The contribution of the kidneys to postabsorptive endogenous glucose production is a matter of controversy. To assess whether this could relate to the use of various isotopical methods with different analytical performance capabilities, we measured glucose kinetics in 12 healthy subjects. Blood samples were taken from the femoral artery and the renal vein after 4 h of [6,6-2H2]glucose infusion (for gas chromatography [GC]/mass spectrometry [MS] analysis), and renal plasma flow was determined with paraaminohippurate. In addition, six subjects received uniformly labeled [13C]glucose (for GC/combustion/isotope ratio MS [IRMS]) and [3-3H]glucose (for counting of radioactive disintegrations). Arterial glucose concentrations (means ± SD) were 4.2 ± 0.1 mmol/l, and endogenous glucose production rates using [2H2]glucose were 2.2 ± 0.1 mg · kg–1 · min–1 or 818 ± 50 μmol/min. Dilution of [2H2]glucose across the kidney was 0.79 ± 1.32%, and renal glucose production (RGP) rates were 27 ± 72 μmol/min. In the six subjects receiving additional tracers, dilutions across the kidney were 2.83 ± 0.72 and 0.54 ± 1.20 (for [U-13C]glucose and [3-3H]glucose, respectively, the dilution with [U-13C] being higher than that with [2H2] (P = 0.007). Corresponding RGP values were 144 ± 39 and 43 ± 76 μmol/min for [U-13C] and [3-3H], respectively. In conclusion, we found that the highly sensitive [U-13C] GC/Combustion/IRMS technique showed consistent dilution of label across the kidney, whereas the less sensitive techniques gave some negative values and smaller RGP rates. Thus, depending on which technique is being used, a fivefold difference in calculated RGP values may be encountered. The methodological variability of our data suggests that extrapolation from regional renal measurements to the whole-body level should be perfumed with caution.
Interest in the potential role of the kidney in the regulation of glucose metabolism in humans has been revived within the past decade (1). The capacity of the renal cortex to produce glucose de novo was first recognized in 1937 (2). Later studies failed to show any significant net renal glucose output in the postabsorptive state in humans (3–6), although it was observed that with prolonged fasting the kidneys contribute ≤50% of endogenous glucose production (3). However, the perception of the kidney as playing a minor role in overall glucose homeostasis postabsorptively was recently challenged by a study using isotope dilution and renal vein catheterization concomitantly; the results showed that in postabsorptive dogs, renal glucose production (RGP) and renal glucose uptake each account for 30% of total whole-body glucose turnover in the presence of net balances close to zero (7). Subsequent human studies from two groups using similar techniques showed that the kidney may be responsible for 15–30% of endogenous glucose production in the basal postabsorptive state, and they also demonstrated that RGP increased in response to hypoglycemia and epinephrine (8–14). However, these results were not confirmed by Ekberg et al. (15), who reported that there was no significant renal dilution of labeled glucose and that postabsorptive RGP at most would be ∼5%. Although currently unclear, some of the controversy could relate to different methodological approaches.
When using the combined glucose label–dilution and arteriovenous balance technique across the kidney, calculations of glucose exchange depend on small across-organ differences of both labeled and total glucose concentrations. These difficulties are further complicated by the high renal blood flow, which will magnify any imprecision in the determination of glucose concentrations and dilution. The gas chromatography (GC)-to-combustion-to-isotope ratio mass spectrometry (IRMS) technique is very precise, because it exploits the high sensitivity of the isotope ratio mass spectrometer (16). To our knowledge, this powerful technique has not been used previously to describe regional renal glucose kinetics.
This study was designed in an attempt to resolve the discrepancies pertaining to assessment of the role of the kidney in glucose homeostasis. We proposed to assess whether the GC/IRMS technique, because of its higher sensitivity, offers advantages compared with more conventional dilution methods. We studied 12 healthy subjects receiving [6,6-2H2]glucose (GC/mass spectrometry [MS] analysis), [3-3H]glucose (for counting of radioactive disintegrations), and uniformly labeled [U-13C]glucose (for GC/IRMS).
RESEARCH DESIGN AND METHODS
We recruited 12 healthy subjects (6 men and 6 women) aged 28.3 ± 1.8 years, with BMI 18–27 kg/m2 and total body weight 68.8 ± 6.9 kg (Table 1). Fasting blood glucose levels were normal (70–100 mg/100 ml) in all participants, and each had normal physical examinations and electrocardiograms, as well as normal hematological, renal, and hepatic function as assessed by biochemical screening. One subject received levothyroxine because of stable primary hypothyroidism. For 3 days before the start of the study, all subjects ingested a standardized weight-maintaining diet (20% protein, 30% fat, and 50% carbohydrate) prescribed by a research dietitian. The protocol was approved by the Mayo Institutional Review Board, and the purpose and potential risks of the study were explained to all subjects. Informed written consent was obtained.
Experimental design.
The studies were conducted after a 12-h overnight fast in the General Clinical Research Center at the Mayo Clinic. The evening before the study, an intravenous catheter was inserted into an antebrachial vein and kept patent with saline. At 8:00 a.m. the following morning, duplicate baseline samples were drawn, and priming doses followed by continuous infusion of [6,6-2H2]glucose (5.0 mg/kg and 5.0 mg · kg–1 · h –1) were given for determination of isotopic enrichment. In addition, 6 of the 12 subjects received primed continuous infusions of [3-3H]glucose (18 μCi and 0.18 μCi/min) and uniformly labeled [13C]glucose (0.1 mg/kg and 0.1 mg · kg–1 · min–1). All infusions lasted 4 h.
At 9:00 a.m., cannulation of the femoral vein, one renal vein, and the femoral artery was performed as described (5,8,17,18). Renal vein catheters were inserted under fluoroscopic guidance, and correct positioning was confirmed by contrast injection. The femoral artery line was used both for blood sampling and for infusion of indocyanine green (0.6 mg/min) to measure blood flow in the leg. Femoral and renal vein catheters were used to collect blood samples. The peripheral antebrachial vein was used for infusion of isotopes and paraaminohippurate (PAH), which was used for measurements of renal plasma flow. Starting at 10:30 a.m., indocyanine was infused for 90 min at a rate of 30 mg/h. PAH was given as a bolus of 10 mg/kg total body weight at 10:30 a.m., followed by a 90-min constant infusion of 1.5 g/h. The study was stopped after 240 min (12:00 p.m.), at which time the catheters were removed and hemostasis obtained. Blood samples were collected at baseline and at five time points (180, 195, 210, 225, and 240 min).
Materials.
[6,6-2H2]glucose (2H, 99 atom percent) and [U-13C6]glucose (13C, 99 atom percent) were purchased from Cambridge Isotope Laboratories (Andover, MA) and [3-3H]glucose was purchased from New England Nuclear Life Science Products (Boston, MA). The chemical, isotopic, and optical purity of these compounds was confirmed before use. Solutions were prepared under sterile conditions in the pharmacy and were shown to be free of bacteria and pyrogens before administration. Cardiogreen (indocyanine) was purchased from Becton Dickinson (Cockeysville, MD), and PAH was purchased from Merck (West Point, PA).
Analysis of samples.
Plasma glucose was measured by a glucose oxidase method (Beckman Instruments, Fullerton, CA) and transformed into blood glucose values using a conversion factor of 0.85. Hormonal assays were performed as previously described (19). Plasma insulin and growth hormone were measured by a chemiluminescent sandwich assay (Sanofi Diagnostics, Chaska, MN), and glucagon and cortisol were measured by radioimmunoassay (Diagnostic Products, Los Angeles, CA). Indocyanine green concentrations were measured by spectrophotometry, and PAH was measured by colorimetry.
Plasma enrichment levels of [6,6-2H2]glucose were determined by GC/MS (5972; Hewlett-Packard, Palo Alto, CA) using the trimethylsilyl-O-methyloxime derivative and monitoring fragment ions at m/z 321 and 319 under electron ionization conditions, as described by Küry et al. (20). The coefficient of variation of this analysis was 2.5% (i.e., 2.5 ± 0.05 molar percent excess).
Plasma enrichment of [U-13C]glucose was determined by IRMS using a Finnigan MAT DeltaS system (Bremen, Germany) fitted with an on-line GC/combustion inlet. The proteins from 40 μl plasma were precipitated with 1 ml of ice-cold ethanol, and the supernatant, after centrifugation, evaporated to dryness using a centrifugal evaporator (Savant, Farmingdale, NY). The methyl boronate derivative of glucose was prepared (21) and separated using a 30 m × 0.32 mm id × 0.25 μm film fused-silica capillary column (DB-1701; J&W, Folsom, CA). The coefficient of variation of this analysis was 1% (i.e., 0.02 ± 0.0002 atom percent excess).
The specific activity of [3-3H]glucose was determined as described previously (22). The coefficient of variation of this measurement was 2 ± 0.24%.
Calculations.
Plasma flows (PFs) from the kidney were calculated as described previously (5) and converted to blood flows (BFs) using the following equation: BF = PF/(1-hematocrit). Measured hematocrit values were 40 ± 1%.
Isotopic plateau was observed from 180 to 240 min. This was assessed based on the observation that when isotopic enrichment values or specific activities of the glucose tracers used in different sites were plotted against time, the ensuing slopes were not different from zero. The mean values of five measurements for any isotope at each plateau were used for all calculations of glucose kinetics.
Whole-body rate of appearance (Ra) for glucose was calculated by dividing the rate of infusion of labeled glucose by enrichment or specific activity.
Net renal glucose exchange was calculated as the product of arteriovenous plasma glucose concentration differences and renal blood flow.
Regional renal production of glucose was calculated in all circumstances using the following equation (15,21):
in which Ra is RGP, BF is total renal blood flow, [glucose]art is the arterial blood glucose concentration, and Even and Eart represent enrichment (or specific activity) in venous or arterial blood. As pointed out by Ekberg et al. (15), the low fractional extractions (<3%) of glucose across the renal bed means that their contributions to the calculation become negligible. For that reason, the parameter was omitted. Negative values for regional glucose production were included in the mean calculations.
Regional renal glucose uptake was determined as the sum of RGP and net renal glucose exchange.
Statistical analysis.
The mean of all five measurements during the 4-h infusion was used for calculations. All values given are means ± SE. Differences among tracers were assessed by Student’s t test.
RESULTS
Parameters of whole-body glucose metabolism and circulating hormones are given in Table 1. Circulating arterial blood glucose concentrations were 4.2 ± 0.1 mmol/l, and the total glucose Ra determined with [2H2]glucose was 2.2 ± 0.1 mg · kg–1 · min–1 (or 818 ± 50 μmol/min) in the 12 subjects (Table 1).
Table 2 shows isotopic enrichment and specific activity obtained with the three tracers in the femoral artery and in the renal vein. In all cases, a small overall dilution of tracer was recorded across the renal bed, ranging from 0.54% with [3-3H]glucose to 2.83% with [U-13C]glucose. The percent dilution was higher with [U-13C]glucose compared with [2H2]glucose (P = 0.007), but not different between [U-13C]glucose and [3-3H]glucose (P = 0.22). As shown in Fig. 1, individual values of dilution for each subject were quite scattered, and a considerable number of subjects exhibited negative values. However, for [U-13C]glucose (followed by GC/IRMS), only positive values were recorded for all six subjects.
Calculated values for renal glucose dynamics are given in Table 3. Arterial and venous glucose concentrations were very similar; some were positive, whereas others were negative, as shown in Fig. 2. This gave rise to a small average net glucose release in the [6,6-2H2]glucose, [3-3H]glucose, and [U-13C]glucose experiments. RGP rates were 27 ± 72, 144 ± 39, and 43 ± 76 μmol/min (for [2H2], [U-13C], and [3-3H], respectively). Only utilization of [U-13C]gave consistently positive values for RGP. Calculated values for the renal contribution to the entire endogenous glucose production varied from 4.9 ± 9.0% ([6,6-2H2]) and 4.3 ± 8.3% ([3-3H]) to 18.3 ± 4.3% [U-13C].
DISCUSSION
This study was undertaken to assess whether methodological differences among the use of various glucose tracers could account for the discrepant results reported on the contribution of the kidney to endogenous glucose production in postabsorptive humans. The main finding was that only utilization of the highly sensitive [U-13C]-GC/IRMS method yielded consistently positive values for RGP, whereas less sensitive methods gave negative values in some cases.
Previous studies examining RGP in the basal state in humans have reported values that range from ∼ 0% (15) to ∼30% (8) of endogenous glucose production. A summary of reported parameters of renal glucose metabolism in humans is given in Table 4. It may be noted that both renal net exchange of glucose (ranging from 0 to 70 μmol/min) and tracer dilution across the renal bed (ranging from 0 to 4.2%) vary considerably. Our findings of a renal net release of glucose between 60 and 90 μmol/min and isotope dilutions ranging from 0.5 to 2.8% are thus comparable to those of the literature (24,25). Conceivably, the excessive variability of results relates to the fact that measured arteriovenous differences are very close to the detection limits of the assays used. The resulting imprecision is amplified manifold by the large renal blood flow of >1 l/min.
In the studies above, extrapolation from data on renal blood flow, arterial and renal venous glucose concentrations, and dilution of labeled glucose molecules across the renal bed to whole-body glucose kinetics rests on a number of assumptions. Total renal blood flow is generally used to convert plasma measurements in one kidney to an estimate of the entire renal contribution. This assumes that the two kidneys are metabolically identical and that conversion from plasma determinations to whole blood by using hematocrit values is appropriate. However, the use of whole-blood flow rates may overestimate actual glucose kinetic events to the extent that blood is not composed of water alone, but also inert dry matter (e.g., plasma membranes, hemoglobin, and plasma proteins). The concentration of water is ∼93% in plasma and ∼70% in blood cells (26,27), meaning that across-organ flux rates in free water may be 15–20% lower than calculated. For this reason, we converted measured plasma glucose values to whole-blood values by multiplying them with 0.85. In addition, hematocrit values may decrease in the course of an investigation because of repeated blood sampling and the infusion of large amounts of water. Furthermore, local loss of [6-H]-labeled glucose during Cori cycling and [3-H]-labeled glucose during triose-phosphate cycling may lead to an overestimation of regional glucose production. Although the role of red cell glycolysis during passage through the renal bed remains to be defined, equlibration of glucose across red cell membranes occurs rapidly, and it seems fair to assume that intracellular water concentrations of labeled and unlabeled glucose are identical to plasma concentrations (28). Alternatively, if it is assumed that no equilibration occurs between red cells and plasma, the calculated values of RGP will be increased by 30%. Finally, it is possible that hemoconcentration due to urine and lymph production may elevate effluent renal vein glucose concentrations, thus spuriously increasing calculated rates for net RGP. Most of these potential limitations will tend to induce an overestimate of the renal contribution to whole-body glucose production.
On the other hand, underestimation of renal glucose output may occur to the extent that there is recycling of labels and that contamination of renal venous blood with caval or gonadal blood occurs. Recycling of deuterium and tritium from labeled water may take place, and synthesis of glucose from 13C-labeled 3-carbon precursors (lactate, pyruvate, alanine, and glycerol) may also contribute. Any possible role of these processes remains uncertain. Recycling of labeled glycogen is probably negligible, because normal human kidney does not contain appreciable amounts of glycogen (29). In this study, renal vein concentrations of PAH ranged from 7 to 16% of the arterial concentrations, virtually excluding the possibility that any significant contamination from caval or gonadal blood occurred.
The above considerations, however, do not explain the variability of results in the literature concerning the contribution of kidney to endogenous glucose production. As pointed out by Ekberg et al. (15), the existing discrepancies could relate to omission or reanalysis of negative values for renal dilution of tracers in some cases. When using the GC/combustion/IRMS technique with high accuracy and precision for the analysis of [U-13C]glucose kinetics across the kidney, we found clear evidence for a small but consistent dilution of labeled glucose. These results could not be reproduced with less sensitive tracer techniques, probably due to lack of analytical precision to some extent. It is intriguing that calculated mean values for RGP were higher with [U-13C]glucose than with [2H2]glucose and [3H]glucose. One would expect intrarenal dilution of deuterium and tritium to be increased due to potential loss of label in glycolysis. Another potential reason for the variable results in the existing literature is that the degree of stress to which the participants had been exposed may have varied. It has been shown that RGP is substantially increased by epinephrine (8).
In conclusion, the data show that tracer dilution in the kidney may vary from 0.5 to 2.8%, depending on the method used, and that the use of the highly sensitive GC/combustion/IRMS technique gives consistently positive results for RGP. This implies that analysis of renal glucose metabolism is very susceptible to methodological noise and that renal contribution to endogenous glucose production is between 4 and 18% in postabsorptive humans. In view of the small arteriovenous differences in isotopic enrichment (or specific activity), future studies should consider larger numbers of subjects to minimize the effect of analytical errors.
Arterial blood glucose concentrations, whole-body glucose rate of appearance by [2H2]glucose, and circulating hormones in 12 subjects undergoing renal vein catheterization
| Glucose (mmol/l) | Ra (mg · kg–1 · min–1)/ (μmo 1/min) | Insulin (pmol/l) | Glucagon (pg/ml) | Growth hormone (ng/ml) |
|---|---|---|---|---|
| 4.2 ± 0.1 | 2.2 ± 0.1/818 ± 50 | 33.1 ± 4.8 | 63.2 ± 3.1 | 0.67 ± 0.30 |
| Glucose (mmol/l) | Ra (mg · kg–1 · min–1)/ (μmo 1/min) | Insulin (pmol/l) | Glucagon (pg/ml) | Growth hormone (ng/ml) |
|---|---|---|---|---|
| 4.2 ± 0.1 | 2.2 ± 0.1/818 ± 50 | 33.1 ± 4.8 | 63.2 ± 3.1 | 0.67 ± 0.30 |
Data are means ± SE.
Isotopic enrichment (molar percent excess) and specific activity (dpm/μmol) in arterial and renal vein plasma and renal vein dilution of tracers
| n | Artery | Renal vein | Dilution (%) | |
|---|---|---|---|---|
| [2H2]glucose | 12 | 2.663 ± 0.311 | 2.657 ± 0.323 | 0.79 ± 1.32 |
| [U-13C]glucose | 6 | 0.0662 ± 0.0016 | 0.0631 ± 0.0014 | 2.83 ± 0.72 |
| [3-3H]glucose | 6 | 489.2 ± 41.2 | 487.0 ± 42.0 | 0.54 ± 1.20 |
| n | Artery | Renal vein | Dilution (%) | |
|---|---|---|---|---|
| [2H2]glucose | 12 | 2.663 ± 0.311 | 2.657 ± 0.323 | 0.79 ± 1.32 |
| [U-13C]glucose | 6 | 0.0662 ± 0.0016 | 0.0631 ± 0.0014 | 2.83 ± 0.72 |
| [3-3H]glucose | 6 | 489.2 ± 41.2 | 487.0 ± 42.0 | 0.54 ± 1.20 |
Data are means ± SE.
Renal blood flow rates, blood glucose concentrations, net glucose balances, and glucose appearance and disappearance rates
| [2H2]glucose | [U-13C]glucose | [3-3H]glucose | |
|---|---|---|---|
| n | 12 | 6 | 6 |
| Renal blood flow (ml/min) | 1233 ± 104 | 1274 ± 122 | 1274 ± 122 |
| Arterial glucose concentration (mmol/l) | 4.21 ± 0.09 | 4.11 ± 0.18 | 4.11 ± 0.18 |
| Renal vein glucose concentration (mmol/l) | 4.27 ± 0.09 | 4.18 ± 0.17 | 4.18 ± 0.17 |
| Net renal glucose exchange (μmol/min) | 61.2 ± 25.5 | 89.8 ± 39.5 | 89.8 ± 39.5 |
| RGP (μmol/min) | 27.2 ± 72.2 | 144.0 ± 39.2 | 43.3 ± 75.9 |
| Calculated renal contribution to endogenous glucose production (%) | 4.9 ± 9.0 | 18.3 ± 4.3 | 4.3 ± 8.3 |
| Renal glucose uptake (μmol/min) | 34.0 ± 73.1 | 54.3 ± 40.5 | 46.5 ± 84.7 |
| [2H2]glucose | [U-13C]glucose | [3-3H]glucose | |
|---|---|---|---|
| n | 12 | 6 | 6 |
| Renal blood flow (ml/min) | 1233 ± 104 | 1274 ± 122 | 1274 ± 122 |
| Arterial glucose concentration (mmol/l) | 4.21 ± 0.09 | 4.11 ± 0.18 | 4.11 ± 0.18 |
| Renal vein glucose concentration (mmol/l) | 4.27 ± 0.09 | 4.18 ± 0.17 | 4.18 ± 0.17 |
| Net renal glucose exchange (μmol/min) | 61.2 ± 25.5 | 89.8 ± 39.5 | 89.8 ± 39.5 |
| RGP (μmol/min) | 27.2 ± 72.2 | 144.0 ± 39.2 | 43.3 ± 75.9 |
| Calculated renal contribution to endogenous glucose production (%) | 4.9 ± 9.0 | 18.3 ± 4.3 | 4.3 ± 8.3 |
| Renal glucose uptake (μmol/min) | 34.0 ± 73.1 | 54.3 ± 40.5 | 46.5 ± 84.7 |
Data are means ± SE.
![FIG. 1. Individual values (percent-wise) of dilution of isotopic enrichment across kidney of [2H2]glucose and [U-13C]glucose as well as the specific activity of [3-3H]glucose, demonstrating that there was a dilution of label across kidney in all subjects when [U-13C]glucose was used as a tracer.](/view-large/figure/2212993/db0410403001.jpeg)
Individual values (percent-wise) of dilution of isotopic enrichment across kidney of [2H2]glucose and [U-13C]glucose as well as the specific activity of [3-3H]glucose, demonstrating that there was a dilution of label across kidney in all subjects when [U-13C]glucose was used as a tracer.
Individual values (percent-wise) of dilution of isotopic enrichment across kidney of [2H2]glucose and [U-13C]glucose as well as the specific activity of [3-3H]glucose, demonstrating that there was a dilution of label across kidney in all subjects when [U-13C]glucose was used as a tracer.
Individual values of arteriovenous differences in glucose (mmol/l) concentrations across kidney, showing that the differences are small and vary from subject to subject.
Individual values of arteriovenous differences in glucose (mmol/l) concentrations across kidney, showing that the differences are small and vary from subject to subject.
Article Information
This study was supported by National Institutes of Health Public Health Service grants R01 DK41973, DK29953, and RR00585. K.S.N. was supported by a David Murdock-Dole Professorship.
We gratefully acknowledge the support of Dr. J. Andrews for catheterizing the artery and vein, Maureen Bigelow for skillful assistance in performing the study, and Peter Berg for technical assistance.
REFERENCES
Address correspondence and reprint requests to K. Sreekumaran Nair, MD, Mayo Clinic and Foundation, Endocrine Research Unit, 200 1st St. SW, Rm. 5-194 Joseph, Rochester, MN 55905. E-mail: [email protected].
Received for publication 23 May 2000 and accepted in revised form 4 January 2001.
N.M. is currently affiliated with Medical Department M, Endocrinology and Diabetes, Kommunehospitalet, University Hospital of Aarhus, Aarhus, Denmark.
