Pre-Prediabetes: Insulin Resistance Is Associated With Cardiometabolic Risk in Nonobese Patients (STOP DIABETES)

Abstract

Context

Prior studies have demonstrated glycemic and cardiometabolic risk in the prediabetic state.

Objective

This work aims to examine if insulin resistance (IR) is associated with markers of glycemic, cardiometabolic, and atherosclerotic risk in nonobese, nonprediabetic individuals compared to insulin-sensitive (IS) individuals matched for body mass index (BMI), sex, and age.

Methods

Of 1860 patients from the STOP DIABETES study, 624 had normal fasting plasma glucose, BMI less than 30, and glycated hemoglobin A_1c (HbA_1c) less than 5.7%. All received an oral glucose tolerance test. Insulin sensitivity was quantitated using the Matsuda index: less than the 25th percentile equals IR (n = 151) and 25th percentile or greater equals IS (n = 473). Measures of dysglycemia and cardiometabolic risk were compared between IR individuals (n = 151) and a subset of IS individuals who were matched for BMI, sex, and age (n = 151). Carotid intima media thickness and carotid plaque were measured in 65 IR and 76 IS individuals.

Results

Compared to matched IS patients, IR nonobese individuals demonstrated increased indicators of glycemic and cardiometabolic risk, including increased 60-minute plasma glucose and percentage of patients with 60-minute plasma glucose greater than 155 mg/dL; increased 120-minute plasma glucose; unrecognized impaired glucose tolerance and type 2 diabetes, decreased disposition index; increased systolic and diastolic blood pressure; elevated plasma triglycerides (TGs); reduced high-density lipoprotein (HDL) cholesterol; increased TGs/HDL ratio, and high-sensitivity C-reactive protein. The presence, size, and number of carotid plaques was greater in the IR group.

Conclusion

Approximately 1 in 4 nonobese patients in this population with normal fasting glucose and HbA_1c were IR. In these nonobese participants, IR was associated with multiple indicators of dysglycemia and cardiometabolic risk.

We have shown that insulin resistance (IR), as well as β-cell dysfunction, is evident long before the onset of conventionally defined prediabetes (1-4) and have referred to this as the pre-prediabetic state. IR is a fundamental pathophysiologic defect in the development of type 2 diabetes mellitus (T2DM) (1-7) and results in compensatory hyperinsulinemia to maintain normal glucose tolerance (NGT) (1-4, 8, 9). However, the compensatory increase in insulin secretion places stress on the pancreatic β cells, leading to a progressive decline in β-cell function (10, 11). As the β-cell response decreases, in the presence of IR (1), progressive hyperglycemia ensues. According to American Diabetes Association (ADA) criteria (12), when fasting plasma glucose (FPG) reaches 100 mg/dL or glycated hemoglobin A_1c (HbA_1c) reaches 5.7%, a prediabetic state is recognized. Many of these prediabetic individuals progress to T2DM (1, 8, 9). Prior to the diagnosis of overt T2DM, both microvascular and macrovascular damage can be demonstrated (13, 14). Importantly, treatment of IR in prediabetic, as well as pre-prediabetic individuals with established vascular disease, has been shown to significantly reduce major adverse cardiovascular (CV) events (15).

The great majority of obese individuals are IR, and the severity of IR increases with increasing body mass index (BMI) (16). Obesity and IR are both associated with the development of prediabetes (FPG = 100-125 mg/dL, HbA_1c = 5.7%-6.4%, or 2-hour PG = 140-199 mg/dL). IR can be present in the absence of obesity (1, 6, 17, 18) and precedes the development of prediabetes. It has also been demonstrated that a 1-hour glucose measurement greater than 155 mg/dL during an oral glucose tolerance test (OGTT) identifies a population at increased risk of progressing to T2DM (19,20). However, despite evidence supporting the use of the 1-hour glucose measurement greater than 155 mg/dL as a predictor of future diabetes (21, 22), the vast majority of clinical practices rely on the FPG concentration and HbA_1c to identify prediabetic individuals. The aim of the present study was to examine how frequently IR and associated measures of cardiometabolic risk are present in IR nonobese, nonprediabetic individuals. The study population represents a retrospective evaluation of patients at risk for diabetes in a community practice in southern California. This study is registered with ClinicalTrials.gov (number NCT03308773) and was approved by the Providence Health and Services Oregon Institutional Review Board.

Materials and Methods

The patients in this study were derived from an internal medicine and endocrinology practice in southern California and represent a subset of the STOP DIABETES population (23). Patients with risk factors for T2DM were offered physiologic testing with an OGTT, as is standard in this practice. The risk factors for T2DM included: family history of T2DM in a first-degree relative, impaired fasting plasma glucose (IFG), HbA_1c greater than 5.6% and less than 6.5%, history of gestational diabetes, delivery of a baby weighing more than 9 pounds (4.1 kg), low high-density lipoprotein (HDL) (<40 mg/dL in men and <50 mg/dL in women), acanthosis nigricans, overweight (BMI = 25.0-30.0), hypertension (≥ 130/90 mm Hg), fasting triglycerides (TGs) greater than 250 mg/dL, ultrasound evidence of fatty liver, polycystic ovary syndrome, atherosclerotic CV disease, or member of a high-risk population (African American, Hispanic/Latino, Native American, Alaskan Native, Asian American, or Pacific Islander). From an initial population of 1860 patients, we identified 622 individuals who were nonobese (BMI < 30) and not prediabetic as defined by the ADA criteria (12) (FPG < 100 mg/dL and HbA_1c < 5.7%). Individuals with a BMI of 30.0 or greater, FPG greater than 99 mg/dL, or HbA_1c greater than 5.6% were excluded. The ethnicity of the population was primarily White and not significantly different between the IR and IS groups (White 86%, Asian 6%, Hispanic 5%, South Asian 2%, African American 1%).

Laboratory testing included a chemistry panel, high-sensitivity C-reactive protein (hs-CRP), HbA_1c, and lipid profile. Following a 10-hour overnight fast, participants received a 75-g OGTT with measurement of plasma glucose, insulin, and C-peptide concentrations at 0, 30, 60, and 120 minutes. From these results, the glycemic response, IS, and β-cell function secretion were quantified as previously described (3, 4, 23, 24).

IS was determined with the Matsuda Index (24) using the 0-, 30-, 60-, and 120-minute PG and insulin concentrations during the OGTT. The Matsuda Index provides a measure of total body IS and is calculated from glucose and insulin levels measured in response to the 75-g OGTT. This index is highly correlated with the gold-standard euglycemic insulin clamp (r = .73; P < .001) (24).

This value was compared to the Matsuda indices derived from a population of patients (n = 675) with NGT from the same practice. IR was defined as Matsuda index less than the 25th percentile when compared to this NGT population. Using this definition, 151 patients from this screened population (<25th percentile) were identified as being IR. From the remaining 473 IS patients, a subpopulation (n = 151) matched for sex, age, and BMI was generated using propensity matching via XLSTAT (Fig. 1). Comparisons between these groups were performed using 2-tailed t test (StatPlus software).

The early insulinogenic index was calculated as the ratio of the increment in plasma insulin to the increment in PG (ΔI/ΔG) from 0 to 30 minutes. Total insulin secretion was calculated as the ratio between the incremental area under the plasma C-peptide (C-pep) curve (0-120 minutes) to the incremental area under the PG concentration (G) curve (0-120 minutes) during the OGTT. The disposition index was calculated as (incremental ΔC-pep/incremental ΔG)_0-120 multiplied by the Matsuda index of insulin sensitivity (3).

All laboratory analyses were performed at Providence Little Company of Mary Medical Center Torrance, California, and included: hs-CRP (latex-particle enhanced immunoturbidimetric assay; Beckman CX-9), HbA_1c (high-performance liquid chromatography; BioRad-Variant Turbo II); plasma insulin and C-pep concentrations (Advia Centaur XP, 2-site sandwich immunoassay using direct chemiluminescence, Seimens), and PG (hexokinase reaction, Beckman Syncron CX-9). Blood pressure (BP) (reclining), height, weight, calculated BMI, medical history, and physical examination were obtained during routine office visits. BP was measured consistent with American Heart Association guidelines. Patients were seated in a chair with their feet flat on the floor and the BP cuff on the upper arm at the level of the heart. A Welsh Allen Connex Spot Meter device with an adult upper arm cuff, appropriate for arm circumference and calibrated twice yearly, was used to measure BP.

A subset (65 IR and 76 IS) of patients received measurements of carotid intima media thickness (CIMT) and carotid plaque size and number using a Siemens S5000 ultrasound and CV syngo Arterial Health Package software with automated edge detection. The protocol included 3 measurements of the far wall of the common carotid artery, 1 measurement of the carotid bulb, and 1 measurement of the internal carotid artery. A vascular ultrasound technician trained in CIMT image acquisition performed the studies and was blinded to the metabolic state of the patient. The criteria for defining carotid plaque are consistent with the recommendations of the American Society of Echocardiography (25).

Maximal common CIMT was used for this analysis. Each study included a sweep of the carotid arterial system to quantify carotid plaque. Plaque was defined as maximal thickness of 1.5 mm or greater and 50% greater than adjacent IMT.

All statistical comparisons between the IR and IS groups were performed with the Mann-Whitney U test to account for nonnormally distributed variables. Since multiple comparisons were measured between groups (Tables 1-3), a Bonferroni adjustment was performed and gave an adjusted α equal to .0263. P values above this threshold should be interpreted accordingly. Spearman rho was employed to examine nonparametric correlations (Jamovi software, version 2.5.4). A linear multivariate analysis including Matsuda index, sex, BMI, age, HbA_1c, systolic BP (SBP), diastolic BP (DBP), low-density lipoprotein cholesterol, hs-CRP, TGs, and HDL cholesterol, was employed to examine the potential for confounders contributing to carotid atheroma.

Clinical outcomes in cardiometabolic research commonly compare the lowest quartile (most IR) group to those above the 25th percentile, and this approach was used in this manuscript. The identification of clinically significant cut points could be of importance and useful to practicing physicians.

Results

A total of 624 of 1860 screened individuals were found to be nonobese (BMI < 30) and nonprediabetic (FPG < 100 mg/dL and HbA_1c < 5.7%). Of these 622 individuals, 151 were IR (Matsuda Index < 25th percentile) and compared to a subset (n = 151) of IS individuals identified by propensity matching for age, sex, and BMI (see Fig. 1).

Glycemic Measures

Compared to the IS group, the nonobese, nonprediabetic IR group demonstrated multiple measures of glucose intolerance: 60-minute PG (147 vs 123 mg/dL; P < .0001), percentage of patients with 60-minute glucose greater than 155 mg/dL (44% vs 3%; P < .0001), 120-minute PG (112 vs 90 mg/dL; P < .0001), mean PG area under the curve (AUC) from 0 to 120 minutes (134 vs 116 mg/dL/min; P < .0001), unrecognized impaired GT (22 [14.6%] vs 7 [4.6%]; P = .005), and unrecognized T2DM (4 [2.6%] vs 0 [0%]; P = .04).

In the IR group, the early insulinogenic index ([ΔI/ΔG]_0-30) (3.68 vs 1.32; P < .0001) and total insulin secretion (0-120 minutes) (228.2 vs 73.9; P < .0001) (see Table 1) were significantly increased. Based on stratification, the Matsuda Index was significantly reduced in the IR group (2.278 vs 7.532; P < .0001). The disposition index, which expresses total insulin secretion per severity of IR, was significantly reduced in the IR nonobese, nonprediabetic group (0.52 ± 0.07 vs 1.83 ± 0.21; P < .0001).

Cardiometabolic Measures

The IR group had significantly higher mean SBP (124 vs 119 mm Hg, P = .006) and DBP (75 vs 72 mm Hg; P < .0001), higher plasma TG concentration (144 vs 100 mg/dL, P < .0001), increased TGs/HDL ratio (3.16 vs 1.84; P < .0001), lower HDL (53 vs 60; P < .0001), and higher hs-CRP (2.1 vs 1.3 mg/L; P = .005) than the IR group (see Table 1).

When glycemic and cardiometabolic measures were compared in the matched IS and IR individuals using BMI cut points of less than 27.5 and 25.0, the results were very similar to those using a BMI cut point of less than 30.0 (see Tables 2 and 3).

Carotid Intima Media Thickness and Carotid Plaque

There were no statistically significant differences in age, sex, BMI, or ethnicity in the subgroups of IR and IS patients who underwent CIMT and carotid plaque assessment. Further, there were no significant differences in SBP, total cholesterol, or non-HDL cholesterol between groups. Other metabolic laboratory tests commonly associated with IR were significantly different between the IR and IS groups, including plasma TGs (166 vs 106 mg/dL; P = .003), HDL cholesterol (50.6 vs 57.8 mg/dL; P = .004), and serum hs-CRP (2.4 vs 1.2 mg/L; P = .04).

In the subset of IR (n = 65) and IS (n = 76) individuals who received a measurement of CIMT, no statistically significant difference in common CIMT was observed (0.819 vs 0.793 mm; P = .93). The percentage of patients with carotid plaque was greater in the IR group (46 vs 28%; P = .02). The number of carotid plaques per patient (0.75 vs 0.42; P = .03) and plaque burden (sum of plaque sizes, 1.83 vs 0.90 mm; P = .01) also was greater in the IR group (Table 4). The percentage of patients with carotid plaque and the number of carotid plaques per patient, plaque size, and plaque burden were increased in the IR vs IS groups in a subanalysis of patients (n = 102) with BMI less than 27.5 (Table 5). A subanalysis of patients with a BMI less than 25.0 (n = 47) demonstrated increases in the number of carotid plaques and plaque burden in IR patients that did not reach statistical significance (Table 6). Evaluating measures of atheroma, a linear multivariate analysis was performed that included the Matsuda Index, sex, BMI, age, HbA_1c, SBP, DBP, low-density lipoprotein cholesterol, hsCRP, TGs, and HDL cholesterol. Regarding the presence of plaque (yes/no), only the Matsuda Index reached statistical significance (P = .006). With regard to plaque number and plaque burden, age (P = .021) and Matsuda index (P = .008) were both found to be statistically significant.

Correlations Between Insulin Resistance vs Glucose Tolerance, Insulin Secretion, and Cardiometabolic Risk Factors

IR was positively correlated with the 60-minute (r = 0.44; P < .0001) and 120-minute (r = 0.52; P < .0001) PG concentrations during the OGTT and total insulin secretion (ΔAUC I_0-120/ΔAUC G_0-120) (r = 0.12; P < .04) (see Table 3). IR was positively correlated with increased hs-CRP, plasma TGs, TGs/HDL cholesterol ratio, SBP and DBP, and BMI, and negatively with HDL cholesterol. IR also was positively correlated with carotid plaque size, plaque burden, and number (Table 7).

Discussion

The major novel finding of the present study is that IR and its associated abnormalities in glycemia, insulin secretion, cardiometabolic risk factors, and increased carotid plaque are present in the pre-prediabetes state based on the ADA definition of prediabetes (12) and occur in the absence of obesity whether one uses a BMI cut point of less than 25, less than 27.5, or less than 30.

IR is a major physiologic defect that precedes the development of T2DM by years to decades (1-7). The prediabetic state is characterized by IR and compensatory hyperinsulinemia, followed by a progressive decline in insulin secretion and β-cell function leading to an overtly diabetic state (1). The incidence of macrovascular complications also is increased in prediabetic individuals (15, 17, 26). The present study results demonstrate that the components of metabolic syndrome and accelerated atherogenesis are well established in IR prediabetic individuals (normal FPG and HbA_1c) in the absence of obesity. Because advanced carotid atherogenesis already is present in the prediabetic state, it is clear that IR, multiple CV risk factors (dyslipidemia, increased DBP/SBP, and a proinflammatory state characterized by increased hs-CRP) and accelerated atherogenesis must start in the pre-prediabetic stage, and we suggest that this terminology be incorporated into the ADA guidelines.

Compared to the IS group, IR NGT individuals had an increased PG excursion during the OGTT and an increased plasma insulin response. However, the disposition index, which provides a measure of β-cell function, was significantly reduced in the pre-prediabetic state. A decline in β-cell function within the NGT range previously has been emphasized by DeFronzo et al (1, 3, 4, 8, 9). One-hour PG greater than 155 mg/dL has been shown to identify an increased risk of progression to T2DM (19, 20), and our results demonstrate that the IR group compared to the IS group was much more likely to have crossed this threshold (44% vs 3%). Of note, the mean 1-hour PG concentration approached this value (155 mg/dL) in the IR group (147 vs 123 mg/dL; P < .0001). Consistent with the presence of IR and reduced β-cell function, we observed an increased incidence of IGT (22% vs 7%; P = .003) and T2DM (3% vs 0%) in the IR group during the 75-g OGTT despite normal fasting PG, normal HbA_1c, and the absence of obesity. Whether one uses a BMI cut point of 30, 27.5, or 25 as an index of obesity, subanalyses in patients with BMI less than 27.5 (IR, n = 102 and IS, n = 101) and BMI less than 25.0 (IR, n = 48 and IS, n = 51) yielded similar and highly significant differences in glycemic variables, β-cell function, and cardiometabolic risk, with the exception of newly diagnosed T2DM (none in either group).

Subanalysis of carotid plaque presence (n = 102), number of plaques, and size of plaques using a BMI cut point of less than 27.5 demonstrated a similar increase in the IR group (P = .02; P = .01; P = .01; respectively). Subanalysis using a BMI cut point of less than 25 did not reach statistical significance for measures of carotid atherogenesis, most likely because of the small number. CIMT and carotid plaque burden are additive with regard to macrovascular risk, and the number of carotid plaques has been associated with increased coronary artery disease risk (27). In the present study, a greater percentage of IR patients demonstrated carotid plaque, as well as an increase in the number of plaques. These findings indicate that the development of atherosclerosis is accelerated in IR individuals and is well established in the pre-prediabetic state in NGT, nonobese individuals. IR has been identified as an important contributor to residual CV risk (5, 6, 28-34), and its targeted treatment presents an opportunity to reduce CV risk (15, 28) beyond treatment of conventional risk factors. Hyperinsulinemia, which represents the compensatory response to IR, has also been shown to be associated with increased CV risk (31).

Our study has some limitations. It is from a single center, is cross-sectional in design, and the ethnicity of the population was primarily White. A strength of the study is the large number of nonobese IR, normoglycemic participants with measurements of multiple CV risk factors establishing a pre-prediabetes cohort.

Conclusion

In patients with risk factors for diabetes, approximately 1 in 4 nonobese, nonprediabetic patients are IR, and this is similar to previously published results with the euglycemic insulin clamp (18). Although IR, independent of classic CV risk factors, has been shown to be associated with increased macrovascular events (5, 6, 28, 29) and dysglycemia (1), to the best of our knowledge, this is the first study from a real-world practice that has demonstrated that IR—in the absence of obesity and prediabetes—is associated with β-cell dysfunction, unrecognized impaired glucose tolerance, and T2DM, multiple measures of cardiometabolic risk, and the presence and atherosclerotic burden of carotid plaques.

Disclosures

J.A. and R.R. are cofounders of RobustForLife. R.D. is on the advisory boards of AstraZeneca, Novo Nordisk, Intarcia, Boehringer Ingelheim, and Corcept; receives research support from AstraZeneca and Merck; and is on the speaker bureaus of Novo Nordisk, Corcept, and AstraZeneca. M.A.G. declares no competing interests.

Data Availability

Some or all data sets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Clinical Trial Information

ClinicalTrials.gov registration number NCT03308773 (registered September 28, 2017).

References

Abbreviations

 
• ADA

  American Diabetes Association

 
• AUC

  area under the curve

 
• BMI

  body mass index

 
• CIMT

  carotid intima media thickness

 
• C-pep

  C peptide

 
• CV

  cardiovascular

 
• DBP

  diastolic blood pressure

 
• FPG

  fasting plasma glucose

 
• HbA_1c

  glycated hemoglobin A_1c

 
• HDL

  high-density lipoprotein

 
• hs-CRP

  high-sensitivity C-reactive protein

 
• IGT

  impaired glucose tolerance

 
• IR

  insulin resistant

 
• IS

  insulin sensitive

 
• NGT

  normal glucose tolerance

 
• OGTT

  oral glucose tolerance test

 
• SBP

  systolic blood pressure

 
• T2DM

  type 2 diabetes mellitus

 
• TGs

  triglycerides