Subtyping Bladder Cancers: Biology vs Bioinformatics

Recent studies employing whole transcriptome expression data and unsupervised analytical methods concluded that bladder cancers can be grouped into basal and luminal molecular subtypes that have implications for prognostication and predicting response to therapy (1–7). Here Mo and colleagues used a supervised approach, based on their knowledge of biomarkers associated with normal differentiation, to assign bladder cancers from public data sets into two major molecular subtypes—basal and differentiated (8). Using this 18-gene biological classifier, they showed that the basal tumors were associated with shorter survival, consistent with their own previous observations (9) and those of the other groups (1–7). However, the relationship between basal subtype membership as defined by the 18-gene signature and poor clinical outcomes was stronger than was observed with some of the other classifiers (1–3), leading them to conclude that their classifier was a better tool for assigning bladder cancers to the basal subtype ("biology trumps pure bioinformatics"). While this may be true, it does not necessarily follow that the most accurate bladder cancer classifier will be the one that consistently shows that basal cancers are the most clinically aggressive. Recent studies concluded that patients with basal tumors obtained the most benefit from neoadjuvant chemotherapy (7,10), so the relative "aggressiveness" of basal tumors within a given cohort could be influenced by the relative proportion of chemosensitive basal tumors that are in it (and the MD Anderson "discovery" and The Cancer Genome Atlas (TCGA) cohorts contained some tumors from patients treated with neoadjuvant or adjuvant chemotherapy) (1,2). Immune checkpoint blockade will probably also complicate molecular subtype associations with survival (11).

http://ift.tt/2ECkBDt