Deuterated water can be used to assess the contribution of DNL to total palmitate in adipose samples and plasma very low-density lipoprotein (VLDL) triglycerides. Orally administered deuterated water is freely permeable across almost all biological membranes, is well mixed in the body, and its enrichment within the body water compartment is easily maintained for extended periods of time because of relatively slow turnover of body water (half-life 7–10 days). Labeled hydrogen atoms from deuterated water (²H₂O) enter into biosynthetic pathways by exchange with cellular water or by specific incorporation of H from NAD(P)H. Deuterated water is a particularly useful tracer for assessment of lipid synthesis because it can be administered for long periods of time, and the label is incorporated into multiple lipid components so that, if desired, the synthesis rate of cholesterol, phospholipid, triglyceride, and fatty acids can be determined in a single experiment. Cell proliferation, i.e., deoxyribonucleic (DNA) synthesis [18, 35] as well as protein synthesis, can also be measured through deuterated water labeling [11, 22, 36] (see protein synthesis).

Deuterated water has been given to humans for more than 70 years without significant adverse effects, and animals have been raised for sequential generations on relatively high levels of deuterium water (e.g., up to 15–20 %), without phenotypic consequences or adverse effects. Additionally, no known risks have been observed in studies examining the effect of deuterated water on male or female reproductive tissues, including sperm function. As mentioned, the only minor adverse effect reported for deuterated water at these doses is transient dizziness when the initial rate of deuterated water intake is too rapid, occurring in approximately 1 in 30 individuals, which typically resolves completely within 3 h. Accordingly, care must be taken in the initial loading rate of deuterated water that is administered.

Intravenous infusion of ¹³C-acetate is another common method used to measure hepatic DNL. Oral or intravenous acetate is efficiently taken up by the liver and the intrahepatic acetyl-CoA pool is labeled. As lipids are synthesized, ¹³C acetyl-CoA is incorporated into the new fatty acids and the fractional synthesis can in principle be calculated from a simple precursor/product ratio. The challenge historically has been the difficulty in accurately determining the isotope enrichment in the intrahepatic acetyl-CoA precursor pool. Additionally, other concerns regarding the lack of equilibrium of acetate across the liver have also been raised. Both of these methodological considerations have been largely resolved. The intracellular acetyl-CoA pool enrichment can be noninvasively determined using mass isotopomer distribution analysis (MIDA) [33, 37] or other similar combinatorial algorithms [38] which largely correct for any transhepatic disequilibrium [39] and have been shown to give results in line with deuterated water labeling methods where equilibration is not an issue. There are many possible ways to utilize this approach for assessing hepatic DNL. ¹³C-acetate can be safely infused for several days, and diurnal changes in DNL can be monitored [40, 41]. For proof of concept studies focused on the regulation of hepatic DNL, we have found the use of a controlled dietary stimulation to deliver very reproducible results in cross-sectional and particularly longitudinal studies where subjects can serve as their own controls.

While obese and diabetic subjects are known to have higher than normal basal fasting levels of hepatic fractional DNL, on the order of 15–20 % contribution to VLDL palmitate, healthy lean individuals have fasting hepatic fractional DNL on the order of 3–5 % after a brief 8–12-h infusion of labeled acetate [40, 42]. Fasting DNL varies considerably within as well as among subjects, due to influences from changes in exercise, alcohol, diet (both caloric content and carbohydrate composition) [30, 43], energy balance, weight, etc. [26]. A high carbohydrate diet can increase DNL from 5 % or less to 20 % in the fasting state, for example. Accordingly, study cohorts should be carefully enrolled to match for these physiological factors, as well as body mass index (BMI), sex, age, etc. when fasting DNL is the measurement of interest. It is known that acute ingestion of ethanol and fructose can increase fractional DNL contribution to circulating fatty acids in VLDL triglyceride from baseline to approximately 35 %, and we have shown that fructose-stimulated DNL assessments are very reproducible compared to fasting DNL [44]. The improvement in variability in the DNL measurement with fructose feeding will require fewer subjects to find certain effects in clinical studies.

Fasting EGP can be assessed using the isotope dilution technique [47]. Originally the method involved the use of radioactive tracers but the introduction of stable isotope-labeled glucose resulted in extensive research exploring the effect of disease and drug treatment on EGP in the postabsorptive state. Isotopic measurement of EGP represents glucose release from both the liver and the kidney, and although the quantitative contribution of renal glucose release (between 5 % and 28 % in healthy subjects) is small compared to that of the liver, one should not equate whole-body isotopically measured EGP with hepatic glucose production alone [48].

A variety of experimental protocols for the measurement of fasting endogenous glucose production have been described in the literature. Methodological differences are noted in the type of glucose tracer, administration of the tracer (continuous infusion vs. bolus), priming technique, tracer infusion time, timing, and number of sample collection and calculation models (steady-state vs. non-steady-state equations) used. It is generally accepted that the use of [U-¹³C₆]-glucose and [6,6-²H₂]-glucose provide the most relevant and easily interpretable measurement of total EGP in context of isotope loss or recycling during partial metabolism of glucose taken up by the liver [49]. Detailed description of the different tracers for the use of glucose production has been published by Wolfe [3]. Often, the tracer is administered as a primed-continuous infusion for 2–5 h, and plasma glucose enrichments are determined by GC/MS during the last 30–60 min of the steady-state period. The rate of EGP is then calculated using the steady-state equation:
In healthy subjects a fixed priming dose corresponding to a 100-min infusion is usually used. However, in insulin-resistant states, where the glucose pool size is increased, the priming dose should be adjusted for plasma glucose concentrations, and a prolonged infusion (4–6 h) is required for valid measurements when using steady-state equations, because of slower turnover rate of the larger extracellular glucose pool. An insufficient priming and infusion time when hyperglycaemia is present has been shown to overestimate EGP due to failure to reach true plateau glucose enrichments in the plasma compartment [50–52]. Indeed, the greater EGP rates in people with type 2 diabetes compared to healthy subjects have been questioned due to tracer protocol design. In our experience, using an adjusted prime and a 5–7-h continuous infusion of [U-¹³C₆]-glucose or [6,6-²H₂]-glucose, we can reproducibly demonstrate an isotopic steady state in essentially all subjects, and we consistently observe higher EGP values in type 2 diabetes compared to healthy subjects confirming published results.

While the primed-continuous infusion of a glucose tracer is the most commonly used method for the measurement of fasting EGP, and considered the gold standard, there may be some instances where a primed constant infusion is not desirable or possible. As an alternative, a bolus injection method has been validated against the primed-continuous infusion method [53]. By this method, a glucose tracer is administered as a bolus (7 mg/ kg body weight of [U-¹³C₆]-glucose given over 1–3 min) and frequent blood samples are collected for 1–3 h. The decay of plasma glucose tracer enrichments is fit to a two- or three-exponential curve using a kinetic simulation module (e.g., SAAM II software [54]), and EGP is then calculated from the fit parameters. While published data are currently limited to healthy volunteers, we have seen very good agreement in patients with type 2 diabetes and impaired glucose tolerance as well (unpublished results).

Several endpoints can be calculated from the direct measurement of EGP. Hepatic insulin sensitivity has been calculated as the inverse of the product of EGP and fasting plasma insulin [55], while the metabolic clearance rate (MCR), which is a measurement of the ability of whole-body tissues to take up glucose in the fasting state, is calculated as EGP divided by glucose concentration.

Several approaches have been described to measure the contribution of gluconeogenesis and glycogenolysis to fasting endogenous glucose production [3]. In this chapter we focus on the method described by Hellerstein et al. involving the infusion of [2-¹³C₁] glycerol and the use of MIDA [56, 57]. This method allows the calculation of gluconeogenesis from all three-carbon sources; lactate, amino acids, and glycerol. A primed-continuous infusion of [2-¹³C₁] glycerol is administered to label the precursor pool (hepatic triose phosphate) for glucose synthesis for several hours (see above), and samples are taken during the last 30–60 min. The [2-¹³C₁] glycerol can be infused with [U-¹³C₆]-glucose or [6,6-²H₂]-glucose for the simultaneous measurement of gluconeogenesis and endogenous glucose production. The principle of MIDA and its application for measuring biosynthesis of a variety of polymers (e.g., glucose) have been discussed in detail elsewhere [37, 58]. In brief, the technique consists of quantifying fractional abundances of different mass isotopomers (i.e., different isotopic isomers) in an intact polymer (e.g., glucose) that contains two or more repeats of a precursor monomeric subunit. Gluconeogenesis can be conceptualized as the polymerization synthesis of two triose-phosphate monomers. The fractional abundances of mass isotopomers (M0 for unlabeled glucose, M1 for singly labeled glucose, and M2 for doubly labeled glucose) in the polymer must then be accurately measured by GC/MS and compared with the natural fractional abundance of baseline values to calculate changes, or excesses, of mass isotopomeric species in the polymer. The precursor pool enrichment (i.e., the hepatic triose-phosphate isotope content, for glucose) can then be back-calculated from the pattern or distribution of mass isotopomeric excesses in the polymer (glucose), using formulae derived from the binomial or multinomial expansions, as described in detail elsewhere [57]. Once the precursor pool enrichment is established, the precursor/product relationship can be applied for calculation of the fraction of polymers that were derived from the endogenous biosynthetic pathway during the labeling period. In the case of gluconeogenesis, the ratio of excess doubly labeled glucose to excess single-labeled glucose species reveals the isotopic abundance of the true precursor pool, from which fractional gluconeogenesis can be calculated. Absolute flux through gluconeogenesis into plasma glucose can then be calculated as the fractional gluconeogenic contribution times the rate of fasting EGP. These two parameters can be assessed simultaneously, as described above. The absolute contribution of glycogenolysis to fasting EGP is calculated as the difference of rate of fasting EGP and rate of gluconeogenesis. Potential methodological problems have been addressed and discussed in detail by Neese et al. [57]. More complex stable isotope techniques like the secreted glucuronate technique have been described for the measurement of intrahepatic glycogen fluxes. This method has been described in detail elsewhere [57, 59].

Hepatic insulin resistance is a key player of whole-body insulin resistance in people with type 2 diabetes [60]. Fasting rates of EGP are often elevated and suppression of glucose production during intestinal absorption of a meal is impaired in type 2 diabetes. The gold standard to determine hepatic insulin sensitivity is the measurement of partial suppression of EGP during a low-dose insulin infusion. Tracer methodology combined with the hyperinsulinaemic euglycaemic clamp allows the simultaneous measurement of fasting EGP, insulin-mediated suppression of hepatic glucose production (hepatic insulin sensitivity), and tissue glucose disposal (peripheral insulin sensitivity) independent of the effects of hyperglycaemia per se. Currently the best methodology to do this is the so-called hot GINF (glucose infusion) method proposed a few decades ago [61, 62]. The method is designed to keep the ratio of the glucose tracer to tracee in plasma constant during the fasting state and during the clamp to minimize errors (or negative values) in EGP measurements due to non-steady-state conditions. Ideally, deviations in circulating tracer enrichments should be kept to <20 % to avoid errors in the EGP results [63]. The method involves the administration of a glucose tracer as a primed-continuous infusion and as part of the exogenous glucose infusion (GINF) used during the clamp. The tracer enrichment of the GINF is calculated to be equal to the expected plasma tracer enrichment at the start of clamp. Originally a radioactive or hot glucose tracer was added to the exogenous glucose infusion, hence the name hot GINF. The continuous infusion of the glucose tracer is started before the clamp to determine fasting EGP as described above. Its rate is then reduced at the start of the clamp and after fasting EGP measures are completed to approximate the expected fall of EGP during the clamp (e.g., the infusion rate of the tracer is reduced to half the starting rate if a 50 % reduction of EGP is expected during the clamp). The anticipated fall in EGP during the clamp depends on several factors such as patient population (e.g., healthy subjects may have a greater suppression of EGP than subjects with type 2 diabetes under similar insulin conditions), clamp insulin infusion rates, and duration of the clamp. Pilot experiments are recommended to determine the optimal tracer design during the clamp if there is not prior experience with the patient population and method. An intrinsic limitation of this approach is therefore that the quantitative effects of an intervention on EGP have to be to some extent known in advance to apply the model. Examples of EGP measured during a clamp in a healthy subject and an individual with type 2 diabetes are shown in Fig. 3.3.