Design Principles of Life, Disease, and Aging

# Hormone circuits: robustness, diabetes, and slow physiological memory *Evidence guide: Mathematical model results and systems-medicine hypotheses, compared with physiology data. No single mechanism explains all diabetes.* Start with the insulin and glucose loop. Glucose rises after a meal. It stimulates beta cells in the pancreas to secrete insulin. Insulin makes tissues such as muscle and fat take up glucose, so glucose falls. A negative feedback loop, acting over minutes to hours. Here is the puzzle. Insulin sensitivity, how strongly tissues respond to insulin, differs widely between people and changes with weight, pregnancy and activity. Yet most insulin-resistant people still keep fasting glucose near about five millimolar. A simple glucose–insulin model cannot explain that. In such a model, lower sensitivity simply raises glucose. The missing piece is slow. Glucose also controls the growth and removal of beta cells. Over weeks to months, the functional mass of beta cells expands when glucose runs high, and shrinks when it runs low. So there are two timescales. Fast: hormone secretion, minutes to hours. Slow: changes in gland mass, weeks to months. In the model, beta-cell mass B changes as d B over d t equals B times growth minus removal, both set by glucose. B can only stop changing when glucose sits at the point where growth equals removal. That locks the long-term glucose set point, a form of integral feedback. Karin and colleagues showed in 2016 that this also produces dynamical compensation. If insulin sensitivity halves, beta-cell mass and insulin roughly double over weeks, and the glucose response to a meal returns to the same shape. Only insulin reveals the change, a pattern also seen in data comparing insulin-resistant and insulin-sensitive non-diabetic people. But law two applies. Beta-cell functional mass cannot grow without limit. When compensation saturates, further insulin resistance raises glucose. In the framework, this corresponds to prediabetes. Insulin resistance is one route, but not the only one. Law three adds a twist. A mutant beta cell that over-reads glucose would sense high glucose, keep dividing, and could take over, pushing glucose dangerously low. Karin and Alon proposed in 2017 that glucotoxicity helps here. Beta cells die not only at low glucose, but also at very high glucose: a U-shaped, biphasic response. A strongly mis-sensing mutant perceives very high glucose, and is removed. The cost is a second, unstable fixed point. If average glucose stays above it for long enough, high glucose kills beta cells, less insulin raises glucose further, and a vicious cycle sets in, resembling progression toward insulin-dependent type-two diabetes. This is a systems-level mechanistic framework from the group. Type-two diabetes has many contributing factors, including genetics, lifestyle and beta-cell stress pathways, and this model is not the sole accepted explanation. In the thyroid axis, an analysis of Israeli medical records showed that TSH can take many weeks to normalize after thyroid hormone levels are corrected. The group's model explains this delay by slow recovery of thyroid and pituitary gland mass. Sources: [sysmed-notes](https://www.weizmann.ac.il/mcb/alon/courses/system-medicine-2022-2023), [topp2000](https://doi.org/10.1006/jtbi.2000.2150), [karin2016](https://doi.org/10.15252/msb.20167216), [polonsky1988](https://doi.org/10.1172/jci113339), [karin2017](https://doi.org/10.15252/msb.20177599), [korem2022](https://doi.org/10.15252/msb.202210919).