be passed equally through families. He also showed that cancers that occur early in life are more likely to have a genetic component than cancers diagnosed in later years.

Risch also discovered that rare cancers, such as thyroid and testicular cancers, are even more likely to be familial than more common types, such as breast and lung cancer. The results re-emphasize the importance of continuing the search for genes associated with cancer, he said.

"The evidence suggests that inheritance plays a role in most, if not all, cancers," said Risch, professor of genetics and (by courtesy) of both statistics and health research and policy. The results of his analysis are published in the July issue of Cancer Epidemiology, Biomarkers & Prevention.

"The conclusions of the previous study are a direct result of the model the researchers selected to analyze the data. If you use an alternate model -- one that I think fits the data better -- you can come to the conclusion that nearly all cancers have a genetic component," Risch said.

Risch analyzed data from the same 45,000 pairs of twins used in the earlier study, which was published in the July 13, 2000, issue of the New England Journal of Medicine. That study assumed most cancers are the result of the additive contributions of many different genes, each of which has only a minimal effect on disease development. The effect of these genes, which cumulatively account for what is termed heritability, were combined with random environmental effects to calculate a person's disease liability.

In this model, the researchers expected cancer to develop if the liability exceeded a predefined threshold value. They found that when they calculated the relative contributions of genetic and environmental factors to the liability value, environment nearly always won out.

In contrast, Risch found that the twin data for this and two other large-scale twin studies could be better explained using a statistical model that assumed the cancers were the result of either only one gene, or of several relatively rare genes. The new model led him to a very different conclusion: cancers are genetically influenced; rare cancers are more likely to be inherited than more common cancers; and cancers with an early onset are more likely to be inherited than those that develop later in life. In other words, a man whose brother develops prostate cancer at a young age is more likely to develop prostate cancer himself.

"What genes do is decrease the age of onset," Risch said. "People who might develop cancer anyway can develop it at an earlier age."

Risch emphasized that his analysis doesn't rule out the effects of environmental factors.

He used a concept called attributable risk to explain how, in some cases, the contribution of genetics and environment can add up to more than 100 percent. For example, a person whose environment includes a lot of sun exposure is more likely to develop skin cancer than a person who never goes outside. However, a light-skinned person is more susceptible to developing melanoma than a dark-skinned person, proving that genes affecting skin color also play a critical role. Because the environmental and genetic factors are synergistic, when added together they can exceed 100 percent of the risk -- an outcome that's not allowed in the threshold/heritability model.

"I think the heritability concept is limited," said Risch. "It's still important to look for genes associated with all of these cancers." SR