If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Thus far studying renal glucose metabolism non-invasively in humans was an unmet need.
•
We used a new approach to analyse renal [18]FFDG-PET data, where renal radioactivity was corrected for tubular radioactivity.
•
With this approach we found that the human renal cortex is an insulin sensitive tissue.
•
This is the first attempt to assess whether [18F]FDG-PET can yield information regarding renal glucose metabolism.
Abstract
Background/purpose
Studying renal glucose metabolism non-invasively in humans is an unmet need. Positron emission tomography (PET) is the current gold standard for measuring regional tissue glucose uptake rates, but the most widely used glucose analog ([18F]FDG) is not a good substrate for sodium-glucose cotransporters (SGLTs). As a consequence, [18F]FDG spills over into the urine and [18F]FDG-PET considerably underestimates published rates of whole renal glucose uptake obtained using the arterial-venous difference technique. Our aim was to assess whether [18F]FDG-PET can be used in the study of renal glucose metabolism in humans.
Methods
We measured individual [18F]FDG radioactivity in the urine and estimated intraluminal [18F]FDG radioactivity concentration; these values were used to correct renal [18F]FDG-PET data acquired ∼90 min from tracer injection under fasting conditions and during an insulin clamp in 9 lean and 16 obese subjects.
Results
We found that the corrected glucose uptake is consistently higher in the medulla than cortex and that both cortical and medullary glucose uptake are higher in lean than obese participants under both fasting and insulinized conditions. Moreover, cortical but not medullary glucose uptake is increased from the fasting to the insulinized condition.
Conclusion
The data show for the first time that [18F]FDG-PET can still provide relevant physiological information on regional renal glucose uptake on the condition that [18F]FDG uptake is corrected for tubular radioactivity.
The kidney is metabolically very active and demands large amounts of energy for its functions, a large portion being used for the active reabsorption of glucose, sodium, and other solutes from the proximal convoluted tubule (PCT) [
]. Along with the marked differences in perfusion and function, several metabolic differences have been described between the renal cortex and medulla. Tubular cells in the cortex are rich in mitochondria and depend predominantly on oxidative metabolism, with fatty acids, ketone bodies, and lactate being the preferred substrates [
]. Insulin receptors are abundantly expressed in the cortex, a region that also contributes to gluconeogenesis thanks to the exclusive expression of key gluconeogenic enzymes (glucose-6-phosphatase, fructose-1,6-biphosphatase, and phosphoenolpyruvate carboxykinase) [
]; here, glycogen is present at higher concentrations than in the cortex.
The study of renal substrate handling in vivo requires catheterization of the renal vein (RV) and an artery (A) (for the determination of substrate A-RV differences) and measurement of renal blood flow. These measures usually show high within-subject variability even under stable metabolic conditions as small A-RV differences are multiplied by a large blood flow (∼1.2 L/min). The case of glucose is further complicated by the fact that the kidneys simultaneously utilize, secrete, and excrete glucose. In fact, the net transrenal glucose gradient typically oscillates around zero in the fasting state, indicating similar rates of concomitant glucose uptake and production. For this reason, the catheterization technique has been combined with the infusion of a glucose tracer (e.g., 3H-glucose) which is taken up by tissues but is neither endogenously made through gluconeogenesis nor recycled to glucose by label rearrangement [
]. With this experimental setup, the fractional tracer extraction across the kidney multiplied by the arterial ‘cold’ glucose concentration and renal blood flow yields the absolute value of total renal glucose uptake; separate estimation of cortical and medullary glucose disposal is not possible.
Positron emission tomography (PET), presently the gold standard for the non-invasive measurement of tissue metabolic rates in vivo, has been widely used in conjunction with the glucose tracer, 18F-labeled-2-fluoro-2-deoxy-d-glucose ([18F]FDG), in studies of glucose utilization in virtually all organs but one, the kidney. This is largely because [18F]FDG, a glucose analog where fluorine-18 substitutes the hydroxyl at the second position in the glucose molecule (hence the acronyms [18F]FDG and 2-FDG have been used interchangeably) is filtered through the glomerulus but virtually not reabsorbed by the sodium-glucose cotransporters (SGLT1 and SGLT2); its accumulation in collecting tubules and pelvis has been thought to interfere with the tissue activity during [18F]FDG-PET scanning [
]. However, to the best of our knowledge attempts to evaluate whether renal [18F]FDG-PET scanning may still provide clinically relevant information have not been performed systematically.
In this study we aimed to correct rates of [18F]FDG uptake for tracer excretion, to compare such rates with published uptake rates obtained by the catheter-tracer technique, and to assess whether [18F]FDG-based regional renal glucose uptake rates can be reliably measured under fasting and euglycemic hyperinsulinemic conditions in healthy lean subjects and obese individuals.
2. Methods
2.1 Study participants and study design
Abdominal [18F]FDG acquisitions were carried out in the fasting state and during a euglycemic insulin clamp on separate days using the same protocol. The details of these studies have been described previously [
SLEEVEPASS: a randomized prospective multicenter study comparing laparoscopic sleeve gastrectomy and gastric bypass in the treatment of morbid obesity: preliminary results.
Brain glucose uptake is associated with endogenous glucose production in obese patients before and after bariatric surgery and predicts metabolic outcome at follow-up.
]. In brief, morbidly obese subjects were recruited from the SLEEVEPASS study, a larger randomized prospective clinical trial comparing different surgical techniques for the treatment of morbid obesity (ClinicalTrials.gov, NCT00793143). Healthy lean controls were also recruited. The exclusion and inclusion criteria have been previously published [
SLEEVEPASS: a randomized prospective multicenter study comparing laparoscopic sleeve gastrectomy and gastric bypass in the treatment of morbid obesity: preliminary results.
]. All subjects gave informed consent and were screened before they were included in the study. Renal [18F]FDG acquisitions were available in 25 subjects (16 obese and 9 lean) during a euglycemic insulin clamp and in 21 subjects (14 obese and 7 lean) also during the fasting state. For the insulin clamp study, two catheters were inserted in the antecubital veins, one for the administration of radiolabeled tracers (and of glucose and insulin on the insulin clamp study day) and the other for arterialized blood sampling. To obtain arterialized venous blood samples, the arm used for venous blood sampling was warmed with a heating pillow throughout the clamp study, as previously done [
Brain glucose uptake is associated with endogenous glucose production in obese patients before and after bariatric surgery and predicts metabolic outcome at follow-up.
]. The euglycemic hyperinsulinemic clamp (i.e. the gold standard method for assessing systemic insulin resistance) was performed as previously described [
]; in brief, a primed-continuous insulin infusion was given at a rate of 40 mU·m−2·min−1, followed by a variable 20 % dextrose infusion, in order to maintain plasma glucose levels steady at 5 mmol/L. At ∼100 ± 10 min of the clamp, [18F]FDG was injected over 15 s, and dynamic scanning started. Frequent arterialized blood sampling was done every 5, 30, and 60 min for the determination of plasma glucose and radioactivity, insulin, and free fatty acids, respectively. Immediately after the completion of the PET scanning, subjects were instructed to void their bladders in order to measure the amount (in MBq) of [18F]FDG excreted in the urine. Plasma and urinary radioactivity concentrations were measured using an automatic γ-counter (Wizard 1480; Wallac, Turku, Finland). The study protocol was approved by the Ethics Committee of the Hospital District of Southwestern Finland.
2.2 PET data analysis
The PET studies were conducted both in the fasting condition and during the insulin clamp on separate days <2 weeks apart using the GE advanced PET camera (General Electric Medica Systems, Milwaukee, WI). Radioactivity from the abdominal area was acquired 90–120 min from [18F]FDG injection for 15 min (3 × 300 s frames). PET images were reconstructed in a 256 × 256 matrix after correction for decay time, dead time, and photon attenuation. Image analysis was performed using Carimas v.2.9 (http://www.turkupetcentre.fi/). To obtain the time-radioactivity curves, the regions of interest (ROI) were manually drawn on PET/CT fusion images in renal cortex and medulla (Fig. 1). In particular, 4–5 consecutive thin ROIs were drawn using the coronal axis images in the region of slightly lower radioactivity just outside the high signal originating from the renal pyramids; this region of interest was considered to represent the renal cortex [
]. Data were analysed using the fractional uptake rate (FUR, 1/min). To obtain glucose uptake rates, FUR values were multiplied by the concomitant plasma glucose values. The lumped constant correction that is applied to account for the difference between glucose and [18F]FDG uptake from the studied tissue is not known for the kidneys and was assumed to be 1[
]. Glucose uptake (GU) rates are expressed as μmol·min−1·100 mL−1. Skeletal muscle GU (psoas muscle) was also calculated as FUR, multiplied by plasma glucose levels and divided by the lumped constant for skeletal muscle (1.2) [
Report of the Task Group on Reference Man. A Report Prepared by a Task Group of Committee 2 of the International Commission on Radiological Protection.
]. Using the individual PET acquisitions and urine radioactivity values, parameters of renal radioactivity were calculated as detailed in Table 1; in particular, corrected regional [18F]FDG uptake rates were obtained by subtracting late tubular [18F]FDG radioactivity from the ROI activity.
Table 1Definition of parameters extracted from [18F]FDG-PET acquisitions.
Parameter
Description
Calculation
Units
PET period
PET period, from [18F]FDG injection to urine collection
PET period, from [18F]FDG injection to urine collection
min
Urine period
Period of urine collection, from pre-study urination to post-PET voiding
], and expressed per kilogram of fat-free mass (μmol·kgFFM−1·min−1), because this normalization minimizes differences due to sex, age, and body weight [
Data are summarized as mean ± SD or median and interquartile range (IQR) for variables with non-normal distribution by Wilks-Shapiro test. Group comparisons were carried out using Mann-Whitney U test for unpaired observations. Paired comparisons by group were performed by repeated-measure ANOVA. A p < 0.05 was considered significant. Analyses were done using JMP version 13.0 (SAS Institute, Cary, NC, USA). Images were created using ggplot package on R Studio [
The lean and obese study groups were well-balanced for sex and age (Table 2). The study groups consisted predominantly of women with data of three men (one lean and two obese) available in the fasting and insulin clamp experiments. As expected, obese subjects had markedly impaired whole-body insulin sensitivity, as indicated by the M value. Skeletal muscle GU was similar in the fasting state, and markedly reduced on the clamp, in the obese vs lean. Total eGFR (mL/min) was higher in the obese subjects.
Table 2Clinical and metabolic characteristics of the study subjects.
Lean
Obese
p
M/W
1/8
2/14
ns
Age (years)
48 ± 6
47 ± 9
ns
Body weight (kg)
67 [13]
118 [18]
<0.0001
BMI (kg/m2)
23.3 [3.5]
43.2 [4.5]
<0.0001
M value (μmol·min−1·kg−1)
41.7 [19.3]
8.4 [2.8]
<0.0001
eGFR (mL/min)
96 ± 12
132 ± 17
0.0007
Plasma glucose (mmol/L)
5.4 ± 0.3
6.0 ± 0.9
ns
Plasma insulin (pmol/L)
30 [6]
90 [70]
0.004
Fasting muscle FUR (1/min)
0.002 [0.001]
0.002 [0.001]
ns
Fasting muscle GU (μmol·min−1·100 mL−1)
1.04 [0.30]
0.97 [0.29]
ns
Clamp muscle FUR (1/min)
0.013 [0.015]
0.004 [0.008]
0.004
Clamp muscle GU (μmol·min−1·100 mL−1)
5.42 [5.77]
1.59 [2.97]
0.006
Entries are mean ± SD or median [interquartile range]; p value for the comparison obese vs lean subjects.
3.1 Urinary [18F]FDG excretion and tubular concentration (Table 3)
In the fasting state, total urine decay-corrected [18F]FDG radioactivity averaged 18 % of the injected dose over ∼120 min, similarly in lean and obese subjects. Urinary [18F]FDG clearance averaged ∼35 % of eGFR in lean individuals and ∼40 % in the obese. Mean tubular [18F]FDG flow averaged ∼240 kBq/min, resulting in estimated mean and late tubular [18F]FDG activity of 70 and 45 kBq/mL, without differences between lean and obese. Renal tissue [18F]FDG clearance was 10–15 times smaller than urinary [18F]FDG clearance, and was significantly impaired in the obese compared to the lean group. On the clamp day, a smaller fraction of the injected [18F]FDG dose was recovered in the urine than on the fasting day, and more so in the lean. In contrast, urinary [18F]FDG clearance was similar between fasting and insulin in either group. Tubular [18F]FDG flow was smaller with insulin compared to fasting, as were measures of [18F]FDG tubular activity. These differences can be attributed to a greater sequestration of the [18F]FDG dose in insulin-stimulated tissues such as skeletal muscle, and the consequent reduction of average plasma [18F]FDG activity. In line with this, tubular flow, mean and tubular late activity were all inversely related to the M value (Supplemental Fig. 1). Of note, insulin-stimulated renal tissue [18F]FDG clearance, while still lower in the obese, was only slightly – and statistically non-significantly – higher than under fasting conditions.
In the fasting state, both cortical and medullary [18F]FDG FUR and uptake rates – whether using corrected or uncorrected values – were higher in lean than obese subjects. The pattern of differences was similar in the clamp experiments (Fig. 2A ). These results held true also when accounting for eGFR for all regions (p < 0.05), except for fasting medullary GU which only trended to be higher in healthy controls compared to patients with obesity (p = 0.07). Insulin appeared to have no stimulatory effect on either cortical or medullary [18F]FDG FUR and uptake rates when using uncorrected values. However, when these rates were corrected for the individual tubular [18F]FDG radioactivity value, the effect of insulin became clear, and statistically significant, for cortical but not medullary ROIs in both groups without canceling the difference in the GU rates between the two groups (73 % increase in the lean participants).
p < 0.05 or less for the paired comparison of medullary vs cortical in both groups (repeated measures ANOVA).
0.002
Entries are median [interquartile range]; FUR = fractional uptake rate; GU = glucose uptake; Corr. GU = value corrected for tubular radioactivity; p value for the group comparison obese vs lean subjects. Complete urinary data were available in 20 subjects during insulin clamp and 13 subjects during fasting conditions.
# p < 0.05 or less for the paired comparison of insulin vs fasting in both groups (repeated measures ANOVA).
§ p < 0.05 or less for the paired comparison of medullary vs cortical in both groups (repeated measures ANOVA).
Fig. 2Bar graph showing corrected cortical and medullary glucose uptake rates in the two groups under two metabolic conditions. Data are mean ± SEM. *p < 0.05 in the comparison between obese and lean participants, #p < 0.05 for the comparison between fasting and insulin clamp and §p < 0.05 for the comparison between cortex and medulla (A). In the pooled fasting (blue circles) and insulin clamp (red circles) data, corrected cortical (B) but not medullary GU (C) correlates with skeletal muscle glucose uptake.
In the pooled data, [18F]FDG uptake was well correlated with skeletal muscle [18F]FDG uptake in the cortex but not the medulla (Fig. 2B–C).
4. Discussion
The general conclusion of the present study is that the [18F]FDG-PET technique can be used to study regional renal glucose uptake on the condition that urine is carefully collected and its [18F]FDG radioactivity measured, and that these measurements are used to correct tissue [18F]FDG uptake for tubular radioactivity. In particular, late acquisition of renal radioactivity (∼90 min following [18F]FDG injection in the present study) is mandatory in order to minimize radioactivity inside the tubuli. Then, the only parameter necessary for the method is total tubular volume, which had to be assumed as we know of no way to measure tubular volume in vivo directly. The fraction we used, 5 % of ROI volume, has been determined by anatomical studies [
Report of the Task Group on Reference Man. A Report Prepared by a Task Group of Committee 2 of the International Commission on Radiological Protection.
]. In our analysis, this value is used as a constant across subjects (lean or obese) and physiological condition (fasting or euglycemic hyperinsulinemia).
The data used for this study yielded several outcomes. With regard to the urinary parameters, urinary [18F]FDG clearance – averaging 35 mL/min or 1/3 of eGFR in our lean subjects – is in the same range (31 mL/min) as that measured for glucose in a group of healthy individuals in whom SGLT reabsorptive capacity had been saturated by clamping plasma glucose at 22 mmol/L [
]. This finding can be taken as direct evidence that in humans 2-FDG has minimal affinity for SGLTs, as previously demonstrated in mice transfected with human SGLTs and GLUT2 [
]. Therefore, 2-FDG uptake into renal tissues must occur from the peritubular circulation through the GLUT2 transporter that is expressed on the basolateral membrane of the proximal convoluted tubule, for which 2-FDG has even higher affinity than d-glucose [
]. Of note, transport of d-glucose from the luminal side to the circulation via GLUT2 is down the concentration gradient created by SGLT-mediated glucose absorption, while transport of 2-FDG via GLUT2 is down an inverse gradient given by the minimal 2-FDG concentration in the cell.
Given the differential transport route of d-glucose and 2-FDG, it is expected that values for renal glucose uptake obtained by the catheter/tracer glucose technique should be different from those calculated by [18F]FDG-PET. To gauge this difference, we compiled the published human studies that have used the catheter/tracer technique (Table 5). The values for total renal glucose uptake in 6 studies in healthy controls in the fasting state show substantial interstudy variability. The corresponding [18F]FDG-PET estimate (obtained by multiplying the renal tissue clearance in Table 3 by the mean fasting glycemia) is 14 μmol/min, which seems to be a considerable underestimate of the catheter data. Importantly, for skeletal muscle the two techniques (Table 5, Table 1) measure very close values in the fasting state (0.84 vs 1.04 μmol·min−1·100 mL−1) as well during similar euglycemic hyperinsulinemia (6.5 vs 5.4 μmol·min−1·100 mL−1); the same has been previously reported for myocardial muscle [
]. Clearly, the physiological differences between muscle (GLUT4-mediated uptake of blood-borne glucose) and kidney (GLUT2-mediated 2-FDG uptake from the peritubular capillary network) account for the discrepant quantitative performance of 2-FDG in the two tissues. Another mechanism leading to an underestimation of renal glucose uptake by [18F]FDG might be dephosphorylation of “trapped” [18F]FDG, given that the kidneys express the enzyme glucose-6-phosphatase. However, this process cannot be identified – let alone quantified – in vivo, and therefore remains a theoretical issue.
Table 5Renal and skeletal muscle rates of glucose uptake as measured by the AV difference/tracer technique.
Unlike the catheter method, however, [18F]FDG-PET can provide quantitative estimates of regional glucose uptake rates in the kidney. Using the corrected uptake rates (Table 4), our analysis yields relevant physiological findings. Firstly, in medullary ROIs [18F]FDG uptake was higher than in cortical ROIs, in agreement with in vitro results [
], showing that the medulla is a preferential user of glucose as an energy substrate while in the cortex free fatty acid use and gluconeogenesis predominate [
]. Secondly, physiological hyperinsulinemia stimulated cortical but not medullary [18F]FDG uptake, in line with evidence that renal ATP is higher in the cortex than in the medulla [
]. Importantly, the 73 % increment of cortical [18F]FDG uptake with insulin is of similar magnitude as the 84 % increase estimated by the catheter/tracer technique for total renal glucose uptake [
] (Table 5). Thirdly, [18F]FDG uptake was markedly reduced in the obese group in both cortex and medulla (Table 4). To a first approximation, this finding is in agreement with the fact that the expression of insulin receptors and the levels of tyrosine phosphorylated receptors are reduced in the renal cortex of insulin resistant rat models [
]. However, insulinization did not significantly stimulate medullary [18F]FDG uptake (Fig. 2), yet medullary [18F]FDG uptake was significantly reduced in the obese both during fasting and following insulin. Obesity is a risk factor for chronic kidney disease [
To the best of our knowledge, this study is the first systematic attempt to assess whether [18F]FDG-PET can yield useful clinical information regarding renal glucose metabolism. However, several limitations must be acknowledged. First, the volumes of fluid (saline and intravenous glucose infusions) given during the studies, and the salt intake of the subjects were not recorded; aggressive hydration has been shown to enhance [18F]FDG elimination in the urine [
]. Second, even though it has been described that patients with obesity and proteinuria have epithelial hypertrophy and increased tubular urinary spaces compared to lean subjects [
], we applied a 5 % correction for the intraluminal tubular volume for both groups, since there are no other published estimates of this quantity in patients with obesity. At any rate, a fixed correction would underestimate rather than overestimate the reported differences in renal glucose uptake values between patients with obesity and lean controls. Also, a more rigorous modelling approach for the estimation of the renal [18F]FDG dynamics would have required simultaneous acquisitions of aorta, kidney and bladder; this was not possible in the present study, but may be possible with the use of the new state-of-the-art PET scanners (FOV ∼1.1 m). These limitations could help planning future studies investigating renal glucose metabolism in vivo, in addition to suggesting the use of glucose analogs, such as 6-fluoro-6-deoxy-d-glucose, that are transported through SGLTs and could therefore closely trace renal glucose transport [
]. It must also be acknowledged that not only is the catheter/tracer technique invasive, but that it is fraught with high variability as null tracer exchange across the kidney has been reported [
]. Under fasting conditions, their renal glucose uptake rates were substantially smaller than what has been previously reported in the literature and in closer agreement to our PET measurements (Table 5). Finally, because of the close proximity of the cortical and medullary ROIs some interference between the cortical and medullary results cannot be excluded – even though this would affect the current results minimally.
In conclusion, albeit 2-FDG does not fully trace native glucose, our ‘advanced’ renal [18F]FDG-PET methodology here applied on patients with obesity and healthy lean controls provides strong support for the notion that the human renal cortex is an insulin sensitive tissue; additional metabolic information can be at hand.
The following is the supplementary data related to this article.
During the insulin clamp, tubular flow (A), mean tubular activity (B) and late tubular activity (C) were inversely related to the M value.
CRediT authorship contribution statement
ER analysed the data and drafted the manuscript. VO and AM analysed the data. PN acquired the data. EF analysed the data and critically revised the text. All authors approved the final version of the text. PN and EF are the guarantors of this work.
Conflict of interest
None.
Acknowledgments
The authors thank the staff of the Turku PET Centre for performing the PET imaging.
Funding
The study was conducted within the Center of Excellence in Cardiovascular and Metabolic Diseases, supported by the Academy of Finland, the University of Turku, Turku University Hospital, Åbo Akademi University, Finnish Diabetes Foundation, Sigrid Jusélius Foundation. ER reports funding from the Emil Aaltonen Foundation, the Finnish Cultural Foundation, the Paulo Foundation, the Maud Kuistilan Muistosäätiö, the Finnish Diabetes Research Foundation, and from the Finnish Medical Foundation.
Ethical approval
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
References
Owen O.E.
Felig P.
Morgan A.P.
Wahren J.
Cahill G.F.J.
Liver and kidney metabolism during prolonged starvation.
SLEEVEPASS: a randomized prospective multicenter study comparing laparoscopic sleeve gastrectomy and gastric bypass in the treatment of morbid obesity: preliminary results.
Brain glucose uptake is associated with endogenous glucose production in obese patients before and after bariatric surgery and predicts metabolic outcome at follow-up.
Report of the Task Group on Reference Man. A Report Prepared by a Task Group of Committee 2 of the International Commission on Radiological Protection.
p value for the difference between obese and lean individuals.
00260-8&id=ga1.jpg){kind=link}
00260-8&id=gr1.jpg){kind=link}
00260-8&id=gr2.jpg){kind=link}