I am analysing an RNA-seq dataset with the below given design in DESeq2

While it was fairly straightforward to get results for Drug vs DMSO within cell lines. I also need to look into the effect of drug common across both the cell lines. I can see two ways of doing this:-

1. Take the intersection of up and down regulated genes from the individual Drug vs DMSO comparisons
2. Use my design as ~CellLine + Treatment and then extract results for Drug vs DMSO. This should in theory give me genes that are significantly up/down across both the cell lines.

The approach [2] should be more sensitive. However taking a look at plotCounts for some of the top significant genes quite a few genes seems to come up with ~similar counts in one of the cell lines and a significant change in the other. Albeit this happens mostly when counts are low in all samples but still this doesn't seem intuitive. Is this a correct way to get the results or should I stick to approach [1]

If you require it is DE in both cell lines, (1) is the better way to go.