An amplifier shouldn't be expected to generate spurious signals at all. You didn't tell the absolute signal level however, and you should also show
the generator spectrum for comparison. I would rather suspect a bad performing PLL generator than the amplifier sourcing the spurious. It
should also be clarified, that the generator output spectrum can't be affected by connecting a load. (In a PLL design, it can, of course).
The analyser shows a span of 1 MHz, not 100 MHz as said in the post. As another point, the display looks like from a TTI PSA
analyzer. I'm not aware of a TTI instrument covering the 6 GHZ range. Is there other, unsaid equipment involved in the
measurement setup?