Insulin resistance assessed by short insulin tolerance test and its association with obesity and insulin resistance-related parameters in humans

The aim of this study was to examine the association of insulin resistance (evaluated by the short insulin tolerance test [SITT]) with parameters related to obesity and insulin resistance. We prospectively recruited controls and patients with type 2 diabetes mellitus (T2DM), subjected them to the SITT, and calculated the K indices of the intravenous insulin tolerance test (KITT(iv)) and the subcutaneous insulin tolerance test (KITT(sc)). We compared KITT(iv) between the volunteers and patients, and examined its correlation with KITT(sc). We also examined the association of KITT(iv) with obesity, insulin resistance-related parameters, and the insulin dose required for glycemic control. A total of 24 participants (seven controls and 17 patients with T2DM) were studied. The mean KITT(iv) was significantly lower in patients with T2DM than in the controls (2.5%{+/-}2.1% vs. 4.5{+/-}1.8%). In all participants, KITT(iv) was significantly correlated with the HOMA-IR values (r=-0.601, p<0.05) but not with KITT(sc) (p=0.62). KITT(iv) was correlated positively with the serum adiponectin concentration, but negatively with the visceral fat area and serum concentrations of tumor necrosis factor- and branched-chain amino acids. In patients with T2DM, KITT(iv) and HOMA-IR values were significantly correlated with the total insulin dose required for glycemic control. Insulin resistance evaluated using intravenous insulin (KITT(iv)) was correlated with the HOMA-IR values, but not with the resistance evaluated using subcutaneous insulin (KITT(sc)). The degree of insulin resistance was associated with biomarkers, such as adiponectin, and with the dose of insulin required for glycemic control.


Introduction
Insulin resistance is a cardinal feature of the pathogenesis of non-insulin-dependent diabetes mellitus [1].Precise estimation of insulin resistance is important to understand the mechanism underlying the impairment of glycemic control and accordingly optimize the treatment for patients with diabetes.
The hyperinsulinemic euglycemic glucose clamp test (HEGCT) is the gold standard for evaluating insulin resistance [2].However, it is complicated and difficult to perform in many patients.The homeostasis model assessment for insulin resistance (HOMA-IR value) is another test that is frequently used for evaluating the insulin resistance.HOMA-IR can be performed readily and is well correlated with the HEGCT [3].However, in patients with diabetes mellitus with high fasting plasma glucose (FPG) levels (who may have a decreased insulin secretion), it is less accurate as an indicator of insulin resistance [4].In addition, HOMA-IR cannot estimate insulin resistance in patients with diabetes who are receiving insulin therapy [4].
The short insulin tolerance test (SITT) is a simple and convenient method for estimating insulin resistance [5,6].In the SITT, regular insulin is administered intravenously, and blood samples are collected sequentially for 15 min; the plasma glucose disappearance rate is then estimated [5,6].The K index of the insulin tolerance test (K ITT ) is calculated from the linear slope of the plasma glucose concentration curve and correlates well with the HEGCT [3,7].The SITT can assess whole-body insulin sensitivity even in patients with decreased insulin secretion or on insulin therapy.However, the SITT carries a risk of hypoglycemia [5,8], and K ITT estimation requires multiple blood sampling.Theoretically, K ITT can be estimated more easily by administering insulin subcutaneously without venous blood sampling, e.g., finger pricking, and using the glucose concentration in finger-prick blood.Furthermore, rapid hypoglycemia is less likely to occur with subcutaneous insulin administration than with intravenous insulin administration.However, the accuracy of K ITT has not been validated.
Previous studies have provided the K ITT values from intravenous SITT (K ITT (iv)) in humans [3,5,6,7,9].Especially in patients with poorly controlled diabetes mellitus, K ITT (iv) is considered more accurate than HOMA-IR because it is not affected by a decreased insulin secretory capacity.However, few studies [10] have examined the association between K ITT (iv) and other parameters related to obesity and insulin resistance (such as visceral fat, adiponectin, and tumor necrosis factor [TNF]-α).Such data will improve the understanding of the pathogenesis of insulin resistance and help optimize the treatment in each patient.Furthermore, the application of K ITT (iv) for predicting the amount of insulin required would be useful in practice.
In the present prospective study, we calculated K ITT (iv) in healthy volunteers and patients with type 2 diabetes mellitus (T2DM).Using KITT(iv) as a reference, we then evaluated the accuracy of HOMA-IR and the K ITT obtained using glucose concentration of fingertip blood (i.e., after subcutaneous insulin injection [KITT(sc)]).We also investigated the relationship between K ITT (iv) and the parameters related .to obesity and insulin resistance, and examined the possible role of K ITT (iv) in predicting the insulin dose required for glycemic control in patients with T2DM.

Study participants
This was a single-center, randomized, crossover study with 17 patients with T2DM (who were admitted to the Hokkaido University Hospital for glycemic control) and seven healthy volunteers (controls).The registration period was from November 2016 to July 2019.
The inclusion criteria for the controls were as follows: age ≥30 years; body mass index, >18.5 kg/m 2 ; and no previous diagnosis of diabetes or impaired glucose tolerance.The inclusion criteria for the patients were as follows: age ≥ 20 years, T2DM diagnosis based on the American Diabetes Association Criteria [11], and FPG level ≥ 140 mg/dL during hospitalization.The exclusion criteria for the controls were as follows: history of hypoglycemia (blood glucose < 60 mg/dL), untreated ischemic heart disease, epilepsy, pregnancy, possible pregnancy, lactation, and researcher-determined ineligibility.The exclusion criteria for the patients were as follows: unstable diabetic retinopathy, stage ≥4 diabetic nephropathy, untreated ischemic heart disease, history of hypersensitivity to human regular insulin, epilepsy, pregnancy, possible pregnancy, lactation, and deemed ineligible by the researcher.

Crossover assignment to the insulin tolerance test with intravenous or subcutaneous insulin
After case enrollment using the central registration system by the research persons, the participants were assigned to undergo either of the following pathways: 1) the SITT first (with intravenous insulin injection), followed by an insulin tolerance test with subcutaneous insulin injection (ITTsc) or 2) the ITTsc first, followed by the SITT.This assigning of the patients was performed on a 1:1 basis by the research office, and the results were communicated to the co-investigators in writing.The two insulin tolerance tests (SITT and ITTsc) were completed within 3 days, irrespective of the order(Fig 1).

Pre-test preparations
The controls arrived at the hospital in the morning after fasting overnight to participate in the study.Patients with T2DM, after admission to the hospital, were started on a diet (25-30 kcal/kg standard .body weight/day); metformin, pioglitazone, and sodium glucose transporter 2 inhibitors were withdrawn on admission.The patients' fasting venous blood was collected the day after admission; if the FPG level was >140 mg/dL, the patient was included in this study.The controls and patients both underwent the SITT and ITTsc after overnight fasting.

SITT procedures
The SITT was performed before breakfast.Venous blood samples were collected for measurement of plasma glucose before and at 3, 6, 9, 12, and 15 min after an intravenous bolus injection of regular human insulin (0.1 U/kg bodyweight; Eli Lilly®, Eli Lilly and Company, IN, USA).Fifteen minutes after the insulin injection, the test was completed by injecting 20 mL of 50% glucose solution.
K ITT [now labelled K ITT (iv)] was calculated from the linear slope of the curve of plasma glucose concentrations at the 3-15-min timepoints using the Lundbaek equation [12].

ITTsc procedures
The ITTsc was performed before breakfast.Finger-prick blood samples were collected for the measurement of blood glucose before, and at 3, 6, 9, 12, 15, 30, 45, 60, 75, 90, and 120 min after a subcutaneous injection of regular human insulin (0.1 U/kg bodyweight).The K ITT (now labelled K ITT (sc)) was calculated from the linear slope of the curve of finger-pick blood glucose concentrations at the 30-120 min timepoints using the Lundbaek equation.

Clinical parameters related to insulin resistance
The waist circumference and waist-to-hip ratio before breakfast were measured during the first ITT.The visceral fat area (VFA) was calculated using the bioimpedance method (BIM; InBody770 ® , InBody Japan Inc., Tokyo, Japan).We measured the fasting serum C-peptide immunoreactivity (CPR); levels of fasting immunoreactive insulin (F-IRI), adiponectin, leptin, and TNF-α; and concentrations of plasma amino acids (AA) and branched chain AAs (BCAA) using the baseline (0 min) blood sample from the SITT or ITTsc (whichever was performed first).The HOMA-IR values were calculated using the following formula: HOMA-IR = (FPG [mg/dL] × F-IRI [μU/mL])/405.F-IRI and the HOMA-IR values were not determined in insulin-treated patients.

Insulin treatment after SITT/ITTsc
After the SITT and ITTsc, all 17 patients with T2DM were treated with multiple daily insulin .injections.Basal and bolus insulin doses were adjusted until the FPG levels were ≤130 mg/dL and the postprandial plasma glucose levels were ≤180 mg/dL.The total insulin dose used for glycemic control was then calculated.Using K ITT (iv) and the HOMA-IR value, equations by which the total insulin dose (/body weight) could be calculated were formed.

Statistical analyses
Data are expressed as mean ± standard deviation.The required sample size was determined to be 19 after assuming a correlation coefficient of 0.6 between K ITT (iv) and K ITT (sc), a power of 80%, and a significance level of 5%.However, the actual sample size was set at 24, assuming a 20% dropout rate.
We compared K ITT (iv) between the controls and patients with T2DM.We then examined the correlation of K ITT (iv) with the HOMA-IR value and K ITT (sc) using Pearson's correlation coefficient analysis.The intraclass correlation coefficient (ICC) was also used to assess the relationship between K ITT (iv) and K ITT (sc).We further analyzed the association of K ITT (iv) with VFA and other parameters related to obesity and insulin resistance using a Pearson's correlation coefficient analysis.In addition, the association of K ITT (iv) and the HOMA-IR value with the daily insulin dose required for glycemic control was evaluated in patients with diabetes, using Pearson's correlation coefficient analysis.The frequency of adverse events was analyzed using the McNemar test.All analyses were performed using JMP (SAS institute, Cary, NC, USA).
Written informed consent was obtained from all individuals prior to participation in the study.
This study was approved by the ethics committee of the Hokkaido University Hospital (date of approval 25 October 2016; approval no.016-0014) and registered in the University Hospital Medical Information Network (UMIN000024453).

Results
The clinical characteristics of the participants are summarized in Table HOMA-IR, homeostatic model assessment for insulin resistance; K ITT (iv), K index of the intravenous insulin tolerance test; K ITT (sc) calculated from the linear slope of the curve of plasma glucose concentrations at the 30-120-min timepoints using the

Lundbaek equation
The mean K ITT (iv) was significantly higher in the controls than in the patients (4.5%±1.8% vs.
In all participants, there was a significant correlation between K ITT (iv) and the HOMA-IR value   (a)K ITT (iv) was significantly correlated with the mean total insulin dose per body weight (r=−0.550,p<0.05); this was estimated using with the following regression equation: total insulin dose/kg = 0.78-0.10× K ITT (iv).(b)The HOMA-IR value was also significantly correlated with the mean total insulin dose per body weight (r=0.806,p<0.05); this was estimated using with the following regression equation: total insulin dose/kg = 0.23+0.15× HOMA-IR.
K ITT (iv), K index of the intravenous insulin tolerance test; HOMA-IR, homeostasis model assessment for insulin resistance Seven and two adverse events occurred in participants undergoing the SITT and ITTsc, respectively.All adverse events were hypoglycemia.No significant difference was observed in the frequency of adverse events between participants undergoing the SITT and the ITTsc.
In the present study, we prospectively studied seven healthy controls and 14 patients with T2DM and evaluated their insulin resistance using K ITT (iv).The following four main results were obtained: 1) the K ITT (iv) of patients with T2DM was 2.5%±2.1%,which was significantly lower than that of the controls (4.5%±1.8%);2) K ITT (iv) was significantly correlated with the HOMA-IR value but not with K ITT (sc); 3) K ITT (iv) was significantly correlated with VFA and the serum glucose, CPR, TNF-α, BCAA/total AA, and adiponectin concentrations; and 4) the total insulin dose required for glycemic control was correlated with K ITT (iv) and the HOMA-IR value, and it was estimated using an equation in patients with T2DM.
HOMA-IR values are widely used as indices of insulin resistance [13,14]; however, their reliability decreases when the insulin secretion is reduced and hyperglycemia occurs [4].In the present study, there was a significant correlation between the HOMA-IR value and K ITT (iv), suggesting its applicability as an index of insulin resistance even in a cohort including healthy volunteers and patients with poorly controlled T2DM.The significant correlation between the HOMA-IR value and K ITT (iv) may be attributed to the fact that the FPG level was not very high and the decrease in insulin secretion was only mild in our patients with T2DM.A previous study revealed that the HOMA-IR value is a useful index for determining insulin resistance at an FPG range of 80-170 mg/dL in obese Japanese patients with T2DM [4].In fact, in this study, the mean FPG level of patients with T2DM was approximately 170 mg/L (Table 1), and the correlation observed between CPR and K ITT supports the fact that the HOMA-IR value can be used as an effective index in this group (Fig 2b).In addition, the association between the two indices was observed even in the controls, which might have led to the significant correlation between the two indices in all study participants.Although the HOMA-IR data in this study were exclusive of patients on insulin therapies for diabetes, Okita et al. [10] reported that K ITT in such patients can be a useful indicator of insulin resistance using the euglycemic clamp test.
We expected that K ITT (sc) would be correlated with K ITT (iv) because the pattern of glucose level changes, unlike the rate of action, in an individual may be similar between intravenous and subcutaneous routes of insulin administration.In addition, the accuracy of finger-prick tests for blood glucose measurement has substantially improved [15].However, contrary to our expectations, there was no significant correlation between K ITT (iv) and K ITT (sc) (ICC=0).A possible explanation for this dissociation may be the instability of the insulin absorption rate due to differences in the skinfold thickness and subcutaneous blood flow [16,17].Exercise, smoking, and body position are also reported to affect insulin absorption [16,17].However, in this study, all patients were resting, had stopped smoking, and were sitting; thus, we considered these factors to have no effect.Previous studies [18][19][20][21] have compared glycemic control between intravenous and subcutaneous routes of insulin administration mainly in intensive care units; however, to our knowledge, there are no crossover studies comparing the two routes in the same individuals.
In this study, K ITT (iv) was correlated with various clinical parameters associated with obesity .and insulin resistance, including VFA, adiponectin, tumor necrosis factor alpha (TNF-α), and branched chain amino acids (BCAA)/total AA.Obesity and visceral fat are closely associated with insulin resistance; thus, the significant correlation between KITT(iv) and the VFA was considered reasonable [10,22].Low adiponectin levels were also associated with an increased risk of insulin resistance [23], and TNF-α is a strong inhibitor of adiponectin promoter activity [24].In non-obese, non-diabetic individuals, no significant correlation was observed between adipose tissue TNF-α mRNA content and K ITT (iv) [25].
However, when including patients with T2DM and a tendency for obesity, positive and negative correlations of K ITT (iv) with adiponectin and TNF-α, respectively, were observed in prior investigations similar to in our study [10].In addition, in our study, K ITT (iv) was negatively associated with BCAAs/total AAs.This corroborates the findings of prior studies that revealed a significant association between insulin resistance and BCAAs [26,27].In fact, BCAAs were reported to accumulate under conditions of insulin resistance [28]; furthermore, a positive association of insulin resistance with valine, one of the three BCAAs, has been reported in Japanese healthy individuals [29].
To the best of our knowledge, this is the first report on a regression equation for calculating the insulin dose required for glycemic control based on K ITT (iv).It is useful for adjusting the insulin dose and avoiding hypoglycemic events.Notably, the correlation coefficient was better for the HOMA-IR value (r=0.806)than for K ITT (iv) (r=−0.550),suggesting that the HOMA-IR value is an easily applicable parameter that helps estimate the insulin dose required for glycemic control.
This study had several limitations.First, the sample size was small and the study was conducted at a single center.Second, there was significant heterogeneity among participants in terms of the age, sex, comorbidities, and treatment regimens (in patients with T2DM).The first and second limitations precluded robust statistical analyses.Finally, we did not verify the reproducibility and reliability of the equations for calculating the total insulin dose using the HOMA-IR value and K ITT (iv).
In conclusion, K ITT (iv) was associated with the HOMA-IR value and insulin resistance-related parameters (such as VFA, adiponectin, and BCAAs).K ITT (iv) is an index of insulin resistance that can be applied in clinical practice; along with HOMA-IR, it may be useful for efficiently adjusting insulin doses.  .

Fig 4 .
Fig 4. Relationship of K ITT (iv) and the HOMA-IR value with the mean total insulin dose/kg.