“…While the dichotomisation into two subtypes has the perceived advantage of simplifying biomarker and functional studies, there is increasing evidence that cells with classical and basal-like features co-exist in the same tumour ( Figure 2B ) ( Puleo et al, 2018 ; Porter et al, 2019 ; Chan-Seng-Yue et al, 2020 ; Hwang et al, 2020 ; Juiz et al, 2020 ; Nicolle et al, 2020 ). This evidence has been generated from analyses of both human tissues ( Puleo et al, 2018 ; Porter et al, 2019 ; Chan-Seng-Yue et al, 2020 ; Hwang et al, 2020 ; Raghavan et al, 2021 ) and ex vivo cultures ( Porter et al, 2019 ; Juiz et al, 2020 ; Nicolle et al, 2020 ; Raghavan et al, 2021 ) and implies that molecular classification systems should account for this phenotypic heterogeneity for a better prediction of patient outcomes. Accordingly, Nicolle and others have recently shown the benefit of classifying patients based on a continuum of phenotypes rather than on two non-overlapping subtypes ( Nicolle et al, 2020 ).…”