Just remember junction temp doesn't have a whole lot to do with CASE temperature only " steady " temperature rises come into play there... the data sheet will have to J to C coefficient, but an instant temp rise will still blow it to smithereens before the case even warms...
Ive had that happen to mosfets , cracking like that under failure... your lucky the energy in those capacitors usually make em go boom...
I don't think the 5 degree change SHOULD have made any difference, unless you were at the limits... which sounds like you were not... maybe re-set it back to 25 and get some readings, a no contact temp probe are not expensive off of fleabay, check case temperature ( not heatsink ) under load and post it here, can back track the junction temp using datasheet, will only give steady temp, not instant, but give you some more info to track the issue down...