I want to add something to this discussion. If u get detailed stats on interMSC HO, analyse them. For eg. if the HO rate from MSCB to MSCA is poor then check for the outgoing HO attempts of MSCB and incoming HO attempts of MSCA. With this u can come to an approx conclusion in which MSC the analysis has to be made. Normally in one node the failure will be high. If success rate is high in MSCA i/g it means there might be problem in MSCB where the definitions of cells of MSCA are not proper.
A good solution is to check all the internal & external neighbor definitions in all MSC and BSC nodes.