But not in all cases. Using “DR” modes typically alters the ISO implementation and substitutes the analog amplification before the ADC, for brightening in the jpeg processing after the ADC to improve the highlight headroom. Whatever ISO Implementation is being used, it won’t usually determine the brightness of the processed raw file anyway, so while whatever ISO value you shot with might affect the raw file in some way, it really doesn’t apply to the raw file in that way (which is what the ISO standard is all about).
DR modes underexpose for a given ISO (by lowering ISO internally) and then brighten the image while preserving highlights. Even in DR modes, increasing ISO can cause raw clipping. In my book, that means that ISO settings apply to raw files.
The only situation where ISO does not apply to raw files is when the ISO setting causes only a tag change (some cameras at higher ISOs, ISO bracketing in some cameras) and leaves the raw data unchanged.
In every camera I’m familiar with (I shoot mostly Fuji), the DR modes operate by altering the ISO implementation to reduce the pre-ADC amplification by 1 or 2 stops (or the actual sensor exposure), often resulting in no “normal” analog pre-ADC amplification at all. This is how highlight headroom is increased, with the 1 or 2 stop “darker” raw files being compensated for in the jpeg processing (with a softer curve in the highlights). Unless your raw editor is compensating automatically (Lightroom and Capture One do via metadata tags), your raw files would always appear 1 or 2 stops darker at default import settings than raw files produced without DR modes (or the jpegs), despite being shot at the same ISO values.
Edit: to clarify, I completely agree that the choice of ISO value can, and usually does, affect the RAW file, but the ISO value designation does not apply to the RAW file.
Even as measurement charts published by photonstophotos.com will invariably show that with an increase of iso-value input the base-iso level DR of a sensor will decrease (dual-gain induced dents included), I'm still somewhat confused over the question whether this holds true for the raw data file as well, given an "iso-invariant" sensor. If brightnesswise it doesn't go more sensitive with an iso increase, then why would/does it go less sensitive when it comes to DR-scope? The main reason why I'll try to keep a low iso-setting and apply ETTR to some feasible extent – while sticking to a DR100 stance – is to retain as much as I can of the sensor's base-iso DR-range level. Hoping that I'm not all too deceived over such prudent care.
If the marks on the graph indicate normal processing (filled in circles) then gain is applied to the raw values as the ISO goes up, and you lose headroom.
Here's an example of a camera that stops applying gain to the ISO values at a certain ISO.
Define "expanded ISO settings". I've seen cameras like the GFX 50S where the ISO settings where the JPEG brightness and raw file numbers are disconnected at what appear to be normal ISO settings.
I have never seen the niceties of ISO implementation (dual conversion gain, programmable gain amplification, digital gain, raw developer gain, or their implications for photography) covered in a camera manufacturer's user manual.
The problem is that the description in many manuals describes an impossibility. It cannot be correct, yet their technical authors wrote it nonetheless.
From a manual:
"Sensitivity can be set higher than the default setting (ISO 200 equivalent) when taking photographs in low light conditions"
"The default setting of ISO 200 equivalent is recommended in most circumstances. Photographs taken at other settings may include a slight amount of noise."
That's very useful :)