Throughout these decades, refined carbohydrates and sugars received little or no attention in discussions of cancer causation. Peter Cleave had suggested in The Saccharine Disease that the refining of carbohydrates might be involved in colon cancer. John Yudkin had noted that the five nations with the highest breast-cancer mortality in women in the late 1970s (in descending order: the United Kingdom, the Netherlands, Ireland, Denmark, and Canada) had the highest sugar consumption (in descending order: the United Kingdom, the Netherlands, Ireland, Canada, and Denmark), and those with the lowest mortality rates (Japan, Yugoslavia, Portugal, Spain, and Italy) had the lowest sugar consumption (Japan, Portugal, Spain, Yugoslavia, and Italy). But in 1989, when the National Academy of Sciences published its 750-page report on Diet and Health, the authors spent only a single page evaluating the proposition that carbohydrates might cause cancer. “There is little epidemiologic evidence to support a role for carbohydrates per se in the etiology of cancer,” they noted. They did add two caveats. One was that “no definitive conclusion is justified…because carbohydrates have often been reported in epidemiologic studies only as a component of total energy and not analyzed separately.” The other was that Richard Doll and Bruce Armstrong had found sugar intake in international comparisons to be “positively correlated with both the incidence of and mortality from” colon, rectal, breast, ovarian, prostate, kidney, nervous-system, and testicular cancer, and that “other investigators have produced similar findings.”

The patterns of cancer incidence, for many cancers, are similar to those of heart disease, diabetes, and obesity, which alone suggests an association between these diseases that is more than coincidental. This was the basis of Cleave’s speculation, of Dennis Burkitt’s, and of those cancer epidemiologists who argued that dietary fat caused breast cancer. But if dietary fat, red meat, man-made chemicals, or even the absence of fiber cannot explain the “strikingly similar” patterns of disease distribution, as the Harvard epidemiologist Edward Giovannucci remarked about colon cancer and Type 2 diabetes in 2001, then something else most likely does.

Those cancers apparently caused by diet or lifestyle and not related to tobacco use are either cancers of the gastrointestinal tract, including colon and rectal cancer, or cancers of what are technically known as endocrine-dependent organs—breast, uterus, ovaries, and prostate—the functions of which are regulated by hormones. This connection between these diet-and life-style-related cancers and hormones has been reinforced by the number of hormone-dependent factors linked to cancers of the breast and the endometrium (the lining of the uterus). All suggest that estrogen plays an important role. All these cancers, with the possible exception of pancreatic and prostate cancer, appear to increase in incidence with weight gain. These associations together imply both a metabolic and a hormonal connection between diet and cancer. This in turn led breast-cancer researchers to focus their attention on the likely possibility that obesity increases the incidence of breast cancer by increasing estrogen production.