Exercise regularly. Avoid extremes in diet - but eat a bit of variety. Get the minimum recommended amounts of vitamins and minerals. Keep your calories and your body fat low.
And if you have bad genetics, you're probably just screwed, and no amount of exercise or diet will fix it.
There is some evidence that certain kinds of atypical diet are slightly less beneficial in terms of mortality. (Low carbohydrate long term. Cause unknown, not CVD or cancer. Very good review, thoroughly recommended. The key word is "all cause mortality and morbidity") And the varied (in certain ways) Western diet is also shown to be quite bad.
Please correct me if I'm wrong, I like to review nutrition papers fairly and thoroughly but have never seen one on this...
Variety in moderation seems to be the only effective proxy to smooth over the constant "used to be good for you - nope, not any more - wait we were wrong it is good" oscillations.
I mean a variety of plant and animal sources, and a variety of macronutrients - to cover all the bases. With the exception of trans fats, and possibly red meat, nearly every food seems to be safe in moderate doses, so there's no reason to avoid or load up on anything in particular.
And when they find out that something like, for example, oily fish, is beneficial - you can look back and say hey, I ate a bit of that too.
I agree that not all forms of variation are good, but variation within what mainstream nutrition science recommends at this point probably mostly is.
tl;dr - no one really knows anything, so play the odds.
This is a bit like financial advice stating "don't become poor" or health advice stating "don't get cancer".
Of course both of these advices are easier to follow for some people than for others.
Exercise and/or don't stuff your face. Pretty simple, really.
And, unlike your finances or genetic predisposition for cancer, you have total control over it. No one is forcing you to over eat.