This seems like an odd way to split the data. Is it not the case that healthy-weight people sometimes gain weight in their old age? Your claim seems unfalsifiable: Whatever mechanism is at play that makes healthy-weight people gain weight later in life could also be at play for people who lost weight, and 30 years later regained it in their old age.

To put it another way, it's a silly bar to use to say that weight-loss programs must move their adherents level of health wrt weight past those of normal slim people, and that any weight gain from a former dieter is proof of a diet's failure of efficacy (while any weight gain from a formerly-healthy-weight, non-dieter is just happenstance).

It's tough to look at just the existing data to tease this out, but if you need to resort to a bar as absurdly stringent as that to make your point, I'm afraid it's not a very convincing one.