AD9162 NCO Leakage

On a custom AD9162 board, measuring NCO "leakage" of about -41dBm.   Setup :  Sending a 50us pulsed 12MHz tone.  NCO is programmed to 1200MHz.  Below is the output. Pulsed tone at 1212MHz and a spur at 1200MHz (NCO frequency).  During deadtime (or when the pulsed tone is not transmitted on the JESD Lanes) all zeros are fed to the DAC. 

I reconfigured the FPGA to transmit all zeros to the DAC and still get the spur at -41dBm

Is there a reason for this spur?