I take a guess that when you say "-10dB", you mean the reflection coefficient looking into the matched circuit, and your question might be about the quality of the match.
-10dB means 10^(0.1*-10dB) = 0.1. So 10% is reflected. In the case of an LNA, adding an input matching network to transform (say) 50 Ohms input impedance to the ideal impedance for best gain from a device may also mean a poor noise figure, because the impedance for minimum noise is almost never the same as for best gain, and a trade-off compromise is chosen.
-10dB may be the best that can be managed. VSWR is between 1.9 and 2.0. I would hope for better than -14dB for an amplifier if it could be done, but this is not always possible. I have seen near -20dB. Very good designed filters can often claim -25 to -30dB in band. A accurately matched microstrip line on 0.508mm Er=3.5 substrate is shown. It still has radiation leaks and effects across its width but the match is better than -35dB.
-20dB, meaning 1% is reflected, would be considered by most to be a very satisfactory value for an LNA amplifier. VSWR would be about 1.22. Most professional microwave LNA product do not often claim VSWR better than 1.5.
These values of match, while OK for an amplifier, would not be good enough for (say) a waveguide filter carrying 500kW radar pulses with perhaps 2kW average power. 10% of such power not going out where it is supposed to might well cause a little trouble and embarrassment!
IIP3 is too low and Gain is also low.Play around the linearity
BigBoss is right. I would add that the noise figure 2.7dB is not typical of what might be expected from a LNA using devices available over the last decade. One can buy X-Band LNA that claim finished amplifier noise figure of 0.8dB At 2.7GHz, you should find it reasonably easy to manage 2.7dB NF. You probably have room to simply match for ideal gain.