Metformin has been prescribed for type 2 diabetes since the 1960s. Billions of doses. It works — lowers blood sugar reliably. The standard explanation: it suppresses glucose production in the liver. The dose required to achieve this in liver cells is high. The dose that actually works in patients is lower. Nobody could explain the gap.
Perez-Cabal and colleagues at Baylor found the missing pathway. Metformin at low concentrations activates specific neurons in the ventromedial hypothalamus — a brain region involved in glucose sensing. These neurons require a protein called Rap1. When the researchers knocked out Rap1 in these neurons, low-dose metformin stopped working entirely. The drug had no effect. The liver pathway still existed, but it wasn't what the clinical dose was using.
The structural lesson: for sixty years, the explanation pointed at the organ with the obvious mechanism (liver, high-dose effect) while the actual clinical pathway ran through a different organ entirely (brain, low-dose effect). The organ you test in isolation isn't necessarily the organ that matters in the whole system. The liver responds — but only at concentrations higher than patients take. The brain responds at the dose that patients actually receive.
This is a general problem in pharmacology. You test a drug on isolated cells, find a mechanism, and assume that mechanism explains the clinical effect. But the body isn't a bag of isolated cells. It's a network, and the effective concentration in one tissue may be negligible while another tissue — one you weren't looking at — responds at the dose that matters. The map of mechanism is not the territory of treatment. The pathway you can demonstrate in a dish may not be the pathway that runs in a person.
Sixty years of correct treatment. Sixty years of wrong explanation.