Hello, it depends on the type of assays you want to do or hypotheses you want to test. Although HepG2 is already tumor, it will still evolve over generations, esp. epigenetically.
Although morphological analysis is an economical and fast way of assessing passage limits, it is not always accurate and in fact tends to *underestimate* genomic or epigenetic changes (due to our mental tendency to save N2 cell stocks).
Some cell types like HUVEC and endothelial lines can only be passaged for a few generations and then start to lose original cell properties including morphology.
***
To really determine what're the limits, it's better to have some biomarkers at the molecular level, so that you can check from time to time, in monitoring the integrity of your model. As for what biomarkers (e.g. by PCR, or even Western blot), again, it depends on what models you are investigating. Reviews and current papers by peers in your subfield should give some hints.
I have worked with macrophage lines, and in general, we try not to use beyond 6 consecutive passages. I would say, for cancer cells, 8-10 passages would be max upper limits of passage.
Consider preparing a very large pool of N2 stocks of low pasage cells initially. We do up to 50-70 aliquots of p2, p3 cells.
HepG2 has proved to be robust enough in our hands, for qPCR studies. Have you changed anything in the culture conditions (new vendor, etc)? Also, morphological changes are very subjective, I would try look for some measurable markers associated with you application.