This paper studies a trade-off between security (intercept probability (IP)) and reliability (outage probability (OP)) for a multi-hop decode-and-forward (DF) relaying protocol in an underlay cognitive radio network, in presence of a multi-antenna eavesdropper. In the considered protocol, all primary and secondary terminals are equipped with multiple antennas, and they employ transmit antenna selection (TAS) (at transmitter sides) and selection combining (SC) (at receiver sides) techniques to enhance the system performance. Relying on channel state information (CSI) of the primary-to-secondary interference links known or unknown, two efficient TAS/SC techniques are proposed for the secondary networks. Moreover, operating on the underlay spectrum sharing mode, the secondary transmitters including source and relays must adapt their transmit power so that OP of the primary network is not higher than a pre-determined threshold. Under impact of Rayleigh fading and co-channel interference, the end-to-end (e2e) OP and IP are expressed by exact closed-form expressions that are verified by Monte-Carlo simulations. Then, both simulation and theoretical results are presented to show the OP-IP trade-off.