Insulin sensitivity index in type 1 diabetes and following human islet transplantation: comparison of the minimal model to euglycemic clamp measures

Abstract

Insulin sensitivity is impaired in type 1 diabetes (T1D) and may be enhanced by islet transplantation, an effect best explained by improved metabolic control. While the minimal model index of insulin sensitivity, S_I, has been used in studies of T1D, it has not before been evaluated against gold-standard measures derived from the euglycemic clamp. We sought to determine how well minimal model S_I derived from an insulin-modified frequently sampled intravenous glucose tolerance (FSIGT) test compared with total body and peripheral insulin sensitivity estimates derived from the hyperinsulinemic-euglycemic clamp in subjects with T1D and following islet transplantation. Twenty-one T1D subjects were evaluated, including a subgroup (n = 12) studied again after intrahepatic islet transplantation, with results compared with normal controls (n = 11 for the FSIGT). The transplant recipients received 9,648 ± 666 islet equivalents/kg with reduction in HbA_1c from 7.1 ± 0.2 to 5.5 ± 0.1% (P < 0.01) and 10/12 were insulin independent. FSIGT-derived S_I was reduced in T1D pre- compared with posttransplant and with normal [1.76 ± 0.45 vs. 4.21 ± 0.34 vs. 4.45 ± 0.81 × 10⁻⁴(μU/ml)⁻¹·min⁻¹; P < 0.01 for both]. Similarly, clamp-derived total body, and by the isotopic dilution method with [6,6-²H₂]glucose, peripheral insulin sensitivity increased in T1D from pre- to posttransplant (P < 0.05 for both). The predictive power (r²) between volume-corrected S_IC and measures of total and peripheral insulin sensitivity was 0.66 and 0.70, respectively (P < 0.00001 for both). That the minimal model S_IC is highly correlated to the clamp-derived measures indicates that the FSIGT is an appropriate methodology for the determination of insulin sensitivity in T1D and following islet transplantation.

insulin resistance is an underappreciated feature of type 1 diabetes (T1D) (7, 13, 32, 33), resulting at least in part from the absence of endogenous insulin secretion, frequent periods of sustained hyperglycemia (12, 21, 31), and impaired sensitivity of lipolysis to inhibition by insulin, resulting in elevated free fatty acids (FFA) (24, 33). Islet transplantation restores endogenous insulin secretion that can correct the hyperglycemia of T1D and normalize FFA metabolism (15, 20, 27), effects that have been associated with improved minimal model indexes of insulin sensitivity derived from an insulin-modified frequently sampled intravenous glucose tolerance (FSIGT) test in T1D recipients of islet transplants. More recently, employing the hyperinsulinemic-euglycemic clamp technique with a stable glucose isotope, our group has confirmed this correction of impaired insulin sensitivity in T1D by islet transplantation and shown that the improvement is mediated by effects at the liver and skeletal muscle (18). Extension of these findings to larger populations and to other therapeutic interventions for T1D may be more readily accomplished using the FSIGT test that can be more easily standardized across clinical sites.

Whereas the minimal model index of insulin sensitivity, S_I, can be derived in T1D subjects using an insulin-modified FSIGT with good parameter resolution (10, 20, 28), S_I has not before been evaluated against measures of insulin sensitivity derived from the gold-standard hyperinsulinemic-euglycemic clamp in T1D as has been performed in subjects with type 2 diabetes (T2D) (22). Because the development of absolute insulin deficiency in T1D is distinct from the relative insulin deficiency present in T2D, a comparative analysis of clamp and minimal model-derived indexes of insulin sensitivity in T1D subjects is critical to the ongoing application of the minimal model approach to the study of insulin sensitivity in this population. The minimal model-derived S_I does not distinguish between insulin action to suppress glucose production (primarily from the liver) and that to enhance peripheral glucose disposal (primarily in skeletal muscle), which requires isotope tracer methodology but does provide an estimate of total body insulin sensitivity that is believed to largely represent peripheral insulin action needed for disposal of the injected glucose load. We sought to determine how well the FSIGT minimal model-derived S_I compared with total body and peripheral insulin sensitivity estimates derived from the hyperinsulinemic-euglycemic clamp with a stable glucose isotope in subjects with T1D and following islet transplantation.

MATERIALS AND METHODS

T1D subjects included those with long-standing C peptide-negative disease complicated by hypoglycemia unawareness and frequent severe hypoglycemia events who had normal kidney function and were initially considered as potential candidates for islet alone transplantation (n = 21). A subgroup of these subjects received intrahepatic islet transplants (n = 12) as part of the Clinical Islet Transplantation (CIT) Consortium protocols being conducted at the University of Pennsylvania (1). The study protocols were approved by the Institutional Review Board of the University of Pennsylvania, and all subjects gave their written informed consent to participate.

The islet transplant recipients included all 11 subjects participating in the CIT07 protocol from our institution that included thymoglobulin and etanercept for induction immunosuppression (19) and one of two participants who experienced early islet graft failure with the CIT05 protocol consisting of thymoglobulin and rituximab for induction and was later retransplanted under an Edmonton protocol with basiliximab induction. The transplant recipients underwent one or two intraportal infusions of islets to achieve insulin independence. Maintenance immunosuppression consisted of low-dose tacrolimus (12-h blood trough target 3–6 μg/l) and sirolimus (24-h blood trough target 10–15 μg/l for the first 3 mo and 8–12 μg/l thereafter).

Healthy nondiabetic control subjects for the FSIGT (n = 11) were selected for gender, age, and body mass index (BMI) from other studies conducted by our group (23, 25) to match that of the T1D subjects.

Metabolic studies.

All 21 T1D subjects underwent FSIGTs and hyperinsulinemic-euglycemic clamps between 2 days and 1 mo apart. When performed the same week, the FSIGT was conducted on the first day, and menstruating women underwent both tests during the follicular phase of their menstrual cycle (26). The 12 subjects who underwent islet transplantation each repeated the FSIGT between 2.5 and 7 mo posttransplant and the euglycemic clamp between 6 and 7 mo posttransplant. The median (interquartile range) of time between the posttransplant FSIGT and euglycemic clamp studies was 3 (2.5–3.5) mo. For both the euglycemic clamp and FSIGT, T1D subjects and insulin-dependent islet transplant recipients were admitted to the University of Pennsylvania Clinical and Translational Research Center (CTRC) the afternoon before study and fasted overnight after 2000 for 12 h before testing. At 2100 subjects were converted from subcutaneous insulin to a low-dose intravenous insulin infusion protocol to target blood glucose at 81–115 mg/dl. Insulin-independent islet transplant recipients and normal controls had the option of fasting overnight either in the CTRC or at home with arrival in the CTRC by 0630 the morning of study. By 0700, one catheter was placed in an antecubital vein for infusions, and one catheter was placed in a hand vein for blood sampling, with the hand placed in a thermoregulated box (∼50°C) or heating pad to promote arterialization of venous blood.

FSIGT.

When used overnight, the insulin infusion was discontinued 20 min before testing. All other medications were withheld until later in the morning. After baseline blood sampling at −10, −5, and −1 min, 0.3 g/kg of 50% glucose was injected over 1 min starting at time (t) = −30 s, and 0.03 U/kg of insulin (1 U/1 ml solution) was injected over 30 s starting at t = 20 min. Additional blood samples were collected at t = 1, 2, 3, 4, 5, 7, 10, 12, 14, 16, 18, 20, 22, 25, 30, 40, 50, 70, 100, 140, and 180 min after the injection of glucose (20, 22, 26). All samples were centrifuged at 4°C, separated, and frozen at −80°C for subsequent analysis. Serum glucose was measured by the glucose hexokinase method using an automated glucose analyzer (Roche Module P; Roche Diagnostics, Indianapolis, IN) and serum insulin by two-site immune-enzymometeric assay using a Tosoh 2000 autoanalyzer (Tosoh Biosciences, San Francisco, CA) at the Northwest Lipid Research Laboratory (University of Washington, Seattle, WA).

Euglycemic clamp.

At t = −120 min a primed (5 mg/kg · fasting plasma glucose in mg/dl ÷ 90 mg/dl over 5 min) continuous (0.05 mg·kg⁻¹·min⁻¹ for 355 min) infusion of the stable glucose isotope tracer [6,6-²H₂]glucose (99% enriched; Cambridge Isotopes Laboratories, Andover, MA) was administered to assess endogenous glucose production before and during the induction of hyperinsulinemia (4, 18). When used overnight, the insulin infusion was continued during this period to maintain stable normoglycemia until t = 0. After baseline blood sampling at −20, −10, and −1 min, at t = 0 min a continuous infusion of insulin was initiated at 1 mU·kg⁻¹·min⁻¹ for 240 min to produce hyperinsulinemia. Subsequently, a variable-rate infusion of 20% glucose was administered according to the glycemic clamp technique (8) to maintain the plasma glucose ∼90 mg/dl. To reduce changes in plasma enrichment of [6,6-²H₂]glucose during the clamp, the 20% glucose solution was enriched to 2.0% with [6,6-²H₂]glucose (4). Morning immunosuppression medications were taken after baseline blood sampling if applicable. Blood samples were taken every 5 min, centrifuged, and measured at bedside with an automated glucose analyzer (YSI 2300; Yellow Springs Instruments, Yellow Springs, OH) to adjust the glucose infusion rate and achieve the desired plasma glucose concentration. Additional blood samples were taken every 20 min for biochemical analysis. All samples were collected on ice in chilled tubes containing EDTA and Protease Inhibitor Cocktail (Sigma-Aldrich, St. Louis, MO), centrifuged at 4°C, separated, and frozen at −80°C for subsequent analysis. Plasma glucose was verified in duplicate by the glucose oxidase method using an automated glucose analyzer (YSI 2300; Yellow Springs Instruments), and plasma insulin was measured in duplicate by double-antibody radioimmunoassay (Millipore, Billerica, MA) as previously described (18). Enrichment of [6,6-²H₂]glucose was measured by gas chromatography-mass spectrometry at Metabolic Solutions (Nashua, NH) or the Metabolic Tracer Resource at the University of Pennsylvania.

Calculations and statistics.

Basal levels of glucose and insulin were calculated from the mean of the baseline samples preceding t = 0 for the FSIGT. Intravenous glucose tolerance was evaluated from the FSIGT by the glucose disappearance rate K_g = ln[glucose]/min × 100, calculated as the slope of the natural log of glucose values between 10 and 40 min with least-squares linear regression (29) using Origin software (Northampton, MA). The FSIGT parameters for the acute insulin response to the injection of glucose (AIR_g), insulin sensitivity (S_I), disposition index (D_I = AIR_g × S_I), and glucose effectiveness (S_G) were derived from Bergman's minimal model using MINMOD Millennium software (6) as previously described (20, 25, 26).

The rate of appearance (R_a) of glucose during the euglycemic clamps was calculated using Steele's nonsteady state equation modified for the use of stable isotopes: R_a = {F − V[(C₂ + C₁)/2][(E₂ − E₁)/(t₂ − t₁)]}/[(E₂ + E₁)/2], where C₂ and C₁ are the glucose concentrations at the times 2 and 1, respectively (in mg/ml), V is the fractional volume of distribution of glucose (40 ml/kg), F is the tracer infusion rate, and E represents the isotopic enrichment at the respective time points (30). The rate of disposal (R_d) of glucose during the euglycemic clamps was calculated using Steele's non-steady-state equation R_d = R_a − V[(C₂ − C₁)/(t₂ − t₁)]. Total body insulin sensitivity was calculated from the final hour of the euglycemic clamp as S_I(clamp) = M/(ΔI × G) where M is the glucose infusion rate (GIR), ΔI is the change in insulin concentration from basal to the steady-state condition during the final hour of insulin infusion, and G is the steady-state glucose concentration during the final hour (2). Accounting for space correction (SC), such that M = GIR − SC where SC = (G₂ − G₁) × 1.9/(t₂ − t₁) calculates the glucose removed from or added to the glucose space between each time interval where G₂ and G₁ are the glucose concentrations at the times 2 and 1, respectively (in mg/dl) (8), resulted in nearly identical measures for S_I(clamp) (r² = 0.997; P < 0.00001). Peripheral (primarily skeletal muscle) insulin sensitivity was calculated using the steady-state tracer-derived glucose disposal as S_IP(clamp) = ΔR_d/(ΔI × G) (3).

Whereas both minimal model (S_I) and clamp [S_I(clamp) and S_IP(clamp)] measures reflect insulin sensitivity that is dominated by extrahepatic effects of insulin (3), they are not directly comparable because they are expressed in different units [(μU/ml)⁻¹ per min and dl·min⁻¹·kg⁻¹ per μU/ml, respectively]. This is due to each parameter being normalized to a different measure of body size, with the volume of distribution of glucose used in the minimal model and body weight in the clamp calculations. To convert them to a common index of insulin sensitivity, S_IC, reflecting the unitary change in insulin to cause a given increment in glucose clearance (dl/min per μU/ml), we followed the procedure reported by Bergman's group (3, 22) such that S_IC(minmod) = S_I × V_D where the volume of distribution of glucose is calculated as V_D (dl) = dose of glucose injected in milligrams divided by G₀ − G_b in mg/dl, where G₀ is the peak and G_b the basal glucose derived from the minimal model, S_IC(clamp) = S_I(clamp) × body weight in kilograms, and S_ICP(clamp) = S_IP(clamp) × body weight in kilograms. Because the conversion factors used for the minimal model and clamp-derived estimates of insulin sensitivity are independent, this procedure excludes bias in the statistical comparison of the two methods for measuring S_IC (3).

All data are expressed as means ± SE. Comparison of results between pre- and posttransplant T1D subjects was performed with paired Student's t-tests or Wilcoxon Match Pairs tests for nonparametric data, and comparison of results between each T1D group and controls was performed with unpaired Student's t-tests or the Mann-Whitney U-test as appropriate using Statistica software (StatSoft, Tulsa, OK). Comparison of the minimal model and clamp-derived indexes of insulin sensitivity were performed by least-squares linear regression using Origin software. Significance was considered at P < 0.05 (2-tailed).

RESULTS

Subject characteristics.

The T1D subjects were of comparable gender distribution, age, body weight, and BMI to the control subjects (Table 1). Those T1D subjects who underwent islet transplantation had a decrease in body weight with resulting decrease in BMI (P < 0.01; Table 1), although values of both remained not different from normal. The T1D subjects had ∼30 years of disease duration and an insulin requirement ∼0.5 U·kg⁻¹·day⁻¹ that was substantially reduced (P < 0.01; Table 1) following receipt of 9,648 ± 666 islet equivalents/kg recipient body weight with 10 of the 12 islet recipients insulin independent at the time of posttransplant assessment. HbA_1c, which was elevated in T1D, was normalized in the group posttransplant (Table 1), of which seven were insulin-independent following one islet infusion, three were insulin-independent following two islet infusions, one was insulin-dependent while awaiting a second islet infusion, and one remained insulin-dependent after two islet infusions. Subjects maintained appropriate levels of tacrolimus and sirolimus (Table 1) except for one who was converted from sirolimus to mycophenolate mofetil (19).

Table 1.

Subject characteristics at the time of FSIGT testing

	T1D
	Allevaluated (n = 21)	Pretransplant (n = 12)	Posttransplant (n = 12)	Normal Controls (n = 11)
Sex (M/F)	9/12	5/7	5/7	6/5
Age, yr	46 ± 2	46 ± 3	47 ± 3*	40 ± 2
Weight, kg	70 ± 3	71 ± 3	66 ± 3*	71 ± 3
BMI, kg/m2	25 ± 1	25 ± 1	23 ± 1*	24 ± 1
HbA1c, %	6.8 ± 0.2	7.1 ± 0.2	5.5 ± 0.1*	ND
T1D duration, yr	31 ± 2	29 ± 4	31 ± 4*
Insulin use, U·kg−1·day−1	0.50 ± 0.03	0.48 ± 0.05	0.02 ± 0.02*
IE/kg			9,648 ± 666
Tacrolimus, μg/l			4.8 ± 0.4
Sirolimus, μg/l			9.2 ± 0.7†

Data are means ± SE; n, no. of subjects.

FSIGT, frequently sampled intravenous glucose tolerance; T1D, type 1 diabetes; M, males; F, females; BMI, body mass index; IE/kg, islet equivalents transplanted/kg recipient body wt whereby an IE approximates a standard islet diameter of 150 μm.

^*P < 0.01 for comparison with T1D pretransplant.

^†One subject was converted from sirolimus to mycophenolate mofetil because of the development of interstitial pneumonia 4 wk posttransplant that subsequently resolved (19).

FSIGT.

Basal glucose was higher in the T1D subjects pre- compared with posttransplant and normal (P < 0.05 for both; Table 2 and Fig. 1A) and was also higher in the posttransplant subjects than in controls (P < 0.05; Table 2 and Fig. 1A), whereas basal insulin was similar in all groups (Table 2). The V_D for the injected glucose was higher in the T1D subjects pre- compared with posttransplant and normal (P < 0.05 for both; Table 2), and K_g was slower in the T1D subjects pre- compared with posttransplant and normal (P < 0.01 for both; Table 2). Islet transplantation restored previously absent endogenous insulin secretion with the first-phase AIR_g not significantly different from normal (Table 2 and Fig. 1B). S_I was reduced in T1D pre- compared with posttransplant and normal (P < 0.01 for both; Table 2). The resulting D_I was not significantly different in the islet transplant recipients compared with controls (Table 2). S_G was also reduced in T1D pre- compared with posttransplant and with normal (P < 0.01 for both; Table 2).

Table 2.

Metabolic parameters during the FSIGT test and euglycemic clamp^a

	T1D
	All evaluated (n = 21)	Pretransplant (n = 12)	Posttransplant (n = 12)b	Normal Controls (n = 11)
Basal glucose, mg/dl	113 ± 5	117 ± 6	100 ± 3c	91 ± 2d,e
Basal insulin, μU/ml	12 ± 2	12 ± 4	11 ± 3	9 ± 1
VD, dl	170 ± 9	162 ± 13	113 ± 8d	129 ± 8c
Kg, %/min	1.05 ± 0.11	0.98 ± 0.16	2.08 ± 0.16d	2.78 ± 0.39d
AIRg, μU·ml−1·min			263 ± 58	473 ± 161
SI × 10−4, (μU/ml)−1min	2.02 ± 0.31	1.76 ± 0.45	4.21 ± 0.34d	4.45 ± 0.81d
DI			994 ± 168	1441 ± 247
SG, min−1	0.010 ± 0.002	0.009 ± 0.002	0.019 ± 0.002d	0.021 ± 0.003d
Steady-state glucose,a mg/dl	87 ± 1	88 ± 2	87 ± 1
ΔInsulin,a,f μU/ml	70 ± 7	74 ± 9	70 ± 7
Steady-state GIR,a mg·kg−1·min−1	6.8 ± 0.4	6.7 ± 0.5	8.3 ± 0.5c
ΔRd,a,f mg·kg−1·min−1	4.8 ± 0.4	4.7 ± 0.5	6.5 ± 0.6c
SI(clamp) × 102,a dl·min−1·kg−1 per μU/ml	0.13 ± 0.02	0.11 ± 0.01	0.15 ± 0.02c
SIP(clamp) × 102,a dl·min−1·kg−1 per μU/ml	0.10 ± 0.01	0.08 ± 0.01	0.12 ± 0.02c

Data are means ± SE; n, no. of subjects. V_D, volume of distribution for glucose injected during FSIGT; K_g, glucose disappearance rate; AIR_g, acute insulin response to glucose; S_I, insulin sensitivity index; D_I, disposition index defined as the product AIR_g × S_I; S_G, glucose effectiveness; GIR, glucose infusion rate; R_d, rate of glucose disposal; S_I(clamp), total body insulin sensitivity; S_IP(clamp), peripheral insulin sensitivity. To convert glucose to mmol/l, multiply by 0.05551; and to convert insulin to pmol/l, multiply by 6.

^aData from the euglycemic clamp have been previously published for the 12 T1D subjects who underwent islet transplantation (18).

^bn = 11 Posttransplant for measures of steady-state glucose, Δinsulin, steady-state GIR, ΔR_d, S_I(clamp), and S_IP(clamp).

^cP < 0.05 compared with T1D pretransplant.

^dP < 0.01 compared with T1D pretransplant.

^eP < 0.05 compared with T1D posttransplant.

^fChange in concentration from basal to the steady state.

Fig. 1.

Plasma glucose (A) and insulin (B) levels in response to the frequently sampled intravenous glucose tolerance test. Glucose (0.3 g/kg) is injected over 1 min at time (t) = 0, and insulin (0.03 U/kg) is injected over 30 s at t = 20 min. Intravenous glucose tolerance improved in type 1 diabetes (T1D) from pre- to posttransplant (P < 0.01), with posttransplant glucose disappearance not different from normal. The acute insulin response to the injection of glucose, which was absent in T1D pretransplant, was restored posttransplant to levels comparable to normal.

Euglycemic clamp.

The steady-state glucose concentration and change in insulin concentration were similar in all groups during the final hour of the clamp, with one posttransplant subject not completing the clamp because of intravenous line failure (Table 2) (18). The GIR was less pre- than posttransplant (P < 0.05; Table 2), with similar results for the change in R_d (P < 0.05; Table 2). As a result, both total body, S_I(clamp), and peripheral, S_IP(clamp), insulin sensitivity were lower in T1D pre- compared with posttransplant (P < 0.05 for both; Table 2).

When the data were adjusted for BMI using a linear mixed model, the change in BMI did not have a significant effect on the change in FSIGT or euglycemic clamp-derived insulin sensitivity measures.

Comparison of FSIGT to euglycemic clamp-derived insulin sensitivity.

Before correction, S_I(minmod) was reasonably predictive of S_I(clamp) (r² = 0.50; P < 0.001; Fig. 2A) and S_IP(clamp) (r² = 0.59; P < 0.0001; Fig. 2B) for all 21 T1D subjects, including those not transplanted (n = 9) and those pretransplant (n = 12). When considering the untransplanted subjects (n = 9) and those posttransplant (n = 11), the predictive value of S_I(minmod) was weaker for S_I(clamp) (r² = 0.19; P < 0.05; Fig. 2C) and S_IP(clamp) (r² = 0.44; P < 0.001; Fig. 2D). This is explained by distinct relationships appearing between the untransplanted and posttransplant subjects such that, when each group is considered separately, S_I(minmod) was more predictive of S_I(clamp) (r² = 0.63 and 0.62, respectively; P < 0.01; Fig. 2C) and S_IP(clamp) (r² = 0.69 and 0.66, respectively; P < 0.01; Fig. 2D) than when considering the untransplanted and posttransplant subjects together.

Fig. 2.

Correlations between total body insulin sensitivity estimated by the minimal model [S_I(minmod)], total body insulin sensitivity based on glucose infusion rate, M, from the euglycemic clamp [S_I(clamp); on the left], and peripheral insulin sensitivity based on glucose disposal rate, R_d, from the euglycemic clamp [S_IP(clamp); on the right]. The predictive relationships of S_I(minmod) to S_I(clamp) and S_IP(clamp) were stronger when considering all 21 T1D subjects (A and B) than when considering those not transplanted and those posttransplant (C and D).

Following correction to common indexes of insulin sensitivity, the common insulin sensitivity index derived from the minimal model, S_IC(minmod), was highly predictive of the common measures S_IC(clamp) (r² = 0.74; P < 0.00001; Fig. 3A) and S_ICP(clamp) (r² = 0.77; P < 0.00001; Fig. 3B) derived from the euglycemic clamp when examining all 21 T1D subjects, those not transplanted (n = 9), and those pretransplant (n = 12). This was also the case when considering individual T1D subjects who were not transplanted (n = 9) and those posttransplant (n = 11) for comparison of S_IC(minmod) with both S_IC(clamp) (r² = 0.66; P < 0.00001; Fig. 3C) and S_ICP(clamp) (r² = 0.70; P < 0.00001; Fig. 3D).

Fig. 3.

Correlations between corrected measures of S_IC(minmod) and S_IC(clamp) (on the left), and S_ICP(clamp) (on the right). The minimal model and clamp-derived measures were each independently converted to a common index of insulin sensitivity, S_IC, reflecting the unitary change in insulin to cause a given increment in glucose clearance and expressed ×10² in dl/min per μU/ml. The resulting predictive relationships of S_IC(minmod) to S_IC(clamp) and S_ICP(clamp) were robust both when considering all 21 T1D subjects (A and B), and when considering those not transplanted and those posttransplant (C and D).

When examining the change in uncorrected insulin sensitivity measures in the transplanted subjects (n = 11), ΔS_I(minmod) correlated with ΔS_I(clamp) (r = 0.61; P < 0.05; Fig. 4A) and more weakly with ΔS_IP(clamp) (r = 0.50; P = 0.07; Fig. 4B). For the change in corrected insulin sensitivity measures from pre- to posttransplant (n = 11), ΔS_IC(minmod) correlated with both ΔS_IC(clamp) (r = 0.60; P < 0.05; Fig. 4C) and with ΔS_ICP(clamp) (r = 0.63; P < 0.05; Fig. 4D). Similar to the results with S_I(minmod) (Table 2), after adjusting for V_D of the injected glucose, insulin sensitivity measured as S_IC(minmod) was lower in T1D pre- compared with posttransplant and with normal (2.71 ± 0.63 vs. 4.72 ± 0.44 vs. 5.36 ± 0.88 dl/min per μU/ml; P < 0.05 for both).

Fig. 4.

A and B: correlations between the changes from pre- to posttransplant in uncorrected measures of ΔS_I(minmod) and ΔS_I(clamp) (A), and ΔS_IP(clamp) (B). C and D: correlations between the changes from pre- to posttransplant in the corrected measures of ΔS_IC(minmod) and ΔS_IC(clamp) (C) and ΔS_ICP(clamp) (D).

DISCUSSION

The present study is the first to demonstrate a high correlation of the minimal model-derived S_IC to euglycemic clamp-derived measures of insulin sensitivity in T1D and islet transplant recipients. These results confirm prior reports of improved insulin sensitivity following islet transplantation for T1D that used minimal model-derived estimates from the FSIGT (15, 20) and validate the minimal model-derived S_I as a measure of both total body and peripheral insulin sensitivity in T1D. In fact, the correlations shown here were slightly stronger with clamp-derived peripheral insulin sensitivity, S_IP(clamp), based on tracer-calculated glucose disposal, R_d, than clamp-derived total body insulin sensitivity, S_I(clamp), based on glucose infusion rate. Importantly, the relationships between the minimal model and clamp-derived estimates were stronger when considering the volume-corrected S_IC, particularly when evaluating untransplanted and posttransplant subject together, an effect explained by the different V_D for the injected glucose between these groups. Indeed, the S_IC measure is more appropriate than the standard S_I when comparing T1D subjects with varying V_D, since similar relationships were present using S_IC whether considering only T1D subjects or including those postislet transplantation.

Prior reports from our group and others have established that the minimal model can adequately resolve the parameter S_I in T1D (10, 20, 28), despite a late rise in glucose during the FSIGT as seen in our study (Fig. 1). Other studies involving T2D have encountered estimates of S_I equal to zero, a problem avoided in the present work we believe by the use of an overnight insulin infusion targeting near-normal glycemia that was present at the start of testing. It is important to note that the minimal model S_I underestimates insulin sensitivity as measured by the clamp when the values were normalized to identical units with the slope for correlations of S_IC(minmod) against S_IC(clamp) and S_ICP(clamp) both being around 0.5 (Fig. 2). A similar underestimation was also reported for the analysis comparing the minimal model and clamp-derived measures of insulin sensitivity in subjects that ranged from normal with those with impaired glucose tolerance, including overt T2D (22). This underestimation is likely explained by the difference in the effect on glucose disposal of an insulin bolus, as given with the FSIGT, compared with a continuous infusion of insulin during the euglycemic clamp since insulin-stimulated glucose disposal increases with time (9). This same difference in insulin administration may also explain the greater magnitude of increase in insulin sensitivity seen with the minimal model than with the euglycemic clamp measures and suggests that the FSIGT may be a more sensitive methodology for identifying changes in insulin sensitivity in T1D. Given these differences in performance characteristics, the minimal model S_I should still be regarded as an index of insulin sensitivity and not be directly substituted for direct measures obtained from euglycemic clamp studies.

This is the first study to show that the impaired S_G seen in T1D (10, 28) can be normalized following islet transplantation. The low S_G derived from the FSIGT in T1D may in part be a consequence of absent insulin secretory function, since greater AIR_g in normal subjects may lead to an overestimation of S_G by the minimal model approach (11). In our prior study of islet transplant recipients where S_G was less than normal, the AIR_g was significantly reduced (20). In the present study, AIR_g was not different from normal, although individual subjects did have reduced responses, and so demonstrates that, in the presence of near-normal β-cell function, there is no impairment of S_G in islet transplant recipients. Glucose effectiveness is comprised of peripheral and hepatic components, with glucose autoregulation of glucose production by the liver contributing importantly to overall endogenous glucose production (5). Because glucose effectiveness describes the capacity for glucose to mediate its own disposal independent from insulin, the normalization of S_G reported here provides additional evidence to the already reported improvements in hepatic and peripheral insulin sensitivity (18) against a detrimental effect of the intrahepatic site of transplantation or immunosuppression regimen on the maintenance of glucose homeostasis in islet recipients.

In conclusion, the minimal model-derived index of total body insulin sensitivity, S_IC, is highly predictive of clamp-derived measures of both total and peripheral insulin sensitivity in T1D and following islet transplantation. When employing the FSIGT to generate the minimal model S_I, simultaneous generation of S_G is possible, with both indexes impaired in T1D and capable of correction as shown here with islet transplantation. The FSIGT parameter AIR_g also provides a measure of islet graft β-cell function that is predictive of insulin requirements (14) and so may also be of interest during postislet transplant metabolic assessment. Thus, our data indicate the usefulness of the minimal model for longitudinal or cross-sectional studies in T1D populations when performance of the euglycemic clamp is not feasible for economic or practical reasons. Finally, the FSIGT may be the desired methodological approach in T1D for additional consideration of potential therapeutic effects on glucose effectiveness and β-cell function.

GRANTS

This work was performed as a project of the Clinical Islet Transplantation Consortium, a collaborative clinical research program headquartered at the National Institute of Diabetes and Digestive and Kidney Diseases and the National Institute of Allergy and Infectious Diseases and supported by National Institutes of Health Research Grants U01-DK-070430 (to A. Naji), R01-DK-091331 (to M. R. Rickels), UL1-TR-00003 (Penn Clinical & Translational Research Center), and P30-DK-19525 (Penn Diabetes Research Center).

DISCLOSURES

No potential conflicts of interest relevant to this article were reported.

AUTHOR CONTRIBUTIONS

Author contributions: M.R.R., K.L.T., and A.N. conception and design of research; M.R.R., C.F., and C.D.-B. performed experiments; M.R.R., S.M.K., J.F.F., M.P.R., K.L.T., and A.N. analyzed data; M.R.R., S.M.K., C.F., J.F.F., M.P.R., K.L.T., and A.N. interpreted results of experiments; M.R.R. and S.M.K. prepared figures; M.R.R. drafted manuscript; M.R.R., S.M.K., C.F., C.D.-B., J.F.F., M.P.R., K.L.T., and A.N. edited and revised manuscript; M.R.R., S.M.K., C.F., C.D.-B., J.F.F., M.P.R., K.L.T., and A.N. approved final version of manuscript.