Can Food Cause Cancer? An Evidence-Based Reading of Bracken Fern, Acrylamide, and Natural Toxins

Food and Cancer: From Hazard Identification to Real-World Risk

The conclusion in one sentence

Whether a food causes cancer cannot be judged merely by asking whether it contains a particular substance. Species, dose, processing method, and human evidence must be assessed together. Bracken fern does pose a ptaquiloside hazard, but this does not mean that every fern or every instance of consumption will cause cancer. PMID: 39772526; PMID: 34512863

The fern controversy: identify the plant first

Professor Ching-Shun Lin noted that the plant with an extensive body of long-term carcinogenicity research is bracken fern, Pteridium aquilinum; all edible ferns should not be treated as one and the same. This fundamental distinction agrees with the current literature. Bracken fern has been classified as a plant that is possibly carcinogenic to humans, and its principal naturally occurring carcinogen is ptaquiloside. PMID: 41681633

Ptaquiloside can form a reactive dienone that subsequently produces mutagenic DNA adducts. The statement that bracken fern contains a genotoxic substance is therefore not an internet rumor; it is supported by chemical mechanisms and experimental evidence. PMID: 39772526

At the human level, the available evidence still comes mainly from observational studies. Associations have been reported between bracken fern consumption and gastric and upper gastrointestinal cancers. These findings justify a cautious view of the risk, but an association cannot be rewritten as a definite causal effect for every individual. PMID: 2337510; PMID: 10396523

Exposure is not limited to eating the fern itself. When grazing animals ingest bracken fern, milk and dairy products may become routes of ptaquiloside transmission. More recent studies also indicate that processing can alter the forms of the compounds present. These are scientific advances that refine the details of risk; they do not overturn Professor Lin’s earlier warning about potential exposure pathways. PMID: 39147486; PMID: 38202326

Ptaquiloside released by bracken fern may also enter soil and water, which is consistent with Professor Lin’s discussion. The claim that “every part of the entire plant contains it,” however, should not be stated categorically: more recent testing did not detect it in rhizomes, showing that differences among plant parts and samples still need to be acknowledged. PMID: 40806289; PMID: 34458862

Cooking can reduce risk, but “it was cooked” is not the whole answer

Traditional processing and thorough cooking can substantially reduce ptaquiloside exposure, and research suggests that everyday exposure after such treatment may fall within a safer range. Inadequate processing or high exposure should nevertheless be avoided. Professor Lin’s precautionary advice not to eat it in large amounts or frequently is therefore well founded, but it should not be inverted into a claim that toxicity must remain unavoidably significant even after thorough cooking. PMID: 36850990

Diplazium esculentum, a fern commonly eaten in Taiwan, is taxonomically distinct from bracken fern; this identification agrees with the literature. Its toxicity and safety in humans have not yet been fully evaluated, however. The more precise conclusion is therefore: “There is insufficient evidence to treat it as bracken fern, but also insufficient evidence to guarantee that there is no reason for concern.” PMID: 34512863; PMID: 36444246

Acrylamide: a clear formation mechanism, but unproven causation of human cancer

Cold storage can cause cold-induced sweetening in potatoes and increase reducing sugars. During subsequent high-temperature frying or baking, those sugars can react with asparagine through the Maillard reaction to form more acrylamide. Professor Lin’s explanation of this pathway agrees with the literature. PMID: 40282310; PMID: 38472765

Yet “acrylamide is present in food” does not mean that “ordinary dietary exposure inevitably causes cancer.” Meta-analyses have found no clear increase in risk for most cancers examined, while a more recent systematic review found a mixture of positive and null results. The most accurate update today is that human observational evidence remains inconsistent and a clear causal relationship has not been established—not that every possible signal has been conclusively ruled out. PMID: 35548558; PMID: 36673439

A cohort study of fried potatoes found that more frequent consumption was associated with higher mortality, whereas the same association was not seen for unfried potatoes. Professor Lin specifically emphasized at the time that this was an association; he did not present acrylamide as a proven cause of death. That boundary is important. PMID: 28592612; PMID: 25403648

Pesticide residues: detection, hazard, and actual risk are different questions

Detecting glyphosate in food does not automatically mean consumers must stop eating it. Human biomonitoring and food-residue surveillance suggest that estimated exposure or overall dietary risk is low in most settings. These data, however, cannot provide an absolute safety guarantee for a particular batch of a product consumed in a particular way. PMID: 40311231; PMID: 42095190

The literature has evolved: a more recent long-term rat study reported increases in tumors in multiple organs under a design that included the acceptable daily intake, while a systematic review concluded that preclinical data remain limited and further research is needed. These findings cannot be used to infer directly that foods with low residues will cause cancer in humans. They do mean, however, that the blanket assurance “low residue guarantees there is no important concern” should no longer be expressed with excessive certainty. PMID: 40490737; PMID: 41600575

Natural does not mean risk-free: Thai basil and Gynura bicolor

There is no established human evidence that Thai basil in ordinary dietary amounts causes liver cancer. Basil can, however, contain methyleugenol and estragole, substances for which genotoxic or carcinogenic evidence exists in rodents. The most defensible formulation after these scientific developments is therefore: “Human risk at normal dietary intake has not been established, but there is hazard evidence at the level of individual constituents.” PMID: 36140952; PMID: 40741360; PMID: 40005118; PMID: 40646165

This also illustrates why chemical names must not be conflated: eugenol, methyleugenol, estragole, and safrole are different substances, and toxicological evidence for one cannot simply be assigned to another. Professor Lin was right to oppose fearmongering based on confusion among similar names, although direct data are still lacking for quantitative claims about ordinary consumption and human excretion. PMID: 36367523; PMID: 40370109

Gynura bicolor may contain pyrrolizidine alkaloids, a class of substances with hepatotoxic and genotoxic properties. Professor Lin’s advice against prolonged consumption in large quantities therefore has a valid hazard basis. Place of origin may affect composition, however, and current evidence cannot convert “a balanced diet” or “occasional consumption” into a defined safety boundary. The advice should be understood as a precautionary principle, not as a validated threshold. PMID: 27623358; PMID: 30704105; PMID: 32517927

Diet studies: a risk association is not an order to stop eating a food

Pooled findings linking pickled-food intake to increased gastric cancer risk, as well as the positive association between kimchi intake and gastric cancer in Korean populations, support Professor Lin’s description of “increased risk.” The literature, however, has not established a clear boundary at which avoiding daily or every-meal consumption becomes safe. PMID: 22499775; PMID: 25339056

Research on fish teaches the same lesson. A large cohort observed that intake of total fish, tuna, and nonfried fish was associated with a higher risk of melanoma, but the study design could not establish that eating fish causes melanoma. Professor Lin highlighted problems with questionnaires and confounding and opposed issuing a direct instruction to eat less fish. That position is consistent with a cautious interpretation of observational research. PMID: 35676377

Supplements and cancer-treatment claims: human evidence is the threshold

Cell and animal findings on Antrodia cinnamomea cannot be treated as reliable proof that it cures cancer or prolongs life. Human cancer trials are limited, and their overall methodological quality is inadequate. Professor Lin’s opposition to exaggerated claims of anticancer efficacy therefore remains consistent with the literature. PMID: 33424591; PMID: 41853122

The statement that there is “absolutely no human clinical evidence,” however, needs to be updated as science advances. Subsequent randomized, double-blind human trials have examined other health conditions; some surrogate outcomes improved, while another trial’s primary outcome was not significant. These studies still do not establish broad medical benefits, much less translate into evidence of efficacy against cancer. PMID: 32657670; PMID: 36476310

Practical principles for evaluating claims

When you encounter the claim that “a certain food causes cancer,” first determine whether it refers to the whole food, a particular species, or one compound within it. Next, ask whether the research involved cells, animals, observational population data, or human trials capable of addressing efficacy and causation. Then check the route of exposure, processing and cooking methods, and frequency of intake. Only after that should individual risk be discussed.

An important contribution of Professor Lin’s articles is their consistent separation of “detection” from “harm,” “carcinogenicity in animals” from “cancer caused by an ordinary human diet,” and “statistical association” from “causation.” Where the literature has advanced, the degree of certainty should be adjusted in light of new evidence. Professor Lin’s original words should not be rewritten, nor should the evolution of science be portrayed as though the professor himself were continually revising the original articles.