Problem integrating data shot with the Optolong L-Extreme dual narrowband filter

I have imaged the Cat's Eye Nebula in two consecutive nights using the same setup with a ZWO ASI294MC Pro one-shot color camera and the Optolong L-Extreme filter. If I stack the data from the first night, the results are as expected. The results from the second night show a strange color cast when I process them even when using only the standard settings and hitting "integrate" right way:

In the first night, I imaged at -10°C - the second night at -20°C for 300s per sub. Individual subs look like this:

Flats-, Darks- and Bias-Frames don't show a visible difference (except less noise @-20°C).

I have tried both, the Adaptive Airy Disk and the Ha-OIII color algorithm - both with the same strange result for the second night's data. 

The normalization map files do differ significantly:

Other targets imaged the second night look fine and can be processed without problems.
Did someone see this behavior before and has an idea, how to deal with this? 

Thanks for any suggestions!

Phillip