You are here

  • Home
  • Archive
  • Volume 2, Issue 1
  • Serum α-hydroxybutyrate (α-HB) predicts elevated 1 h glucose levels and early-phase β-cell dysfunction during OGTT

Article Text

Article menu
XML
Epidemiology/Health service research
Serum α-hydroxybutyrate (α-HB) predicts elevated 1 h glucose levels and early-phase β-cell dysfunction during OGTT
  1. Stephen A Varvel1,
  2. James V Pottala1,2,
  3. Dawn L Thiselton1,
  4. Rebecca Caffrey1,
  5. Tara Dall1,
  6. Maciek Sasinowski1,
  7. Joseph P McConnell1,
  8. G Russell Warnick1,
  9. Szilard Voros1,
  10. Timothy E Graham3
  1. 1Health Diagnostic Laboratory, Inc., Richmond, Virginia, USA
  2. 2Department of Internal Medicine, Sanford School of Medicine, University of South Dakota, Sioux Falls, South Dakota, USA
  3. 3Department of Medicine, Biochemistry and Nutrition, University of Utah, Salt Lake City, Utah, USA
  1. Correspondence to Dr Stephen A Varvel; svarvel{at}hdlabinc.com

Abstract

Objective Serum α-hydroxybutyrate (α-HB) is elevated in insulin resistance and diabetes. We tested the hypothesis that the α-HB level predicts abnormal 1 h glucose levels and β-cell dysfunction inferred from plasma insulin kinetics during a 75 g oral glucose tolerance test (OGTT).

Research design and methods This cross-sectional study included 217 patients at increased risk for diabetes. 75 g OGTTs were performed with multiple postload glucose and insulin measurements over a 30–120 min period. OGTT responses were analyzed by repeated measures analysis of variance (ANOVA). Multivariable logistic regression was used to predict 1 h glucose ≥155 mg/dL with α-HB added to traditional risk factors.

Results Mean±SD age was 51±15 years (44% male, 25% with impaired glucose tolerance). Fasting glucose and insulin levels, but not age or body mass index (BMI), were significantly higher in the second/third α-HB tertiles (>3.9 µg/mL) than in the first tertile. Patients in the second/third α-HB tertiles exhibited a higher glucose area under the receiver operating characteristics curve (AUC) and reduced initial slope of insulin response during OGTT. The AUC for predicting 1 h glucose ≥155 mg/dL was 0.82 for a base model that included age, gender, BMI, fasting glucose, glycated hemoglobin (HbA1c), and insulin, and increased to 0.86 with α-HB added (p=0.015), with a net reclassification index of 52% (p<0.0001).

Conclusions Fasting serum α-HB levels predicted elevated 1 h glucose during OGTT, potentially due to impaired insulin secretion kinetics. This association persisted even in patients with an otherwise normal insulin–glucose homeostasis. Measuring serum α-HB could thus provide a rapid, inexpensive screening tool for detecting early subclinical hyperglycemia, β-cell dysfunction, and increased risk for diabetes.

  • Oral Glucose Tolerance Test
  • Biomarkers
  • Insulin Resistance
  • Pancreatice Insulin Secretion

This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/

https://doi.org/10.1136/bmjdrc-2014-000038

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Key messages

    • Fasting serum α-hydroxybutyrate (α-HB) levels predicted elevated 1 h glucose during oral glucose tolerance test, potentially due to impaired insulin secretion kinetics.

    • This association persisted even in patients with otherwise normal insulin–glucose secretion kinetics.

    • Measuring serum α-HB could thus provide a rapid, inexpensive screening tool for detecting early subclinical hyperglycemia, β-cell dysfunction, and increased risk for diabetes.

    Introduction

    The US Centers for Disease Control estimates that approximately 30% of individuals with diabetes are not aware that they have the condition.1 Prediabetes, which confers a 50% lifetime risk for diabetes, affects more than 38% of US adults (>90 million people), and yet less than 10% of those affected have ever received a formal diagnosis.2 ,3 Prediabetes and diabetes (type 2) share the fundamental pathophysiological features of insulin resistance and pancreatic β-cell dysfunction.4–6 Even without hyperglycemia, the presence of insulin resistance and β-cell dysfunction increases the risk for cardiovascular disease, retinopathy, neuropathy, microalbuminuria, certain cancers, pulmonary disease, depression, dementia, hospitalization, and early death.7–12 It is clear that detection and treatment of these earliest, fundamental stages of the disease process could provide a more effective strategy for prevention. Consequently, efforts are underway to develop convenient, sensitive, and predictive methods for earlier diagnosis of metabolic disease.13–15

    Several circulating metabolites identified via an unbiased metabolomics approach have recently been associated with insulin resistance or impaired glucose tolerance (IGT).16 ,17 One of these small molecule organic metabolites, α-hydroxybutyrate (α-HB), was found to have the highest correlation—of 485 screened metabolites—with the glucose disposal rate as assessed by the hyperinsulinemic-euglycemic clamp, and was elevated in patients without diabetes who had impaired fasting glucose or IGT based on a 2 h oral glucose tolerance test (OGTT).18 Importantly, α-HB and a second metabolite linoleoylglycerophosphocholine (L-GPC, a putative lipid signaling molecule also associated with insulin sensitivity) have recently been shown to predict progression of dysglycemia and development of type 2 diabetes (T2D).19 Furthermore, α-HB, but not L-GPC, was associated in these studies with impairments in surrogate indices of pancreatic β-cell function. Consistent with this, in vitro studies of cultured INS1e insulinoma cells showed that preincubation with α-HB tended to dose-dependently decrease insulin secretion in response to stimulation by glucose and arginine.19 Therefore, elevated α-HB could identify a subset of individuals at greater immediate risk for deterioration of glycemic control and progression to T2D.

    The biology of α-HB is not fully understood. This organic ketoacid is produced in the liver as a byproduct during the formation of α-ketobutyrate (a product of either threonine catabolism or methionine metabolism via cystathionine) under conditions of excess glutathione demand resulting from high oxidative stress. In addition, conditions that promote a high ratio of dihydronicotinamide adenine dinucleotide/nicotinamide adenine dinucleotide (NADH/NAD+), such as the increased oxidative stress state characteristic of insulin resistance, may increase production of α-HB. Hence, the idea of α-HB as a marker of oxidative stress holds some intuitive appeal as a potential pathophysiological link between increased α-HB and early impairment in pancreatic β-cell function, especially given the proposed role of oxidative injury in β-cell dysfunction (for a recent review, see Montane et al20).

    The primary objective of this study was to examine the relationship between the fasting α-HB level and insulin–glucose homeostasis in routine clinical practice patients, and to assess whether α-HB might detect underlying metabolic abnormalities in overtly normal patients. It has been shown that 1 h postload plasma glucose ≥8.6 mmol/L is a significant risk factor for progression to diabetes.21 We hypothesized that patients with higher fasting α-HB levels would be more likely to have 1 h postload plasma glucose ≥8.6 mmol/L and evidence of impaired β-cell function.

    Research design and methods

    Subjects

    Between March 2012 and May 2013, 217 patients underwent a 75 g OGTT and fasting blood collection to evaluate glucose tolerance at several outpatient centers across the USA (Madison, Wisconsin; Jackson, Mississippi; Montgomery, Alabama; Charleston, South Carolina; Seattle, Washington; and Salt Lake City, Utah). Clinical indications for testing included obesity, a history of first-degree family members with diabetes, and the presence of one or more components of the metabolic syndrome, including elevated blood pressure, high serum triglycerides, low high-density lipoprotein cholesterol (HDL-C), and either impaired fasting glucose or elevated glycated hemoglobin (HbA1c) in the non-diabetic range (ie, <48 mmol/mol (<6.5%)). The study protocol was approved by the Copernicus Group Institutional Review Board (IRB), Durham, North Carolina. All analyses involved de-identified data and were covered by a waiver of consent and authorization. Use of discarded clinical samples of patients in Salt Lake City, Utah, was additionally approved under a separate review and waiver process by the University of Utah Human Subjects Institutional Review Board.

    Laboratory/assays

    Blood samples were sent by overnight courier on dry ice to Health Diagnostic Laboratory, Inc (HDL, Inc, Richmond, Virginia, USA) for measurement of glucose, insulin, C-peptide, lipids, and α-HB. To ensure quality preservation and handling of the samples, blood was collected in BD Vacutainer SST ‘Tiger Top’ tubes with a clot activator and gel, allowed to clot for no more than 30 min, and centrifuged for 15 min at 3000 rpm before shipping. α-HB and L-GPC were determined by electrospray ionization liquid chromatography-mass spectrometry. Fasting glucose (mmol/L) was measured by an ultraviolet method, while HbA1c levels (mmol/mol) were determined by high performance liquid chromatography. Antiglutamic acid decarboxylase (anti-GAD) antibody levels were measured by an ELISA technique. Insulin and C-peptide were assessed by electrochemiluminescence immunoassay. Low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein choleserol (HDL-C) and triglycerides were measured using direct enzymatic assays.

    Oral glucose tolerance testing

    OGTT was performed according to the standardized protocol. Fasting blood samples were collected before administration of glucola (75 g glucose solution), which was consumed within 5 min. Additional blood samples were collected at either (1) 30, 60, 90, and 120 min, or (2) 60 and 120 min from completion of the glucola. All patients were restricted from eating, drinking, or smoking during the testing period. Two estimates of whole-body insulin sensitivity were derived from the OGTT data—the Matsuda Index22 and the ‘Clamp-like Index’ (CLIX), the latter as previously described ((serum creatinine (×0.85 if male)/(mAUCglucose× mAUCc-peptide))×6600), where mAUCglucose equals the total AUCglucose/120 and mAUCc-peptide equals the total AUCc-peptide/120.23 In a subset of patients where creatinine values were missing (29%), group mean creatinine values were utilized. In patients with no diabetes, CLIX correlates highly with the glucose disposal rate (ie, M value) measured during the euglycemic hyperinsulinemic clamp. An OGTT-derived version of the disposition index (AUCinsulin/glucose×Matsuda Index) was used to estimate β-cell function, reflecting insulin secretion relative to insulin resistance.24

    Statistical methods

    The study included 217 patients with suspected diabetes risk. Cross-sectional differences in patient characteristics were tested among fasting α-HB tertiles using one-way analysis of variance (ANOVA), and χ2 tests for continuous and categorical data, respectively. Repeated measures ANOVA models were used to analyze the glucose and insulin mean response profiles over the 3-time point or 5-time point 2 h OGTT. The insulin response-dependent variable was transformed using the natural transformation to improve the normality and homoscedasticity of the residual errors. Additionally, one-way ANOVA models were used to test the area under the curve (AUC) for the repeated glucose and insulin measurements among α-HB tertiles using the trapezoidal rule and either the 3-time point or 5-time point data as available. Residual diagnostics were conducted to verify model assumptions.

    Multivariable logistic regression was used to test the association (ie, odds ratio) and incremental improvement in discrimination (ie, c-statistic) for classifying patients with 1 h glucose ≥8.6 mmol/L when α-HB was added to the following list of potential confounding variables: age, gender, body mass index (BMI) categories (normal <25, ≥25 overweight <30, obese ≥30 kg/m2), fasting glucose, log(HbA1c), log(triglycerides), HDL-C, and LDL-C. HbA1c and triglycerides were natural logarithm transformed to reduce leverage of extreme observations. When testing the usefulness of a novel biomarker, the American Heart Association recommends reporting the marker's statistical association, discrimination, calibration, and reclassification performance.25 The Hosmer-Lemeshow test was used as a measure of model calibration.26 In addition, the reclassification of patients with 1 h glucose ≥8.6 mmol/L was tested when α-HB was added to the fully adjusted logistic regression model by calculating the integrated discrimination improvement (IDI), which can be described as the average increase in sensitivity given no change in specificity. Lastly, the continuous net reclassification index (NRI), which measures the percentage of patients who had model probabilities changed in the correct direction (ie, increased for those with events and decreased for non-events) due to the addition of α-HB to the fully adjusted model, was also tested.27 ,28 SAS V.9.3 (Cary, North Carolina, USA) was used for all analyses, and the critical level α=0.05 was used to prescribe statistical significance.

    Results

    Table 1 displays patient characteristics according to α-HB tertiles (<37.5 µmol/L (3.9 µg/mL), 37–53.8 µmol/L (3.9–5.6 µg/mL), >53.8 µmol/L (>5.6 µg/mL)). Fasting α-HB levels were not associated with age, gender, BMI, HbA1c, L-GPC, or lipid levels (table 1). However, mean fasting glucose, insulin, and C-peptide values were higher in the second and third α-HB tertiles compared with the first tertile. The OGTT results also revealed higher mean levels for 1 h and 2 h glucose, and lower scores on the Matsuda Index, CLIX, and an OGTT-derived version of the disposition index (AUCinsulin/glucose×Matsuda Index) in the highest two tertiles compared with the lowest (p<0.05).

    View this table:
    Table 1

    Patient characteristics and clinical factors grouped by α-HB tertiles (μg/mL)

    High fasting α-HB levels are associated with impaired insulin secretion and glucose homeostasis

    The mean glucose and log(insulin) levels at 30 min time points are shown by α-HB tertiles in figure 1. Since there were no significant differences between the second and third α-HB tertiles at any time point (see online supplementary figure 1A and table S1, minimum p=0.22), they were combined for a simpler interpretation. Figure 1A shows that the mean glucose levels were similar at baseline, but then in the group of patients with α-HB ≥3.9 µg/mL were elevated by the 30 min time point and remained higher compared to that in those with α-HB <3.9 µg/mL. The association between circulating glucose and α-HB was also revealed by the partial AUCglucose values (ie, that part of the AUCglucose >2.8 mmol/L (50 mg/dL)), for which the mean (SD) values were 127 (75), 172 (67), and 179 (92) units for increasing α-HB tertiles. The AUCglucose for patients in the second and third α-HB tertiles was significantly higher than the AUCglucose of patients in the first tertile, minimum p=0.0008), and there was no difference between AUCglucose for the second and third α-HB tertiles (p=0.56); providing further support for combining these groups (figure 1B).

    Figure 1
    Figure 1

    (A) Oral glucose tolerance test (OGTT) glucose and (C) natural logarithm of insulin responses over time with SEM bars comparing the first tertile to the second and third tertiles combined in routine clinical practice patients (N=217); *p value <0.05. The F-test for the α-hydroxybutyrate (AHB) group by time interaction is given above the panel. (B) OGTT glucose response area under the curve (AUC) above 50 mg/dL. (D) 30 min rise in log(insulin).

    The mean log(insulin) response during the OGTT by α-HB groups is shown in figure 1C. All α-HB groups exhibited similar peak insulin secretion levels during the OGTT (60 min minimum p=0.29; online supplemental figure 1B); however, the response for those with α-HB <3.9 µg/mL was different than with α-HB ≥3.9 µg/mL. Specifically, insulin levels rose and declined more rapidly in the former (figure 1C). The 30 min geometric rise (95% CI) for insulin was 61.2 (51.6 to 70.8) pmol/L (10.2 (8.6 to 11.8) μU/mL) and 51 (41.4 to 60) pmol/L (8.5 (6.9 to 10) μU/mL) for those <3.9 µg/mL and ≥3.9 µg/mL, respectively (p=0.0003; figure 1D).

    α-HB predicted elevated OGTT 1 h postload glucose level

    We used multivariable logistic regression to test the ability of fasting α-HB to classify patients having elevated 1 h postload glucose level ≥8.6 mmol/L during the OGTT. Considering fasting α-HB alone, a patient was 1.38 times as likely for each 1 µg/mL (9.6 µmol/L) increase (OR, 95% CI 1.20 to 1.59, p<0.0001) to have 1 h glucose ≥8.6 mmol/L. This strong relationship remained unchanged after adjusting for age, gender, BMI, fasting glucose, log(insulin), log(triglycerides), HDL-C, and LDL-C, OR=1.38 (95% CI 1.17 to 1.63, p=0.0002). The model was well calibrated, that is, the predicted counts of patients having elevated 1 h glucose was similar to the actual counts across event probability deciles (Hosmer-Lemeshow p=0.52). The c-statistic (area under the receiver operating characteristic (ROC) curve) was 0.82 for the base model adjusted for the multiple potential confounding variables listed above, and increased to 0.86 when α-HB was added (change in c-statistic 95% CI 0.01 to 0.07, p=0.015, figure 2). The addition of α-HB also resulted in a significant IDI of 0.078 (p<0.0001). This was due to increases in the average sensitivity and average specificity of 0.047 (p=0.0091) and 0.032 (p=0.0001), respectively. In addition, the continuous NRI was 0.52 (p=0.0001), that is, 52% of the patients had their probability of having 1 h glucose ≥8.6 mmol/L changed in the correct direction. These findings indicate that α-HB is a useful, independent predictor of elevated OGTT 1 h glucose in at-risk patients from routine clinical practice.

    Figure 2
    Figure 2

    Receiver operating characteristic (ROC) curves for classifying patients having a 1 h glucose ≥155 mg/dL during oral glucose tolerance test. The area increased by 0.039 (95% CI 0.008 to 0.070, p=0.015) when α-hydroxybutyrate (AHB) was added to age, gender, body mass index, fasting glucose, log(fasting insulin), log(triglycerides), high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol in the logistic regression model.

    α-HB predicts subclinical hyperglycemia in patients with normoglycemia

    Two subgroups were formed from the overall patient cohort. The first (N=137) excluded patients with fasting glucose ≥5.6 mmol/L (100 mg/dL) and HbA1c ≥39 mmol/mol (5.7%), representing those patients who would be identified as normoglycemic based on typical fasting glycemic measures. The second subgroup (N=93) further excluded patients with fasting insulin >72 pmol/L (12 (µU/mL), 2 h glucose ≥7.8 mmol/L (140 mg/dL), and an anti-GAD antibody positive in order to create a cohort which would typically be considered normal even if more extensive testing was performed. Figure 3A, B show the ROC curves for classifying patients with 1 h postload glucose ≥8.6 mmol/L for these two subgroups of patients using fasting glucose, HbA1c, or α-HB. The area increased by 0.10 (p=0.059), 0.09 (p=0.087), or 0.17 (p=0.0044), respectively, for the subgroup of patients with normal levels of fasting glucose and HbA1c. Similarly, α-HB exhibited strong predictive capability within the second cohort of patients with normoglycemia (figure 3B). In these patients, with low-risk levels for all the above biomarkers, the area increased by 0.15 (p=0.016), 0.15 (p=0.038), or 0.24 (p=0.0007), using only fasting glucose, HbA1c, or α-HB to classify 1 h postload glucose ≥8.6 mmol/L.

    Figure 3
    Figure 3

    Comparison of receiver operating characteristic (ROC) curves for predicting oral glucose tolerance test (OGTT) 1 h postload glucose ≥155 mg/dL in (A) patients with fasting glucose <100 mg/dL and glycated hemoglobin (HbA1c) <5.7% (N=137, with 33 events); the area increased by 0.10 (p=0.059), 0.09 (p=0.087), or 0.17 (p=0.0044) when using fasting glucose, log(HbA1c), or α-hydroxybutyrate (AHB) alone, respectively; or in (B) patients additionally with insulin ≤12 (U/mL, 2 h glucose <140 mg/dL, and a negative antiglutamic acid decarboxylase (anti-GAD) antibody (N=93, with 15 events). The area increased by 0.15 (p=0.016), 0.15 (p=0.038), or 0.24 (p=0.0007), respectively.

    Discussion

    Prediabetes is one of the strongest clinical risk factors for diabetes and vascular disease. In prediabetes and the early stages of diabetes, fasting glucose and HbA1c are often highly correlated, reflecting basal hepatic glucose output in the fasting state. The 75 g OGTT provides important, additional, diagnostic information by challenging insulin–glucose homeostasis with conditions that simulate the usual postprandial state, in which there is induction of insulin secretion, insulin-stimulated glucose uptake by skeletal muscle and other tissues, and suppression of hepatic glucose production. As such, the OGTT can uncover defective glucose regulation that is not observed in the fasting state. Consistent with this, up to 40% of individuals with abnormal 2 h OGTT glucose values have normal fasting glucose levels.29 Hence, fasting blood glucose and HbA1c values are not adequate substitutes for the OGTT. Nevertheless, the OGTT is not routinely administered in most clinical practice settings, in part due to the time and technical effort it requires.

    Recent studies are revealing that there is more to be learned from an OGTT than 2 h glucose values, which primarily reflect the rate of glucose clearance as a function of peripheral insulin resistance. Pancreatic β-cell dysfunction, the more proximal cause of diabetes, is directly reflected in impaired early-phase insulin response, and thus steeper early rises in glucose values during the first hour of the OGTT. Consistent with this, it has been demonstrated in two large prospective studies that the 1 h postload plasma glucose value obtained during the 75 g OGTT is a strong, independent predictor of risk for diabetes, even in individuals with normal (ie, not prediabetic or diabetic) fasting or 2 h glucose levels (normal glucose tolerance; NGT).30 In the San Antonio Heart Study and the Botnia Study, baseline 1 h glucose levels predicted diabetes incidence better than did either fasting or 2 h glucose, and were strongly associated with indices of β-cell function and insulin resistance.21 ,31 ,32

    Cross-sectional data from several recent studies have reinforced the clinical importance of elevated 1 h glucose levels in individuals with otherwise normal glucose regulation. For example, among the 39% of patients with NGT in the GENFIEV study, an elevated 1 h glucose level predicted β-cell dysfunction, insulin resistance, and increased numbers of cardiovascular risk factors such as dyslipidemia and elevated blood pressure.33 Others have shown that patients with NGT and elevated 1 h glucose levels have features of insulin resistance or impaired β-cell function by diverse measures: acute insulin response and disposition index (calculated from the frequently sampled intravenous glucose tolerance test (IVGTT)), decreased glucose disposal on hyperinsulinemic–euglycemic clamp, elevated markers of chronic inflammation (hsCRP and fibrinogen), and increased levels of hepatic transaminases.34–37 Moreover, increased 1 h glucose predicts increased cardiovascular risk. In patients with hypertensive NGT, elevated 1 h glucose levels are associated with higher blood pressure, increased vascular stiffness, and left ventricular hypertrophy.38 ,39 Thus, elevated 1 h postload glucose may represent a unique ‘subclinical’ form of hyperglycemia that is not accounted for in current models for clinical detection of prediabetes or diabetes, but which plays an important role in the pathophysiology of diabetes and diabetes-related cardiovascular complications.

    In the present study, we found that fasting serum levels of α-HB predicted insulin resistance, impaired insulin secretion kinetics, and abnormal glucose regulation during the 75 g OGTT—in particular, elevated 1 h post-load plasma glucose levels. Importantly, the predictive ability of α-HB for 1 h postload glucose levels is observed even in individuals with ostensibly normal insulin–glucose homeostasis by other standard clinical criteria. These findings indicate that α-HB has clinical utility for predicting subclinical hyperglycemia, even under conditions where other commonly used indicators of impaired insulin–glucose homeostasis are ostensibly normal. This suggests that measurement of serum α-HB could provide a rapid, inexpensive screening tool for detecting early, subclinical stages of the disease process leading to prediabetes/diabetes. In addition, measurement of α-HB in conjunction with fasting plasma glucose and HbA1c could enhance the ability to detect hyperglycemia that might otherwise only be apparent by OGTT testing in diabetic and non-diabetic individuals.

    Our findings are consistent with prior observations that elevated serum α-HB levels can identify individuals with impaired insulin–glucose homeostasis, and further indicate that this relationship is independent of BMI and other traditional diabetes risk factors. In the present study, no differences between α-HB tertiles were observed for age, gender, BMI, or HbA1c. α-HB was also shown to be independent of other novel biomarkers of diabetes risk such as L-GPC. The observation that α-HB does not associate with other traditional factors associated with diabetes risk but is associated with elevated 1 h glucose in the OGTT suggests it has truly independent value in identifying patients at risk for developing diabetes.

    The results of the present study suggest that elevated serum α-HB is associated with delayed insulin secretory kinetics and insufficient insulin secretion during the OGTT, as differences were observed in the early rise in insulin levels during the OGTT and the OGTT-derived version of the disposition index. These findings are consistent with prior observations that serum α-HB correlates inversely with indices of β-cell function during OGTT in the RISC and Botnia study cohorts.19 The importance of interpreting estimates of β-cell function within the context of prevailing insulin resistance has been well established, as insulin resistance is normally compensated for by increased insulin secretion.40 In the present study, a direct estimate of insulin secretion during the OGTT, AUCinsulin/glucose, trended lower according to α-HB tertile, though did not achieve statistical significance (similar to AUCc-peptide/glucose, data not shown). However, when adjusted for insulin resistance, the observed deficits were striking. It should be noted that the original formulation of the disposition index as reported by Kahn et al41 utilized an estimate of insulin secretion obtained over only the first 8 min of an IVGTT, which was relatively independent of the estimate of insulin sensitivity obtained over the next few hours. However, when the same insulin excursion is used in the estimate of insulin secretion and insulin resistance, as in the OGTT-derived version of the disposition index used here, interpretation becomes more complicated. Additional experiments investigating α-HB utilizing an IVGTT protocol will be helpful in this regard.

    In conclusion, we report the discovery that elevated α-HB is strongly predictive of abnormal 1 h postload plasma glucose levels and decreased insulin secretion in the spectrum of clinic patients undergoing OGTT for evaluation of suspected metabolic risk. In addition, we confirm prior reports that serum α-HB is associated with insulin resistance based on 2 h OGTT glucose levels. Remarkably, we find that α-HB is highly predictive of OGTT glucose responses, and calculated indices of insulin sensitivity and β-cell function—even among a subset of patients selected for having ostensibly normal indicators of insulin–glucose homeostasis. Hence, α-HB may have broad applications as a tool for detecting states of subclinical hyperglycemia, insulin resistance, and β-cell dysfunction that could indicate increased risk for diabetes and cardiovascular disease.

    Acknowledgments

    The authors wish to thank all the physicians who participated in this study: Sam Fillingane DO, Jackson, Mississippi; Tara Dall, MD, Delafield, Wisconsin; Tim Graham, MD, Salt Lake City, Utah; Graham Reedy, MD, Auburn Washington.

    References

    1. ↵
      Centers for Disease Control and Prevention. National diabetes fact sheet: national estimates and general information on diabetes and prediabetes in the United States, 2011. Atlanta, GA: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, 2011.
    2. ↵
      1. Go AS,
      2. Mozaffarian D,
      3. Roger VL,
      4. et al
      . Heart disease and stroke statistics—2013 update: a report from the American Heart Association. Circulation 2013;127:e6–245.
      OpenUrlFREE Full Text
    3. ↵
      1. DeFronzo RA,
      2. Abdul-Ghani MA
      . Preservation of beta-cell function: the key to diabetes prevention. J Clin Endocrinol Metab 2011;96:2354–66.
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Graham TE,
      2. Kahn BB
      . Tissue-specific alterations of glucose transport and molecular mechanisms of intertissue communication in obesity and type 2 diabetes. Horm Metab Res 2007;39:717–21.
      OpenUrlCrossRefPubMed
    5. ↵
      1. Groop PH,
      2. Forsblom C,
      3. Thomas MC
      . Mechanisms of disease: pathway-selective insulin resistance and microvascular complications of diabetes. Nat Clin Pract Endocrinol Metab 2005;1:100–10.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Bornfeldt KE,
      2. Tabas I
      . Insulin resistance, hyperglycemia, and atherosclerosis. Cell Metab 2011;14:575–85.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Alexander CM,
      2. Landsman PB,
      3. Teutsch SM,
      4. et al
      . NCEP-defined metabolic syndrome, diabetes, and prevalence of coronary heart disease among NHANES III participants age 50 years and older. Diabetes 2003;52:1210–14.
      OpenUrlAbstract/FREE Full Text
    8. ↵
      1. Tung TH,
      2. Shih HC,
      3. Tsai ST,
      4. et al
      . A community-based study of the relationship between insulin resistance/beta-cell dysfunction and diabetic retinopathy among type II diabetics in Kinmen, Taiwan. Ophthalmic Epidemiol 2007;14:148–54.
      OpenUrlCrossRefPubMed
    9. ↵
      1. Tabak AG,
      2. Herder C,
      3. Rathmann W,
      4. et al
      . Prediabetes: a high-risk state for diabetes development. Lancet 2012;379:2279–90.
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      1. Rasgon NL,
      2. Kenna HA
      . Insulin resistance in depressive disorders and Alzheimer's disease: revisiting the missing link hypothesis. Neurobiol Aging 2005;26(Suppl 1):103–7.
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Kan C,
      2. Silva N,
      3. Golden SH,
      4. et al
      . A systematic review and meta-analysis of the association between depression and insulin resistance. Diabetes Care 2013;36:480–9.
      OpenUrlAbstract/FREE Full Text
    12. ↵
      1. Reasner CA
      . Reducing cardiovascular complications of type 2 diabetes by targeting multiple risk factors. J Cardiovasc Pharmacol 2008;52:136–44.
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Borai A,
      2. Livingstone C,
      3. Ferns GA
      . The biochemical assessment of insulin resistance. Ann Clin Biochem 2007;44(Pt 4):324–42.
      OpenUrlAbstract/FREE Full Text
    14. ↵
      1. Herder C,
      2. Kowall B,
      3. Tabak AG,
      4. et al
      . The potential of novel biomarkers to improve risk prediction of type 2 diabetes. Diabetologia 2014;57:16–29.
      OpenUrlCrossRefPubMed
    15. ↵
      1. Kolberg JA,
      2. Gerwien RW,
      3. Watkins SM,
      4. et al
      . Biomarkers in type 2 diabetes: improving risk stratification with the PreDx (R) Diabetes Risk Score. Expert Rev Mol Diagn 2011;11:775–92.
      OpenUrlCrossRefPubMed
    16. ↵
      1. Wang TJ,
      2. Larson MG,
      3. Vasan RS,
      4. et al
      . Metabolite profiles and the risk of developing diabetes. Nat Med 2011;17:448–53.
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. Cobb J,
      2. Gall W,
      3. Adam KP,
      4. et al
      . A novel fasting blood test for insulin resistance and prediabetes. J Diabetes Sci Technol 2013;7:100–10.
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. Gall WE,
      2. Beebe K,
      3. Lawton KA,
      4. et al
      . Alpha-hydroxybutyrate is an early biomarker of insulin resistance and glucose intolerance in a nondiabetic population. PLoS ONE 2010;5:e10883.
      OpenUrlCrossRefPubMed
    19. ↵
      1. Ferrannini E,
      2. Natali A,
      3. Camastra S,
      4. et al
      . Early metabolic markers of the development of dysglycemia and type 2 diabetes and their physiological significance. Diabetes 2013;62:1730–7.
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. Montane J,
      2. Cadavez L,
      3. Novials A
      . Stress and the inflammatory process: a major cause of pancreatic cell death in type 2 diabetes. Diabetes Metab Syndr Obes 2014;7:25–34.
      OpenUrl
    21. ↵
      1. Abdul-Ghani MA,
      2. DeFronzo RA
      . Plasma glucose concentration and prediction of future risk of type 2 diabetes. Diabetes Care 2009;32(Suppl 2):S194–8.
      OpenUrlFREE Full Text
    22. ↵
      1. Matsuda M,
      2. DeFronzo RA
      . Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp. Diabetes Care 1999;22:1462–70.
      OpenUrlAbstract/FREE Full Text
    23. ↵
      1. Anderwald C,
      2. Anderwald-Stadler M,
      3. Promintzer M,
      4. et al
      . The Clamp-Like Index: a novel and highly sensitive insulin sensitivity index to calculate hyperinsulinemic clamp glucose infusion rates from oral glucose tolerance tests in nondiabetic subjects. Diabetes Care 2007;30:2374–80.
      OpenUrlAbstract/FREE Full Text
    24. ↵
      1. Retnakaran R,
      2. Shen S,
      3. Hanley AJ,
      4. et al
      . Hyperbolic relationship between insulin secretion and sensitivity on oral glucose tolerance test. Obesity 2008;16:1901–7.
      OpenUrlCrossRef
    25. ↵
      1. Hlatky MA,
      2. Greenland P,
      3. Arnett DK,
      4. et al
      . Criteria for evaluation of novel markers of cardiovascular risk: a scientific statement from the American Heart Association. Circulation 2009;119:2408–16.
      OpenUrlAbstract/FREE Full Text
    26. ↵
      1. Hosmer DW,
      2. Hosmer T,
      3. Le Cessie S,
      4. et al
      . A comparison of goodness-of-fit tests for the logistic regression model. Stat Med 1997;16:965–80.
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Pencina MJ,
      2. D'Agostino RB Sr.,
      3. D'Agostino RB Jr.,
      4. et al
      . Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27:157–72.; discussion 207-12.
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Pencina MJ,
      2. D'Agostino RB Sr.,
      3. Steyerberg EW
      . Extensions of net reclassification improvement calculations to measure usefulness of new biomarkers. Stat Med 2011;30:11–21.
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. Karve A,
      2. Hayward RA
      . Prevalence, diagnosis, and treatment of impaired fasting glucose and impaired glucose tolerance in nondiabetic U.S. adults. Diabetes Care 2010;33:2355–9.
      OpenUrlAbstract/FREE Full Text
    30. ↵
      1. Abdul-Ghani MA,
      2. Abdul-Ghani T,
      3. Ali N,
      4. et al
      . One-hour plasma glucose concentration and the metabolic syndrome identify subjects at high risk for future type 2 diabetes. Diabetes Care 2008;31:1650–5.
      OpenUrlAbstract/FREE Full Text
    31. ↵
      1. Abdul-Ghani MA,
      2. Williams K,
      3. DeFronzo RA,
      4. et al
      . What is the best predictor of future type 2 diabetes? Diabetes Care 2007;30:1544–8.
      OpenUrlAbstract/FREE Full Text
    32. ↵
      1. Abdul-Ghani MA,
      2. Lyssenko V,
      3. Tuomi T,
      4. et al
      . Fasting versus postload plasma glucose concentration and the risk for future type 2 diabetes: results from the Botnia Study. Diabetes Care 2009;32:281–6.
      OpenUrlAbstract/FREE Full Text
    33. ↵
      1. Bianchi C,
      2. Miccoli R,
      3. Trombetta M,
      4. et al
      . Elevated 1-hour postload plasma glucose levels identify subjects with normal glucose tolerance but impaired beta-cell function, insulin resistance, and worse cardiovascular risk profile: the GENFIEV study. J Clin Endocrinol Metab 2013;98:2100–5.
      OpenUrlCrossRefPubMed
    34. ↵
      1. Marini MA,
      2. Succurro E,
      3. Frontoni S,
      4. et al
      . Insulin sensitivity, beta-cell function, and incretin effect in individuals with elevated 1-hour postload plasma glucose levels. Diabetes Care 2012;35:868–72.
      OpenUrlAbstract/FREE Full Text
    35. ↵
      1. Succurro E,
      2. Marini MA,
      3. Arturi F,
      4. et al
      . Elevated one-hour post-load plasma glucose levels identifies subjects with normal glucose tolerance but early carotid atherosclerosis. Atherosclerosis 2009;207:245–9.
      OpenUrlCrossRefPubMedWeb of Science
    36. ↵
      1. Succurro E,
      2. Arturi F,
      3. Grembiale A,
      4. et al
      . One-hour post-load plasma glucose levels are associated with elevated liver enzymes. Nutr Metab Cardiovasc Dis 2011;21:713–18.
      OpenUrlCrossRefPubMed
    37. ↵
      1. Sesti G,
      2. Fiorentino TV,
      3. Succurro E,
      4. et al
      . Elevated 1-h post-load plasma glucose levels in subjects with normal glucose tolerance are associated with unfavorable inflammatory profile. Acta Diabetol 2014.
    38. ↵
      1. Sciacqua A,
      2. Miceli S,
      3. Carullo G,
      4. et al
      . One-hour postload plasma glucose levels and left ventricular mass in hypertensive patients. Diabetes Care 2011;34:1406–11.
      OpenUrlAbstract/FREE Full Text
    39. ↵
      1. Sciacqua A,
      2. Maio R,
      3. Miceli S,
      4. et al
      . Association between one-hour post-load plasma glucose levels and vascular stiffness in essential hypertension. PLoS ONE 2012;7:e44470.
      OpenUrlCrossRefPubMed
    40. ↵
      1. Ahren B,
      2. Pacini G
      . Importance of quantifying insulin secretion in relation to insulin sensitivity to accurately assess beta cell function in clinical studies. Eur J Endocrinol 2004;150:97–104.
      OpenUrlAbstract
    41. ↵
      1. Kahn SE,
      2. Prigeon RL,
      3. McCulloch DK,
      4. et al
      . Quantification of the relationship between insulin sensitivity and β-cell function in human subjects: evidence for a hyperbolic function. Diabetes 1993;42:1663–72.
      OpenUrlAbstract/FREE Full Text

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

      Files in this Data Supplement:

    Footnotes

    • Contributors SAV, TEG, JVP, SV, MS, RC, GRW, and TD designed the study. JVP analyzed the data. SAV, TEG, JVP, and DLT wrote/reviewed/edited the manuscript and interpreted the findings. RC interpreted the findings and contributed to the discussion. SV, MS, JVM, TD, and GRW interpreted the findings.

    • Funding This work was supported by Health Diagnostic Laboratory, Inc.

    • Competing interests All authors are employees of Health Diagnostic Laboratory, Inc (HDL), except for TG who received consulting fees from HDL.

    • Ethics approval Copernicus Group IRB.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement No additional data are available.