I don't think this is true. "Only" a quarter of people that try heroin become addicted. Similar numbers with alcohol (15% of people that try alcohol develop a problem with it), but given alcohol is more widespread, there is some selection bias because only a subset of people try heroin, whereas nearly everyone tries alcohol. I would not be surprised if the overall addiction risk is very similar between alcohol and heroin.
I think it's much more likely some sort of genetic trait underlying it.