Hello Montadar my friend, a post from you after a long time. You must be busy.
Here is what I know about the PAPR problem with DS-CDMA.
As you know we have quadrature modulation schemes such as QAM being used in most communication systems these days, so in those systems, it can happen that the I channel and the Q channel can combine constructively to produce a signal which is of a higher value of power as compared to average power of cdma symbols.
This, I think leads to a higher value of PAPR and in order to remove this, both channel are randomized using two different scrambling sequences (CDMA2000, I think)
I think you get the idea, but please verify this as I am kind of rusty on this matter.