Up- and down-regulated in relation to a non-cancer control? So it's higher expressed than control in one experiment and shows less expression than control in another experiment? Or are you talking about expressed in one experiment and not expressed in another? Are your experiments from the same cancer tissue? Same experiment type (both qPCR, RNA-seq, ...)? Or are they from the same cancer, but from different persons? Some more information would be great here.
But even without that information, I would think that it can happen... think about environmentally affected gene regulation, e.g. stress-induced expression changes in microRNAs. It does not automatically mean that the change in expression is because of the cancer.
Up- and down-regulated in relation to a non-cancer control? So it's higher expressed than control in one experiment and shows less expression than control in another experiment? Or are you talking about expressed in one experiment and not expressed in another? Are your experiments from the same cancer tissue? Same experiment type (both qPCR, RNA-seq, ...)? Or are they from the same cancer, but from different persons? Some more information would be great here.
But even without that information, I would think that it can happen... think about environmentally affected gene regulation, e.g. stress-induced expression changes in microRNAs. It does not automatically mean that the change in expression is because of the cancer.
Your question is very confusing, is it pertaining to expression in cancer/non-cancer tissue from same subject or multiple subjects. without the study design the community will give giving you suggestions which may not be of help.
Yebo. Everything is possible, if not in the way it is than in the way it looks for sure. There is not a single step in the process which you couldn't screw. There were a number of possible reasons already suggested here for why this could be observer as an artifact. The reason might be also as simple as different count of cancer cells in the samples from different cancer studies or even in the same study, same cancer, different replicate samples.
There are also reasons to why this could be a real observation, not an artifact. Studies in cancer biology rarely make any account of time. I'm about to submit a manuscript with analysis of almost all known miRNA expression profiles over time. The bottom line is all of these miRNA are oscillating following circadian and/or metabolic cycles. Samples taken at different time may capture the same miRNA at higher or lower parts of the waveform. Cancer is also associated with profound shifts in metabolism, which may also affect the phase of miRNA oscillation.
The main issue with studying cancer is that each cancer is probably unique, and also different actual tumours within the same individual are probably different as well. If you read the "Hallmarks of Cancer" paper by Hanahan and Weinberg, they propose that certain phenotypes are needed for a normal cell to generate a malignant tumour in an individual or animal. One example would be that they are resistant to apoptosis, however there are many ways a cell can develop this trait, but essentially it is the trait that is important so there will be differences in the way that particular cells arrive at that trait. Additionally given that a malignancy is usually the result of 12 or more genetic changes then this would mean that there will be a great deal of heterogeneity between tumours. This is the clinical picture because if all tumours of one type were all the same they would be a lot easier to treat. That is not to say that there are not common mechanisms, which there are, for example ras mutations, it is just not that cut and dried.
It is possible for the same miRNA to be up-regulated and down-regulated in your cancer, particularly if it has nothing to do with causing the disease itself or if it is a consequence of different stages/time-point of the disease.