You say that the low light ISO score of the K5IIs is quite a low score compared to other APS-C sensor DSLRs available. It is actually quite high, with a difference of only the 3.8% from the D7100.
I think the data may be misinterpreted.
How many cameras do you test for each model? Do we have a mean, median and variance for the values? If you only test one, then it could be within the statistical variation of the sensor.
Disclaimer: I do not have a Pentax and do not work for Pentax.