Deriving Fasting Glucose-Insulin Steady States

Why Derive a Fasting Equilibrium?

A fasting glucose and insulin measurement is a snapshot of a system with continuing fluxes. Glucose is produced and removed, while insulin is secreted and cleared. At a steady state, the concentrations do not change, but the underlying fluxes need not be small: production must balance removal, and secretion must balance clearance. Writing those balances explicitly gives a mechanistic way to connect fasting concentrations with the processes that maintain them.

The derivation below uses a deliberately compact model. It is useful for tracing equilibrium shifts, not for representing every physiological control or predicting every fasting profile. Production and disposal are aggregate rates, consistent with the earlier material on organ-resolved glucose mass balances and hepatic glucose production. The secretion term is a local approximation to glucose-responsive secretion.

Model and Equilibrium Conditions

Let glucose production be \(P_0/(1+I/K_H)\). Here \(P_0\) is the production rate in the absence of insulin suppression, and \(K_H\) is the insulin concentration giving half of that rate in this chosen suppression function. Let glucose disposal be \(D_0+k_GG+k_II\): \(D_0\) is insulin-independent disposal, and the other terms represent glucose-dependent and insulin-dependent disposal. These terms are expressed as glucose concentration-equivalent rates, in mg/dL/h.

Let total insulin secretion be \(S_b+q(G-G_{\mathrm{ref}})\), a local linear relationship over the glucose range being considered. \(S_b\) is the secretion rate at reference glucose \(G_{\mathrm{ref}}\), and \(q\) is the local glucose responsiveness. Insulin clearance is \(k_{\mathrm{cl}}I\). Secretion and clearance rates are in mU/L/h.

The assumptions are that the subject is in a fasting interval with no meal-derived glucose input; the listed processes capture the dominant balances for the question at hand; the parameters remain fixed over the interval; and the linear secretion approximation is used only where its predicted secretion is nonnegative and physiologically plausible. The positive parameters \(K_H\) and \(k_{\mathrm{cl}}\) ensure positive suppression and clearance scales. The equations also assume a common time unit and that the glucose fluxes have already been scaled to concentration-change rates.

The second balance is about total secretion. The glucose-responsive contribution is added to secretion at the reference glucose; it is not a replacement for that component. Accordingly, insulin clearance must be compared with basal-plus-responsive secretion, not with \(S_b\) alone.

Reducing the Coupled Balances to One Equation

The insulin balance can be rearranged to express insulin as a function of glucose:

Substituting this relation into the glucose balance leaves one equation in one unknown. Define the net glucose balance along the insulin-equilibrium curve as

A steady state is a root \(G^*\) of \(F(G)=0\), with \(I^*=I(G^*)\). If \(q>0\), then \(I'(G)=q/k_{\mathrm{cl}}>0\). For positive parameters, the derivative is

Thus, wherever the secretion relationship is valid and insulin is nonnegative, \(F\) decreases strictly with glucose. If it changes from positive to negative over the admissible glucose range, it has exactly one root there: continuity gives existence, and strict decrease gives uniqueness. A mathematical root outside the valid range of the secretion approximation is not automatically a meaningful physiological equilibrium.

How Parameter Changes Shift the Equilibrium

The scalar equation makes directional predictions possible. Since \(F'(G)<0\), a parameter change that raises net glucose balance at the old equilibrium generally requires a higher glucose concentration to restore balance. For example, increasing \(P_0\) raises production at fixed concentrations. Implicit differentiation gives a precise result:

Thus higher unsuppressed glucose production raises equilibrium glucose in this model. Because \(I'(G)>0\), it also raises equilibrium insulin. Similarly, increasing \(K_H\) weakens suppression at fixed positive insulin: \(\partial F/\partial K_H>0\), so \(dG^*/dK_H=-(\partial F/\partial K_H)/F'(G^*)>0\). Reducing \(k_I\), the coefficient for insulin-dependent disposal, also reduces glucose removal at fixed insulin and shifts the equilibrium toward higher glucose.

These directions depend on the stated model and on holding the other parameters fixed. They do not establish that a measured association between glucose and insulin was caused by any one parameter change. Other mechanisms can change together in an individual.

Stability, Diagnostics, and Alternatives

An algebraic equilibrium need not be stable in every possible model. For the dynamics above, the Jacobian at an equilibrium has diagonal entries \(-k_G\) and \(-k_{\mathrm{cl}}\), an upper-right entry \(b=-P_0/[K_H(1+I/K_H)^2]-k_I<0\), and a lower-left entry \(q>0\). Its trace is negative and its determinant is \(k_Gk_{\mathrm{cl}}-bq>0\). For a two-variable continuous-time system, these conditions imply that both eigenvalues have negative real parts. Therefore the equilibrium is locally stable under these model assumptions: small disturbances decay near the equilibrium. This conclusion is local and does not guarantee stability in a more detailed model with delays, additional feedbacks, or changing inputs.

When calculating an equilibrium, check the following before interpreting it:

• Verify both residuals. Substitute the unrounded concentrations into both balances and confirm that production matches disposal and total secretion matches clearance, within rounding error.
• Check admissibility. Reject negative concentrations and roots that imply negative secretion or lie outside the range where the linear secretion approximation is credible.
• Check units. Each side of a balance must have the same rate units. A concentration cannot be added directly to a flux without a parameter that converts it to a rate.
• Check the operating condition. If glucose or insulin is still trending, or fasting inputs are changing, the measurement may not represent a steady state.
• Check numerical behavior. A numerical root finder should be tested from more than one starting point or bracketed over the admissible range. In this monotone model there is at most one admissible root, but a more complicated nonlinear model may have multiple roots.

A direct algebraic solution is convenient when substitution produces a manageable polynomial, as in the examples. A numerical root solver is usually more practical for nonlinear secretion or production functions; it should report a root with a small balance residual and within the model’s valid range. Dynamic simulation is an alternative when the question concerns how the system approaches a steady state, rather than only where the balances vanish. A single fasting pair \((G,I)\) does not generally identify all the parameters: different combinations of production, disposal, secretion, and clearance can satisfy the same two balances. Independent data or perturbations are needed to distinguish those mechanisms.

Common Mistakes and Model Limits

• Equating basal secretion alone with clearance. The wrong reasoning is that the insulin balance is \(S_b=k_{\mathrm{cl}}I\). The correct balance uses \(S_b+q(G-G_{\mathrm{ref}})=k_{\mathrm{cl}}I\); basal secretion is only one part of total secretion.
• Balancing production against only one disposal term. At equilibrium, production equals the sum \(D_0+k_GG+k_II\), not just insulin-dependent disposal.
• Treating an algebraic root as automatically physiological. A root must be nonnegative and consistent with the assumptions and operating range of the model.
• Reading a directional prediction as a unique explanation. Higher glucose at equilibrium is compatible with several parameter changes; the equilibrium alone cannot identify which process changed.

This compact model omits many sources of variation, including time-varying inputs, additional hormonal regulation, and nonlinear tissue responses. Its value is narrower: it makes explicit what must balance at fasting equilibrium and shows how a stated parameter change shifts that balance. Use it as a mechanistic simplification, not as a complete account of fasting physiology.