In assessing risk, theoretically, common factors are sought between persons exhibiting a condition, then research done to find how those common factors may (or may not) impact risk.
If there exists a significant number of women who have had abortions, and those abortions caused an unnatural change in stem cell biochemistry, with biochemical differences between miscarriages and induced abortions, and that change in biochemistry can be linked to tumor development, then causality may be established.
There is no guarantee that having an abortion will cause cancer, any more than that not having an abortion will guarantee that a woman will not contract breast cancer, only the indication that the cell changes brought about by induced alterations of biochemistry (via abortion) increase the risk of contracting breast cancer.
If other risk factors are present as well, this only serves as an indicator of probability in a large population that any individual will be more likely (than someone who does not have these risk factors) to contract the disease.
Where this type of assesment becomes scary, aside from the potential for bogus, agenda driven research and enriching trial lawyers, is when you seek insurance and your medical records (and previous generation's medical records) are there on the computer for all to see. Even with (especially with?)socialized medicine, you become a bad medical investment, and do not receive coverage or care.
In this case, however, the agenda bias is against finding a link, socially, medically, in liability terms, and from the research folks who find aborted babies a convenient source of stem cells and other tissue on which to conduct research.