Who thought this?
Cancer cells in vivo are known to have clonal evolution . This paper is specifically talking about in vitro growth on a petri dish. The problem is when a biopsy is taken it includes some cancer cells and some normal tissue. The cancer cells need to be isolated, then multiplied to have enough material to work with. This paper suggests this process results in heterogenous progeny - differing in key mutations related to immunotherapy.
(disclaimer: not anywhere near an expert, software guy in bioinformatics)
Why would anyone think they would not harbour significant levels of genetic variation? This makes it sound like they just spent 40 years misinterpreting data because they didnt do some basic algebra.