Since this is more of a programming question,
I'm going to use "C notation" for the hex values.
Combining 2 bytes to get a 16-bit integer:
theInteger = (upperByte << 8) + lowerByte;
upperByte = 0xfc
lowerByte = 0x92
theInteger = (upperByte << 8) + lowerByte
= (0xfc << 8) + lowerByte
= ( 0xfc00 ) + 0x92
dissecting a 16-bit integer into two bytes:
upperByte = theInteger >> 8;
lowerByte = theInteger & 0xff;
theInteger = 0xfc92
upperByte = theInteger >> 8
= 0xfc92 >> 8
lowerByte = theInteger & 0xff
= 0xfc92 & 0xff
I suspect the "& 7f" in that formula should be something more like "& ff".
If it were really "& ff", then reversing the formula is easy -- subtract one from the integer, and then dissect the integer into two bytes.
However, if the "& 7f" is correct, then it's a little more complicated -- no matter what initial byte values we have, that formula (with the "& 7f") will never generate certain integers. In particular, it will never generate the integer 0x8000.
Also, there are two different byte pairs that that formula (with the "& 7f") will generate the same final value. In particular, the byte pair upperByte=0x4c and lowerByte=0xff gives the final integer value 0x7f4d, the same as upperByte=0x4c and lowerByte=0x7f.
Is that really what you wanted?