Never rely on a secondary source; ever moreso when it's written on a meta-study.
the whole ihealth thing is based off two conflicting papers - one in mice and another in rats. Neither included work in a human model or trial. Both species of rodent responded differently - the rats saw improved insulin sensitivity and lower liver fat while the mice responded oppositely.
There is evidence from Belury 2003 that CLA improves insulin sensitivity in humans, which in turn relates to body mass. This would imply that the rat model is more inline with human data.
The point? Not all species necessarily react identically to a given substance, and there is good support for the correlation of moderate CLA to insulin sensitivity and body mass.