Looking at http://www.farnell.com/datasheets/2299421.pdf I don't see a gain figure just a maximum of 500µA output.
In high voltage operation, it is recommended that IBIAS remainwithin 200 μA to 1 mA. This ensures that the bias circuit is turned on, allowing the device to function as expected. At the same time, the current through the bias circuit/regulator is limited to 1 mA. Refer to Figure 19 and Figure 21 for IBIAS and V+ information when using the AD8212 in a high voltage configuration
Don't know if you can get a guard trace around it.
This is what the LT3080 datasheet says (page 10) about the set pin, which is also high Z
"With the low level current used to generate the reference voltage, leakage paths to or from the SET pin can create errors in the reference and output voltages. High quality insulation should be used (e.g., Teflon, Kel-F); cleaning of all insulating surfaces to remove fluxes and other resi-dues will probably be required. Surface coating may be necessary to provide a moisture barrier in high humidity environments
Board leakage can be minimized by encircling the SET pin and circuitry with a guard ring operated at a potential close to itself; the guard ring should be tied to the OUT pin. Guarding both sides of the circuit board is required. Bulk leakage reduction depends on the guard ring width. Ten nanoamperes of leakage into or out of the SET pin and associated circuitry creates a 0.1% error in the reference voltage. Leakages of this magnitude, coupled with other sources of leakage, can cause significant offset voltage and reference drift, especially over the possible operating temperature range.If guardring techniques are used, this bootstraps any stray capacitance at the SET pin. Since the SET pin is a high impedance node, unwanted signals may couple into the SET pin and cause erratic behavior. This will be most noticeable when operating with minimum output capacitors at full load current. The easiest way to remedy this is to bypass the SET pin with a small amount of capacitance from SET to ground, 10pF to 20pF is sufficient."
Thankfully your circuit is well above that (
200uA min vs
10uA (3080)typical, but its something to consider because I know you do automotive), which is hot and dirty. I think if you keep dropping bias current, you will get non-linear behavior, and eventually some kind of garbage. That could mean nonlinear noise modulation of gain if its set low. That might be ugly but I am not sure, and maybe with their techniques you can extend the chip functionality to lower bias current. Not a current source but still may be of use.