The problem with “narrow[ing] down the characteristic substance” is that all foods have a variety of substances which interact in different ways with the human metabolism. As others have pointed out, this may vary due to interactions with other chemicals, rate of absorbtion and breakdown during digestion, and methods of preparation–cooking, for instance, which we do to make many foods more palatable or digestable, destroys many naturally occuring nutrients and creates substances that can be carcenogenic or otherwise detrimental. Antioxidants, for instance, have been touted for their benefits over the last few years, despite the fact that most don’t survive digestion and have a questionable impact on health. While it’s true that an antioxidant will, under lab conditions, combine with a free radical and thus prevent it from attacking proteins in the body, in an actual metabolic environment there are so many interactions occuring that it’s hard to tell exactly what effect any individual compound is going to have. In excess quantities antioxidants can actually create free radicals, as can the monounsaturated fats that everyone is so hip-happy about. (Check out the rapeseed oil debacle in Spain the early 'Eighties and you’ll never look at a bottle of Canola oil the same way again.)
It’s useful to bear in mind that, for the most part (baring co-evolved and domesticated species) these “foods” aren’t evolved to offer nutrition but develop these substances for their own purposes (to act as natural pesticides, or aid in some stage of development, or attract pollen carriers, or whatnot) and are nutritionally beneficial or not to humans in a purely random fashion. Not everything you can put in your mouth is good for you even if it is “all-natural”.
Nutritional studies are epidemological experiements that transpire over long time periods (often years or even decades) with understandibly loose controls; often, the study parameters are created retroactively and the data filtered to match the parameters rather than created beforehand; after all, you’re hardly going to dictate to a large group of people what foods they can eat, how much, and how it can be prepared over the next 20 years. As a result, study results (which are often presented as factual conclusions in the popular media) are inferrential indicators, not hard and fast conclusions.
As others have indicated, phytochemicals comprise a vast array of different compounds, some of which may be beneficial, others may be harmful, most probably offering no significant interaction at all. Promoting “phytochemicals” as being good for you is like condemning all steroids as being bad just because a small subset of metabolic steroids (when taken in excessive quantities) can have negative consequences. I daresay, given our still limited understanding of the causes of these afflictions, that no one has yet “proven anti-cancer and anti-heart disease effects” of any phytochemical compounds. There may be correlative links, or even suggested causivity, but “proof” of such claims is a dubious claim at best, and while they might offer some small margin of benefit against heart disease or cancer, the big factors, such as general nutrition (avoiding saturated fats and getting your grains and vitamins), unhealthy behaviors (smoking, excessive drinking, exposure to toxic compounds), and genetics (nothing you can really do there), are going to predominate. If you are genetically prone to, say, breast cancer, you certainly want to avoid other behaviors which are correlated with cancer, but the odds that you’ll get cancer are largely controlled by your genes, regardless of how many antioxidant supplements you gulp down.
As tremorviolet suggests, the best thing to do, if you’re concerned about getting these in your diet, is to eat unprocessed and uncooked (or barely cooked–steamed or lightly grilled) vegetables and fruits. Not only will you get the (unknown and highly speculative) benefits of your phytochemicals but you’ll also get plenty of good old fashioned vitamins, minerals, and fiber, which ain’t nothin’ but good for you.
Stranger