Do morphological observations matter and what does the future hold?
There is no doubt that interobserver variability make morphology an imprecise tool (Cserni, 1999; Dunne and Going, 2001), but there is no evidence so far that any combination of molecular attributes can be used to predict the morphological, behavioural and drug response profiles of all breast cancers. Certainly array-based studies have shown distinct expression patterns related to drug response and to survival data, but as yet do not supplant morphology in routine use. Many of the distinct morphologies described have known distinct molecular attributes; if it looks different, it probably is different. Recognition of distinct lesions is important to factor into the study of breast cancer biology, in particular the design of study cohorts and trial analysis since they may present distinct molecular targets and/or response profiles. It may be that forms of molecular profiling techniques will ultimately replace morphological analysis in tumour classification (Bell, 1999; Alizadeh et al, 20
