in cell A, do you have a poor CSSR as well ?
the co channel adjacency could be a major problem, especially if ARFCN 105 is the BCCH freq of cell A or the BCCH freq of the neighbour cell.
if the radio KPI (cdr, cssr, ho out succ, tch duration avg, tch erlang) are fine, then I doubt the problem comes from parameters or radio.
I would focus on hw.
Try to lock each TRX one by one, for few hours. See if when one TRX is locked, the inc ho gets better.
Also, do drive tests, to detect feeder swap. (but that would also be seen on the radio kpi !)