecutive years, whereas without vaccination around twenty-three would have been expected. In total, the vaccine is estimated to have saved around two hundred lives.
And right alongside it — a failure in coverage numbers. By age fifteen, 76–86% of girls had been vaccinated, below the WHO target of 90%; among boys, coverage was 71–80%, with no formal target in place. In Australia, where the program has been running since 2007 and HPV prevalence among those covered has fallen by 90%, coverage among fifteen-year-olds slid from 85.7% to 79.5% over four years. A good vaccine does not work where it is not administered.
The third example is Linqu in China, where they tested an antibiotic rather than a vaccine. In a county with high stomach cancer incidence, 3365 residents were assigned in a 2×2×2 design: a two-week course of H. pylori eradication therapy, vitamins, garlic, and placebo. Twenty-two years after the intervention, the protective effect of eradication on incidence held: odds ratio 0.48. Mortality from stomach cancer fell in all three groups, including the garlic group, though garlic had no effect on incidence at all. A two-week course of antibiotics, assessed twenty years later — that is an unusually long distance for an intervention study.
Epstein–Barr virus — 260,000 cases, and second place after hepatitis C in the number of things left undone. Around 95% of adults are infected; there is no licensed vaccine, preventive or therapeutic, and no specific antivirals either. The first bet, on the surface protein gp350, produced antibodies and reduced the incidence of infectious mononucleosis, but did not stop infection itself. Moderna has brought two mRNA products into clinical trials — one preventive, one therapeutic — and this is so far the most advanced point in a field where the virus’s carcinogenicity has been known for more than sixty years and nothing has been licensed.
Merkel cell polyomavirus is a rare case where causality has been proven almost like a textbook demonstration. It was discovered in 2008 through “digital transcriptome subtraction”: looking for foreign transcripts in a tumor. The virus was found in roughly three quarters of Merkel cell tumors, and its DNA is clonally integrated — meaning infection preceded the expansion of the tumor clone. And viruses from tumors carry stop mutations that cut off the helicase domain of the large T antigen while preserving the retinoblastoma protein-binding motif. Such a virus physically cannot be an accidental passenger: it can no longer replicate, and without integrating into the genome it would simply disappear. So far, this gives medicine little — the cancer is rare, and there is no vaccine — but the forensics of causality here are cleaner than for most known carcinogens.
The third is H. pylori itself: 760,000 cases, first place. There is no vaccine against it and none in sight for the foreseeable future, but unlike viruses it can be cured with antibiotics. That is precisely why it stands alongside vaccines on the WHO list of priorities: the tool is simple, while coverage is constrained by the organization of healthcare, not science.
The most interesting point is the old logic of “destroyed one bacterium, grew another cancer.” H. pylori is associated with a reduced risk of esophageal adenocarcinoma, while the prevalence of infection is declining; hence the hypothesis that mass eradication feeds the growing incidence of esophageal cancer. They checked everyone who received eradication therapy across five Scandinavian countries between 1995 and 2018: 661,987 people, 5.5 million person-years of follow-up, 550 cases of esophageal adenocarcinoma. The standardized incidence ratio was 0.89 — meaning no additional risk, and 11–24 years after treatment it was lower than the background rate: 0.73.
The caveat is of exactly the same kind as with IARC’s percentages: Gastroenterology published a letter calling for broader studies and greater methodological rigor. Registry cohorts do not close the question once and for all, and that is the normal status of the question, not a scandal.
The second divergence is Japanese. Japan suspended its active recommendation of the HPV vaccine in June 2013 and restored it only in April 2022 — nine years later. Coverage among girls aged 11–16 fell as low as 0.3% in 2016; by 2020, it had risen only to 15.9%. The reasons cited are an adverse-event reporting system changed in April 2013, which allowed reports not only from doctors: a flood of reports about vaccinations from previous years, a precautionary decision, media noise, and legal disputes. The modeled price of the pause: 24,600 to 27,300 additional cervical cancer cases and 5,000–5,700 deaths. Japan, in other words, has its own bill — not from a failed program, but from fear.
And one separate thread I did not pull: the same Epstein–Barr virus, beyond tumors, turned out to be the leading cause of multiple sclerosis. In a cohort of more than ten million US military personnel, the risk of multiple sclerosis rose 32-fold after EBV infection and did not rise after cytomegalovirus or other viruses. This is already beyond oncology — the story of an agent working on several fronts at once.