How you determine it 'no effect'.

Have you try to lower the power supply voltage (a little bit lower than vrst) when you run the test and then recover the power again?

I am wondering if below part of code would work (check you should left shift 6 or 5):

VDM0CN & (0x01 << 6);

Please be noted that customer should take any risk caused if they don't follow our guidance.

You may try to disable the supply monitor reset function first and have a try (that was not recommend, especially when you want to write flash. flash operation like write/erase may cause more critical issue/disaster if you violate Silabs guidance).

https://www.silabs.com/community/mcu/8-bit/knowledge-base.entry.html/2004/02/15/flash_corruption-u4VR.html

Do you have any reason behind why you care the last current ADC sample (one sample only)?

I would suggest you run the example code first (may not need any change to run it on your device).

If you have problem to run it on your board, could you upload the schematic?

It would be same in EM0 ,EM1 and EM2 mode.

what is your measured result? 

I don't have time to test this. I encourage you test this out by yourself if you have EFR32xxx or radio board.

Anyway, I don't encourage you design like this because it evidently violate the spec (DS), you should take risk by your own if you violate the spec.

You could take a look at KBA below (I am not sure if this cause permanent damage to BGM123):

https://www.silabs.com/community/mcu/32-bit/knowledge-base.entry.html/2015/10/19/exceeding_gpio_absol-wkLF

I am also wondering if the GPIO could be pull/latch high or low without a valid IOVDD.

How you determine it 'no effect'.

Have you try to lower the power supply voltage when you run the test?

I am wondering if below part of code would work (check you should left shift 6 or 5):

VDM0CN & (0x01 << 6);

Please be noted that customer should take any risk caused if they don't follow our guidance.

How you determine it 'no effect'.

Have you try to lower the power supply voltage when you run the test?

I am wondering if below part of code would work (check you should left shift 6 or 5):

VDM0CN & (0x01 << 6);

For debug mode, the MCU may not enter some low power mode.

Try to stop the debugging and then measure the current.

You may try to disable the supply monitor reset function first and have a try (I suppose it should work as you expected. that was not recommend, especially when you want to write flash. flash operation like write/erase may cause more critical issue/disaster if you violate Silabs guidance).

https://www.silabs.com/community/mcu/8-bit/knowledge-base.entry.html/2004/02/15/flash_corruption-u4VR.html

Do you have any reason behind why you care the last current ADC sample (one sample only)?

Sorry, I guess you are using the BRD4173A radio board board to run the test. I have though you may use a custom board made by yourself to run the test.

I don't have that board. I have a BRD4162A, i tried the BLE 'soc-empty' project (for BRD4173).

it could run on BRD4162A.

If I understand it correctly, you would like realize something like below picture showed: