Heres how PR works (from my understanding):
The average damage, kills, and win rate for Gearing respectively are 43k, 0.72, 48.5% (pulled from wowsnumbers)
First the damage you dealt in that game is converted into a ratio, kind of like an "effectiveness" rating. Basically, it's comparing you to the average stats of players.
rDmg = 15k/43k = 0.348
rWins = 1/0.485 = 2.062
rFrags = 0/0.72 = 0 (I'm going to assume you got 0 kills in this match)
Next these values are normalized (and if negative, assumed to be 0)
nDmg = (rDmg-0.4)/(0.6) = -0.08667 = 0
nWins = (rWins-0.7)/(0.3) = 4.539
nFrags = (rFrags-0.1)/(0.9) = 0
***(Pay attention to the zero's found above)
Finally, PR is calculated with this formula:
PR = 700*nDMG + 300*nFrags + 150*nWins = 4.539*150 = 681
In this case, only the "value" you get from winning contributes to your PR. Damage and Kills give 0 to your PR for that game
Evidently, the PR calculation is pretty obscure but clears up a little with explanation :)
Anyways, just a couple things to note here:
Because kills are always integer values, and usually average kills is in the range of (0.5,1.5), having 0 kills straight up contributes nothing to your PR rating, since after normalization, the value of nFrags will likely be 0
DMG is valued very highly, as the other comments have pointed out. Quantitatively, the coefficient for nDMG is 700, which is over twice the coefficient for kills (300). In other words, if in one game you have a damage value that is a certain threshold below the average damage, your damage that game will contribute also nothing to your PR rating.
Wins share the same issue as kills: Within one game, the value for Winrate can either be 0 or 1 (loss or win). As such, if you had lost that one game in gearing, the value of nWins would likely have also been 0, meaning that your PR could have been 0 overall. That is very scary if you ask me since evidently, you were not AFK that match.
Lastly, don't value PR over one game too much. As I've explained above, the values for your personal average wr/kills is too volatile with low sample size of games, since a 0 value can easily negate part of your PR calculation.
TL;DR
Having 0 kills or 0 wins greatly affects your PR, since it negates part of the PR calculation entirely. Don't worry about PR from a low sample size of games, since it can be very low despite you doing valuable team play in game.