I'll just throw this one out there as I found it while looking for something else:
Metformin paradoxically worsens insulin resistance in SHORT syndrome
SHORT syndrome is a (very, very rare) genetic failure of insulin signalling at the PIK3 regulatory subunit 1 level. Insulin binds to its receptor but signalling fails due to a single downstream gene defect giving severe insulin resistance. People with this syndrome are, needless to say, very thin. They maintain normoglycaemia using a very high level of insulin which does, given a high enough concentration, produce normoglycaemia. There doesn't appear to be any problem with secreting insulin from the pancreas. In fact, to overcome the failure of insulin signalling during her OGTT the patient's pancreas produced a plasma insulin of 688mIU/l*. In new money that is just under 5000pmol. As in roughly ten times what you might expect. Severe, but not quite insuperable, insulin resistance.
*The paper specifies insulin in mIU/ml. I'm assuming this is a typo or a font failure because clinical insulin concentrations are usually expressed as microIU/ml or mIU/l. Obviously if it really is 688mIU/ml the concentrations will be 1000 times those quoted. Gulp. People really should use the SI system.
Back to the patient. The obvious thing to do is to give an insulin sensitising agent, number one in popularity nowadays being metformin.
This turned out to be a bit of a boo boo.
During an OGTT under metformin the patient's insulin resistance worsened and mild hyperglycaemia ensured but this was despite a plasma insulin concentration which was simply too high to measure. The lab could measure up to around 7000pmol (pax typos) and it looks like the curve went MUCH higher than that.
That is despite metformin's predictable and recently found ability to suppress insulin release from mouse islets.
From the Protons perspective metformin blunts insulin signalling via blockade of mitochondrial glycerol-3-phosphate dehydrogenase. Its beneficial effects to increase insulin sensitivity come from reduced exposure to insulin signalling in peripheral cells. The peripheral cells of a person with SHORT syndrome barely see insulin signalling at all even without the metformin. You would hardly expect further blocking any residual insulin signalling to help matters. It doesn't.
It's the sort of paradox which only happens when you are in the wrong paradigm of metformin's mechanism of action. Might have been a chance to make progress...
Addendum: The lady in question did not seem to enjoy her experience with the medics too much:
"As we intended to check the effects of this approach, an extended 75 g OGTT was performed on metformin 4 days later. This showed dramatic and paradoxical worsening of insulin tolerance with insulin concentrations above the upper assay detection limit (Fig. 1b). Metformin treatment was discontinued. She was discharged home on Dydrogesterone and vitamin D supplementation. We planned to perform investigations on other family members, and particularly on her younger brother, but despite several reminders they failed to attend clinic appointments as well as declined admission to the hospital".