Beta-cell mass and function
β-Cell mass can be accurately evaluated only through autopsy studies [130,131]. Complex in vivo tests aiming at measuring maximal secretory capacity combining different stimuli have been proposed [94,132] as an alternative. However, their ability to discriminate between defects in function and mass, as well as their feasibility, is a matter of debate . As thoroughly reviewed by Robertson , the acute response to an i.v. glucose bolus in normoglycemic subjects and the acute response to an i.v. arginine bolus in hyperglycemic subjects do correlate with β-cell mass. However, the severity of the defects observed with these tests in T2DM  contrasts with the relatively preserved mass size observed at autopsy suggesting that in vivo investigations cannot distinguish between defects of mass and function. More recently, external detection of β-cell mass through metabolic tracers has been attempted but, up to now, the reliability of these methods remains limited .
Autopsy studies have consistently demonstrated that the whole body β-cell mass in a normal adult man is approximately 0.9 g and displays a wide inter-individual variability (95% CI: 0.5–1.3 g). Assuming 1 ng per cell and a fasting insulin production of 75 pmol min−1 m−2 this means that glucose homeostasis in man relies upon “only” 900,000,000 cells, each producing approximately 90,000 molecules per minute and five to sixfold more in the stimulated condition .
Obesity is a marginal factor in determining total β-cell mass as it accounts for a 0.2 g increase for any 10 units of BMI. The dimension of this cell pool expands slowly and gradually from 0.2 to 20 years of age  remaining relatively stable thereafter being only marginally influenced by aging [130,131]. In rodents pregnancy is associated with β-cell proliferation until mid gestation to drop to the nonpregnant stage after parturition .
The reasons for the large individual variability in β-cell mass in adults ranging from 500 to 1500 million cells is not fully understood but a genetic influence cannot be ruled out as suggested by the observation that several type 2 diabetes susceptibility genes are involved in β-cell development (PDX1, PTF1A, or HNF1B) and apoptosis (INS, HNF4A, EIF2AK3, WFS1, or FOXP3) .
The workload imposed on each single cell will vary in inverse proportion to the initial asset and, as a corollary, the β-cell functional reserve must be wide enough to guarantee normal glucose homeostasis in most individuals. As evident from autopsy [130,131] and pancreatic surgery  studies, this functional reserve is approximately twofold, with hyperglycemia usually ensuing for reduction of >50% of the original β-cell mass. After hemi-pancreatectomy similar glucose values are achieved in response to either oral or i.v. glucose in spite of 50% lower insulin response suggesting that the reduction of β-cell mass translates into—or mimics—a diffuse and severe β-cell dysfunction. The small impact on glucose tolerance is likely to depend on the simultaneous reduction of glucagon  and/or increase in insulin sensitivity.
Interestingly, in a similar experimental setting the C-peptide:glucose ratio was found to correlate with the residual β-cell mass . However, the explained variance was less than 50% and it was present only when the subjects with diabetes, those with impaired and those with normal glucose tolerance were evaluated altogether.
These observations suggest that a reduction in β-cell mass can translate into impaired function. However, such an effect tends to become evident only in extreme conditions (i.e., existing insulin resistance), whereas, in a more physiologic setting, the relationship between mass and function is not present or is rather loose.
The evidence of heterogeneous, and sometimes selective, β-cell dysfunctions in subjects with mild impairments in glucose homeostasis , points to a minor impact of β-cell mass on physiologic glucose homeostasis. This is also supported by the demonstration that mild-to-severe degrees of β-cell dysfunctions can be reversed by intensive blood glucose-lowering therapy  or bariatric surgery . In an elegant study performed in nondiabetic subjects undergoing partial pancreatectomy  the same ∼50% reduction of the β-cell mass, resulted in different degrees of β-cell dysfunction depending on the subject’s insulin sensitivity, underscoring the discrepancy between mass and function. Finally, in a very recent study in a rodent model of T2DM  a significant fraction (75%) of β cells were found to be present but not producing insulin in response to glucose, confirming the dissociation between function and mass.
Insulin clearance is usually expressed in two ways, depending on the site of entry of the hormone into the circulation: (i) exogenous (or peripheral) insulin clearance, which can be determined experimentally through a euglycemic glucose clamp as the ratio between exogenous insulin infusion and arterial insulin concentration at steady-state, and (ii) endogenous (or prehepatic) insulin clearance, which is calculated as the ratio between endogenous insulin secretion and arterial insulin concentration at steady-state.The latter is an important determinant of plasma insulin concentration in physiologic conditions; however, its direct experimental determination is hampered by the difficulty in gaining access to the portal vein where endogenous insulin is secreted. Thus, its determination typically rests on the indirect calculation of insulin secretion through C-peptide deconvolution. Assembling data from different studies and our own data, in Figure 8.5 we provide estimates of the values of whole body and major organs insulin clearance in a subject both in the fasting state and during a 2-h OGTT; the two conditions will be discussed separately.
Insulin clearance in fasting conditions
In quantitative terms the fraction of portal insulin that is removed by the liver in its first pass is approximately 65%, ranging between 50 and 70% [50,146–148]. Once into the systemic circulation, insulin is cleared again by the liver with a similar efficiency and to a lesser extent by the skeletal muscles and the kidneys. The overall contribution of the liver (first pass plus recirculation) is therefore dominant (approximately 90%).
Information on the biologic variability of endogenous fasting insulin clearance is lacking and can be indirectly inferred from the data on exogenous insulin clearance. Exogenous insulin clearance displays a large inter-individual variability, which is chiefly (∼50%) explained by genetic factors [149,150] and, to a lesser extent (∼20%), by the negative effect of abdominal obesity and liver fat , and insulin resistance . The role of insulin clearance in the maintenance of glucose homeostasis is commonly interpreted as compensatory, with lower values being observed in those who are more insulin-resistant and would take advantage of increased insulin levels. This may not hold true since experimental animals in which insulin clearance was reduced by knocking out the insulin degrading enzyme , after a transient mild improvement in glucose metabolism developed insulin resistance and diabetes. Similarly, fasting hyperinsulinemia and mild degrees of insulin resistance developed in pancreas-transplanted patients with organ venous drainage into the systemic circulation compared to those with portal drainage . In physiologic conditions there is no convincing evidence that endogenous insulin clearance plays a role in the control of fasting glucose homeostasis .
Insulin clearance in the fed state
In the fed state endogenous insulin clearance is lower compared to fasting (Figure 8.5) as a consequence of the saturation of hepatic insulin removal for elevated insulin concentrations and possibly of other unknown factors.The contribution of the muscle is reduced  while that of the kidney is increased  although together they still contribute no more than 20% to overall insulin clearance. The 25–30% reduction with respect to the fasting rate displays a wide inter-individual variability (from 10 to 40%), which is largely, though not entirely, explained by the different insulin concentrations (lower clearance for higher levels). A recent study  has revealed the mechanism for the saturation and a source of variability: mice lacking the transporter (ZnT8), which enriches the insulin vesicles in zinc, demonstrate that this ion is responsible for the inhibition of the clathrin-dependent insulin endocytosis in the liver and subjects with a single nucleotide polymorphism in the gene encoding for this transporter also display an increased insulin clearance during an OGTT. Little information is available on the nongenetic factors that influence the reduction in insulin clearance observed in the fed state. A role of the central nervous system is possible  as well as signals other than GLP-1 from the gut  to the liver via nitric oxide generation [23,160].
The presence of reduced insulin clearance in all the conditions of hyperinsulinemia/insulin resistance (diabetes, obesity, fatty liver) has led to the conclusion that the changes in clearance are compensatory: they guarantee higher peripheral levels at similar secretion rates. This association, however, is largely dependent on the saturation phenomenon and it is thus very difficult to extrapolate the contribution of insulin clearance to glucose homeostasis independently of the prevailing plasma insulin levels.
On the other hand, there is evidence that insulin clearance can influence glucose tolerance independently of other factors. For instance, an acute 30% increase in insulin clearance, produced by systemic nitric oxide inhibition, is able to deteriorate glucose tolerance in normal subjects . Furthermore, surgical alteration of insulin clearance in pancreas-transplanted patients with portal or systemic pancreatic drainage  produces homeostatic changes that are independent of β-cell function.
Therefore, in the fed state insulin clearance, by modulating insulin levels, appears to contribute to glucose homeostasis in concert with, and not in response to, insulin sensitivity and secretion.