[devonsolomon]

The article is titled as a mystery and then framed as a mystery while answering the question: lifestyle factors heavily influence cancer.

The article demonstrates this by a known factor, tabacco. Other lifestyle factors are similarly predictive and already known, like diet and pollutants (which the article mentions).

The article curiously doesn’t mention detection bias: the tendency for increased access to diagnostic tools or medical evaluations to lead to higher rates of detection/incidence. This is almost certainly true of comparing rural Armish to city dwellers, and has a demonstrated enough effect to explain the difference in non-tobacco related cancer cases between the two groups - and much more likely an influence than genetic protective differences.

The genetic angle is likely to be so minimal to be trivial, IMO.

Long story short: this isn’t very new or interesting.

[rurban]

I find it very interesting, because just recently I argued with the director of a clinic who said genetics control everything. And I vehemently disagreed. He was the expert, but still complete wrong. And also politically disturbing.

[biomcgary]

I'm a computational biologist working for a health prediction biotech. Genetics only explains about 20% of disease variance (on average across many disease, additional caveats apply :-).

[devonsolomon]

It’s true that although there’s long been ample evidence that cancer is a preventable disease (most cancers anyway), the medical community upholds the ‘randomness’ myth.

I believe this is because 1. Doctors don’t want to blame their patients for becoming sick. 2. Many of the lifestyle factors are societally systemic and feel too big to try change. 3. The money is in treatment, not prevention.

Edit: (See: https://pmc.ncbi.nlm.nih.gov/articles/PMC2515569/#:~:text=Th....)

[devonsolomon]

Just to give a case in point on detection: if you autopsy men over 80 that die of unrelated causes, 80% have evidence of prostate cancer (see Dr Gleason study).

You can imagine how sensitive detection/incidence amongst this cohort is to detection bias.