I think the expression may not be necessarily same among different breast cancer subtypes. Hence it is still premature to say it can be used for diagnosis
I would be very cautious before using some microRNAs for diagnosis, particularly for those miRNAs lacking sufficient data in other independent studies published so far. Next, you should consider relevant number of cancer patients, clinicopathological differences among them (as noted Xin Hong above), relevant number of control patients, validation of the method, suitable normalization, etc. Particularly, you should have a diagnostically relevant type of clinical samples (i.e., other than tissue samples). Best wishes. Ludek