Enhanced Predictive Capability of a 1-Hour Oral Glucose Tolerance Test

Diabetes Care Volume 41, January 2018 171

Enhanced Predictive Capability Manan Pareek,1,2,3 Deepak L. Bhatt,1 Mette L. Nielsen,2 Ram Jagannathan,4 of a 1-Hour Oral Glucose Tolerance Karl-Fredrik Eriksson,5 Peter M. Nilsson,5 Michael Bergman,6 and Test: A Prospective Population- Michael H. Olsen2,3 Based Cohort Study

Diabetes Care 2018;41:171–177 | https://doi.org/10.2337/dc17-1351

OBJECTIVE To examine whether the 1-h blood glucose measurement would be a more suitable screening tool for assessing the risk of diabetes and its complications than the 2-h measurement.

RESEARCH DESIGN AND METHODS We conducted a prospective population-based cohort study of 4,867 men, randomly selected from prespecified birth cohorts between 1921 and 1949, who underwent an oral glucose tolerance test with blood glucose measurements at 0, 1, and 2 h. Sub- jects were followed for up to 39 years, with registry-based recording of events. Discriminative abilities of elevated 1-h (‡8.6 mmol/L) versus 2-h (‡7.8 mmol/L) 1Brigham and Women’s Hospital Heart & Vascu- glucose for predicting incident type 2 diabetes, vascular complications, and mortality lar Center, Harvard Medical School, Boston, MA 2 were compared using Kaplan-Meier analysis, Cox proportional hazards regression, Cardiovascular and Metabolic Preventive Clinic, fi Department of Endocrinology, Centre for Individ- and net reclassi cation improvement. ualized Medicine in Arterial Diseases, Odense University Hospital, Odense, Denmark RESULTS 3Cardiology Section, Department of Internal RISK METABOLIC AND CARDIOVASCULAR Median age was 48 years (interquartile range [IQR] 48–49). During follow-up (median Medicine, Holbaek Hospital, Holbaek, Denmark 4 33 years [IQR 24–37]), 636 (13%) developed type 2 diabetes. Elevated 1-h glucose Center for Healthful Behavior Change, Depart- – P < ment of Population Health, New York University was associated with incident diabetes (hazard ratio 3.40 [95% CI 2.90 3.98], 0.001) School of Medicine, New York, NY and provided better risk assessment than impaired glucose tolerance (Harrell con- 5Department of Clinical Sciences and Lund Uni- cordance index 0.637 vs. 0.511, P < 0.001). Addition of a 1-h measurement in subjects versity Diabetes Centre, Lund University, Skane˚ stratified by fasting glucose provided greater net reclassification improvement than University Hospital, Malmo,¨ Sweden 6Division of Endocrinology, Diabetes and Metab- the addition of a 2-h measurement (0.214 vs. 0.016, respectively). Finally, the 1-h olism, Department of Medicine, and Diabetes Pre- glucose was significantly associated with vascular complications and mortality. vention Program, New York University Langone Health, New York University School of Medicine, CONCLUSIONS New York, NY The 1-h blood glucose level is a stronger predictor of future type 2 diabetes than the Corresponding author: Manan Pareek, mpareek@ 2-h level and is associated with diabetes complications and mortality. bwh.harvard.edu and [email protected]. Received 6 July 2017 and accepted 4 October 2017. Type 2 diabetes is associated with significant morbidity and mortality and represents a This article contains Supplementary Data online major burden on health care systems worldwide (1,2). Several randomized clinical trials at http://care.diabetesjournals.org/lookup/ provide evidence that type 2 diabetes can be prevented or at least postponed with suppl/doi:10.2337/dc17-1351/-/DC1. lifestyle modification and drug therapy, which makes identifying high-risk individuals © 2017 by the American Diabetes Association. particularly important (3–6). Traditionally, prediabetes has been defined as impaired Readers may use this article as long as the work is properly cited, the use is educational and not fasting glucose (IFG) or impaired glucose tolerance (IGT) during a 2-h oral glucose for profit, and the work is not altered. More infor- tolerance test (OGTT), and interventional studies thus far have predominantly included mation is available at http://www.diabetesjournals subjects with IGT (7,8). However, not all subjects with prediabetes develop type 2 .org/content/license. 172 One-Hour Oral Glucose Tolerance Test Diabetes Care Volume 41, January 2018

diabetes, and conversely, a significant the final study population consisted of [IQR]) (non–normally distributed vari- number without prediabetes progress to 4,867 men (Supplementary Fig. 1). In ad- ables). Categorical variables are presented type 2 diabetes (9). Accumulating longitu- dition, 132 women recruited between as counts and corresponding percentages. dinal evidence, pioneered in particular by 25 January 1977 and 6 January 1984 un- Kaplan-Meier analysis with the log-rank Abdul-Ghani and colleagues (10–18), sug- derwent baseline OGTT. One was excluded test and Cox proportional hazards regres- gests that the 1-h postload glucose level owing to known diabetes and 3 owing to sion with Harrell concordance index (C in- during OGTT with a cutoff $8.6 mmol/L emigration, but none of them underwent dex), assuming an uncensored policy for (155 mg/dL) may be an early marker of intervention, leaving a total of 128 female handling ties, were used for assessment IGT and subsequent type 2 diabetes that subjects. The Malmo¨ Preventive Project of discriminative ability for postload glu- is potentially more useful than either fast- was approved by the ethics committee cose measurements, including compari- ing or 2-h glucose levels. Furthermore, of Lund University and conducted in ac- sons between predefined risk groups from a pathophysiological perspective, cordance with the Declaration of Helsinki. (27). Hazard ratios (HRs) were reported the 1-h glucose level is intriguing owing All participants gave informed consent. unadjusted and adjusted for age, BMI, to its seemingly strong correlation with IFG, triglycerides, and family history of markers of both insulin secretion and sen- Baseline Variables diabetes. Furthermore, the ability of 1-h sitivity (10,19,20). However, practical clin- Participants used a self-administered and 2-h postload glucose measurements ical implications of using a 1-h glucose questionnaire to provide information on to enhance prognostication in addition to measurement for prediction of type 2 di- lifestyle and medical history, including a fasting blood glucose was tested with cat- fi abetes and its associated complications history of diabetes in rst-degree relatives, egorical net reclassification improvement are less clear. Therefore, the aim of this cardiovascular disease, and current medi- (28). All glucose measurements were as- fi study was to examine whether the 1-h cation. Prevalent diabetes was de ned as sessed in a binary fashion, using the fol- blood glucose measurement would be a self-reported diabetes or according to the lowing cut points: IFG, fasting blood more suitable screening tool for risk assess- 1985WorldHealthOrganizationcriteria(24). glucose $5.6 mmol/L (100 mg/dL); ele- ment than the 2-h blood glucose alone. Blood glucose and serum lipids were ob- vated 1-h blood glucose, $8.6 mmol/L $ tained after an overnight ( 10 h) fast. (155 mg/dL); and IGT, 2-h BG $7.8 RESEARCH DESIGN AND METHODS Blood glucose was analyzed using the glu- mmol/L (140 mg/dL) (7,10). Analyses – All subjects were participants of the cose oxidase (1974 1977) or the hexoki- were performed at 12 years and at max- – Malmo¨ Preventive Project (1974–1992), nase (1977 1992) method. Serum lipids imal available follow-up, respectively. ’ a population-based case-finding program were analyzed using the local laboratory s This was because of the possibility of standard methods. A 2-h OGTT was per- with the objective of identifying high-risk 2 the intervention among high-risk individ- adults suitable for preventive measures. formed by ingestion of 30 g glucose/m uals during the first 12 years after recruit- Inhabitants of Malmo,¨ Sweden, belonging body surface area (DuBois formula), with ment affecting the study outcomes to prespecified birth cohorts (1921–1949) glucose levels determined at 0, 20, 40, 60, 90, because even though we attempted to were invited for an examination of cardio- and 120 min (25). Analyses of postload minimize the implications of the interven- vascular risk factors, alcohol abuse, and glucose levels were focused on measure- tion by excluding individuals who were breast cancer. Progressively older men ments at 1 h and 2 h. subjected to it, there may have been a were recruited later during the course Outcomes spillover effect to the control subjects, of the study. The participation rate was We used national and local registries to given the population-based nature of – 71% (21 23). During the initial phase of record clinical end points. Besides the di- the study. Results obtained from the the study (9 September 1974 to 31 May agnosis itself, the date of diagnosis is also women were reported separately. There- 1978 [both inclusive]), 7,200 men were coded in these registries. All events (type 2 fore, unless explicitly stated otherwise, all consecutively included, of whom 5,364 diabetes, myocardial infarction, diabetic analyses and conclusions were based on fi underwent baseline OGTT. After exclu- retinopathy, and diabetes with peripheral the larger male population. The signi - sion of subjects with a missing or invalid vascular complications, including ulcer) cance level was 5% (two sided), and no 1-h glucose measurement (7 of 5,364) or were defined according to the relevant adjustments for multiple comparisons known diabetes (100 of 5,364), and those ICD-7 to ICD-10 codes (Supplementary were made, as the study was considered who emigrated (71 of 5,364), 5,182 indi- Table 1). Reported validities in the Swed- exploratory. All analyses were performed viduals were left. Of these, a subset of ish National Inpatient Register were high with IBM SPSS Statistics 23 (IBM, Armonk, d individuals at high risk including those for all diagnoses (26). Mortality follow-up NY) and Stata/IC 15 (StataCorp, College with hypertension, hyperlipidemia, diabe- was based on the national registry on Station, TX). fi tes, and IGT, with the latter de ned per causes of mortality at the Swedish Central RESULTS local standards as a 2-h blood glucose level Bureau of Statistics. As follow-up was lim- $ fi 7.0 mmol/L (126 mg/dL) con rmed on a ited to 31 December 2013, the potential Table 1 shows baseline anthropometric, d separate day (315 of 5,182) underwent minimum and maximum follow-up times clinical, and laboratory characteristics of the intervention. The intervention for IGT were 35 and 39 years, respectively. study participants. All 4,867 subjects were comprised of dietary advice and in- male, with a median age of 48 years (IQR creased physical activity, most often in- Statistical Methods 48–49; range 27–52), mean BMI 24.8 6 cluding frequent visits at an outpatient Continuous variables are presented as 3.1 kg/m2, mean body surface area 1.93 6 clinic for up to 12 years after inclusion. mean 6 SD (normally distributed vari- 0.15 m2, and mean total cholesterol level We also excluded these subjects; thus, ables) or median (interquartile range 5.8 6 1.0 mmol/L (225 6 40 mg/dL). care.diabetesjournals.org Pareek and Associates 173

Table 1—Baseline characteristics of the study population, stratiﬁed according to glycemic category Total NGT/1h-normal NGT/1h-high IGT/1h-normal IGT/1h-high P Study population 4,867 3,139 (65) 1,564 (32) 32 (1) 132 (3) Age, years 48 (48–49) 48 (48–49) 48 (48–49) 48 (47–49) 48 (48–49) ,0.001a Active smoking 2,582 (53) 1,583 (50) 934 (60) 8 (25) 57 (43) ,0.001b Family history of diabetes 655 (13) 375 (12) 254 (16) 6 (19) 20 (15) 0.001b Sedentary lifestyle 2,757 (57) 1,740 (55) 914 (58) 20 (63) 83 (63) 0.09b BMI, kg/m2 24.8 6 3.1 24.6 6 3.0 25.1 6 3.3 25.7 6 3.6 25.7 6 3.3 0.001c Systolic blood pressure, mmHg 130 6 16 128 6 14 134 6 17 133 6 19 143 6 21 ,0.001c Diastolic blood pressure, mmHg 88 6 10 87 6 10 90 6 10 90 6 14 94 6 14 ,0.001c Fasting blood glucose, mmol/L [mg/dL] 4.6 6 0.5 4.5 6 0.5 4.8 6 0.5 4.8 6 0.6 5.0 6 0.6 ,0.001c [83 6 10] [81 6 9] [87 6 10] [87 6 11] [90 6 11] 20-min blood glucose, mmol/L [mg/dL] 7.6 6 1.4 7.3 6 1.3 8.2 6 1.3 6.7 6 1.2 7.9 6 1.4 ,0.001c [137 6 25] [131 6 23] [148 6 24] [121 6 22] [142 6 26] 40-min blood glucose, mmol/L [mg/dL] 9.0 6 1.8 8.2 6 1.3 10.5 6 1.5 8.0 6 1.2 10.5 6 1.6 ,0.001c [162 6 32] [147 6 24] [188 6 27] [144 6 22] [189 6 28] 1-h blood glucose, mmol/L 8.0 6 1.9 6.8 6 1.1 10.0 6 1.2 7.4 6 0.7 11.0 6 1.6 ,0.001c [mg/dL] [143 6 35] [123 6 20] [180 6 22] [133 6 13] [198 6 28] 2-h blood glucose, mmol/L 5.2 6 1.3 4.9 6 1.1 5.4 6 1.2 8.1 6 0.3 8.6 6 0.7 ,0.001c [mg/dL] [94 6 23] [89 6 20] [98 6 21] [146 6 6] [155 6 12] Total cholesterol, mmol/L 5.8 6 1.0 5.8 6 1.0 5.9 6 1.0 5.7 6 1.0 6.1 6 1.0 ,0.001c [mg/dL] [225 6 40] [223 6 39] [228 6 40] [222 6 40] [238 6 39] Triglycerides, mmol/L 1.4 (1.0–1.8) 1.3 (1.0–1.7) 1.5 (1.1–2.1) 1.4 (1.1–2.3) 1.6 (1.1–2.1) ,0.001a [mg/dL] [121 (92–163)] [116 (89–152)] [133 (97–182)] [120 (99–205)] [140 (101–184)] Continuous variables are presented as mean 6 SD (normally distributed variables: BMI, blood pressure, blood glucose, and total cholesterol) or median (IQR) (non–normally distributed variables: age and triglycerides). Data for the study population (ﬁrst row) and categorical variables (active smoking, family history of diabetes, and sedentary lifestyle) are presented as counts and corresponding percentages. aKruskal-Wallis rank test. bPearson x2 test. cOne-way ANOVA.

A history of a sedentary lifestyle (leisure 636 (13%) individuals had been diagnosed 1-h blood glucose levels (Table 3 and time mostly spent on sedentary activi- with type 2 diabetes, corresponding to Supplementary Table 3). ties, including reading, television, and incidence densities of 1.2 and 4.8 per cinema) was reported in 2,757 (57%), 1,000 person-years, respectively. Cu- Discriminative Ability The presence of an elevated 1-h blood glu- and 655 (13%) had a first-degree relative mulative incidences were lowest in sub- cose level was associated with significantly with diabetes. Normal postload blood glu- jects with NGT/1h-normal, highest in greater discriminative ability than IGT cose levels (NGT/1h-normal) were found subjects with IGT/1h-high, and greater based on 2-h blood glucose at both 12 in 3,139 (65%), elevated glucose at 1 h in subjects with NGT/1h-high than in years (C index 0.698 vs. 0.553, P , 0.001) only (NGT/1h-high) was found in 1,564 those with IGT/1h-normal (Table 2 and and 39 years (C index 0.637 vs. 0.511, P , (32%), and IGT only (IGT/1h-normal) was Supplementary Table 2). 0.001). Additionally, the presence of IFG found in in 32 (1%), and 132 (3%) had both or elevated 1-h blood glucose was associ- elevated 1-h glucose and IGT (IGT/1h- Hazard Risk The Kaplan-Meier plot shows the unad- ated with a greater risk of type 2 diabetes high). Collectively, 1,696 (35%) individuals justed type 2 diabetes–free survival for than IFG or IGT at 12 years (HR 7.22 [95% had elevated glucose at 1 h, and 164 (3%) risk categories based on postload blood CI 3.93–13.26], P , 0.001, C index 0.720, had IGT. Glucose levels at 1 h and 2 h were glucose measurements (Fig. 1A and vs. HR 4.64 [95% CI 2.69–7.99], P , 0.001, C moderately correlated (Pearson r = 0.332, Supplementary Figs. 3 and 4). Congruent index 0.600; P for difference ,0.001) and P , 0.001) (Supplementary Fig. 2). with the above, both crude and adjusted 39 years (HR 3.31 [95% CI 2.82–3.88], P , Cumulative Events HRs for development of type 2 diabetes 0.001, C index 0.636, vs. HR 2.28 [95% CI Median follow-up time was 33 years (IQR were significantly greater in risk cate- 1.83–2.84], P , 0.001, C index 0.537; P for 24–37). At 12 and 39 years, 65 (1%) and gories including subjects with elevated difference , 0.001).

Table 2—Cumulative incidence and incidence density of type 2 diabetes, stratiﬁed according to glycemic category Total NGT/1h-normal NGT/1h-high IGT/1h-normal IGT/1h-high Study population, n (%) 4,867 3,139 (65) 1,564 (32) 32 (1) 132 (3) Type 2 diabetes cumulative incidence, n (%) 12 years 65 (1) 17 (0.5) 39 (2) 0 9 (7) 39 years 636 (13) 259 (8) 343 (22) 4 (13) 30 (23) Type 2 diabetes incidence density, per 1,000 person-years 12 years 1.2 0.5 2.2 0 6.3 39 years 4.8 2.9 8.8 4.4 9.6 174 One-Hour Oral Glucose Tolerance Test Diabetes Care Volume 41, January 2018

highermortalityriskthanthosewith NGT/1h-normal, whereas those with IGT/ 1h-normal did not (Table 3, Supplementary Table 5, and Fig. 1B). The risk of myocardial infarction or fatal ischemic heart disease was also greater among subjects with elevated 1-h glucose levels versus NGT/1h-normal. Finally, elevated 1-h glucose levels were associated with greater risks of retinopathy and peripheral vascular complications. No differences were detected between IGT/1h-normal and NGT/1h-normal (Supplementary Tables 6 and 7).

Glucose Measurements at Earlier Time Points To highlight the importance of 1-h blood glucose measurements, we performed supplemental analyses comparing glucose levels obtained at 20 and 40 min with those obtained at fasting, 1 h, and 2 h in predicting incident type 2 diabetes and all-cause mortality (Supplementary Tables 8 and 9). Standardized HRs for glucose measurements obtained at time points earlier than 2 h were generally greater among the younger and thinner halves of the study population, but only the interaction terms between BMI and 40-min and 1-h glucose, and between age and 40-min glucose, were statistically sig- niﬁcant (Supplementary Table 10).

Female Subjects Median age at baseline was 39 years (IQR 38–40; range 29–56). Median follow-up was 36 years (IQR 35–36). Mean blood glucose concentrations were 4.7 6 0.5 mmol/L (84 6 8mg/dL)atfasting, 7.5 6 1.9 mmol/L (135 6 34 mg/dL) at 1 h, and 5.8 6 1.2 mmol/L (104 6 22 mg/dL) at 2 h. Four (3%) had an elevated Figure 1—A: Kaplan-Meier plot showing unadjusted risk of type 2 diabetes. B:Kaplan-Meierplotshowing 2-h glucose concentration, and 41 (32%) unadjusted survival. hadanelevatedlevelat1h.Accordingly,86 (67%) were categorized as NGT/1h-normal, Reclassification Guideline-Based Screening 38 (30%) were categorized as NGT/1h-high, The abilities of elevated 1-h and 2-h blood Restricting analyses to subjects in whom 1 (1%) was categorized as IGT/1h-normal, glucose levels to reclassify risk in addition contemporary guidelines suggest screen- and 3 (2%) were categorized as IGT/1h- to fasting blood glucose measurements ing for diabetes (8), i.e., subjects aged $45 high. None had developed type 2 diabetes were tested in separate models. Two risk years or subjects with a BMI $25 kg/m2 at 12 years. At complete follow-up, 13 categories were applied, and reclassifica- and at least one additional risk factor (. 99% (10%) had been diagnosed with type 2 dition could only be present if postload glu- of our study population met the criteria abetes, corresponding to an incidence den- cose levels were elevated in subjects with for screening) did not alter the results. sity of 3.2/1,000 person-years. Cumulative normal fasting glucose. As such, the addi- incidences (and incidence densities per tion of 1-h glucose was associated with Diabetes Complications and Mortality 1,000 person-years) in the four categories greater net reclassification improvement Similar overall patterns were observed were 8% (2.5), 13% (4.2), 0% (0), and 33% than the addition of 2-h glucose at both when we used all-cause mortality as the (13.6), respectively. Compared with the 12 years (0.308 vs. 0.066) and 39 years outcome, i.e., subjects with elevated 1-h NGT/1h-normal group, HRs were 1.72 (95 CI (0.214 vs. 0.016) (Supplementary Table 4). blood glucose levels had a significantly 0.55–5.43), P = 0.35, for the NGT/1h-high care.diabetesjournals.org Pareek and Associates 175

– Table 3—Crude and adjusted HRs for incident type 2 diabetes and all-cause hypertrophy (19,20,29 31). However, mortality, stratified according to glycemic category very few reports of the clinical conse- Risk groupa Crude HR (95% CI) P Adjusted HR (95% CI)b P quences have been published (32–35), and although associations with macrovas- Type 2 diabetes cular events and mortality have been 12 years NGT/1h-high 4.75 (2.69–8.39) ,0.001 3.87 (2.16–6.93) ,0.001 shown, only one study of 1,945 subjects IGT/1h-normalc ddddincluded glucose levels at 2 h and indi- IGT/1h-high 13.76 (6.13–30.87) ,0.001 9.00 (3.83–21.16) ,0.001 cated that 1-h glucose levels predicted 39 years all-cause mortality among subjects with NGT/1h-high 3.39 (2.88–3.99) ,0.001 2.93 (2.48–3.46) ,0.001 NGT (35). Our study demonstrates for – – IGT/1h-normal 1.56 (0.58 4.18) 0.38 1.17 (0.43 3.15) 0.76 the first time an association between el- IGT/1h-high 3.71 (2.54–5.41) ,0.001 2.76 (1.87–4.06) ,0.001 evated 1-h blood glucose levels and ad- All-cause mortality verse cardiovascular prognosis, including 12 years NGT/1h-high 1.41 (1.13–1.75) 0.002 1.35 (1.07–1.69) 0.01 microvascular complications, in subjects IGT/1h-normal 1.57 (0.50–4.90) 0.44 1.42 (0.45–4.45) 0.55 with NGT, while at the same time showing IGT/1h-high 1.77 (1.03–3.05) 0.04 1.58 (0.91–2.74) 0.11 the lack of such an association in subjects 39 years with IGT only. Furthermore, the 1-h blood NGT/1h-high 1.29 (1.19–1.39) ,0.001 1.29 (1.19–1.39) ,0.001 glucose level predicted mortality, whereas – – IGT/1h-normal 0.81 (0.49 1.32) 0.39 0.80 (0.49 1.31) 0.38 isolated IGT did not. IGT/1h-high 1.30 (1.05–1.60) 0.02 1.27 (1.02–1.57) 0.03 Although the 2-h OGTT has received a b The NGT/1h-normal group serves as baseline (comparator group). The multivariable Cox less emphasis by the American Diabetes proportional hazards regression model was adjusted for age, BMI, IFG, triglycerides, and family history of diabetes. cThere were no events in this small group at 12 years. Association (8), the majority of evidence for intervention among subjects with prediabetes comes from studies of overweight or obese individuals with IGT (3–6). group and 6.68 (95% CI 0.82–54.52), upon intervention have not been estab- Furthermore, both fasting glucose and P = 0.08, for the IGT/1h-high group. lished. Our study therefore extends previ- HbA1c have limited sensitivity and speci- ous findings by using a simple and clinically ficity for detecting subjects at risk, espe- CONCLUSIONS applicable approach, directly comparing cially because of high false-negative rates, Our results demonstrate that in middle- risks in groups stratified according to post- and the current study provides evidence aged men, for whom screening for type 2 load blood glucose levels, in a larger popu- that the 1-h OGTT is superior to fasting diabetes would be recommended, the 1-h lation sample with longer follow-up duration. glucose alone for identifying high-risk blood glucose level is a significant predic- An elevated 1-h blood glucose level was as- subjects (7,10,12,13,15,16,36–38). Hence, tor of future type 2 diabetes, regardless of sociated with a significantly increased risk of a more sensitive strategy is needed, which the 2-h blood glucose level. In addition, incident type 2 diabetes at both short-term atthesametimeissimpletouseinapri- the 1-h blood glucose level has greater and long-term follow-up and further pro- mary care setting and less time consuming detection rates (higher sensitivity) than vided significant net reclassification im- and more convenient than the 2-h OGTT. the 2-h blood glucose level and can cor- provement in subjects categorized per The current study extends the results rectly reclassify subjects with traditionally fasting blood glucose. Consistent with prior from a previous report from the Malmo¨ defined prediabetes, especially with findings, subjects in the IGT/1h-normal Preventive Project and shows that the longer duration of follow-up (high speci- group constituted a minority (18), and, im- 1-h time limit provides not only the best ficity). Importantly, the 1-h blood glucose portantly, few individuals categorized as compromise in terms of time consump- level is also associated with diabetes com- such progressed to diabetes, without any tion (performs equivalent to or better than plications and mortality. significant difference from persons in the glucose measurement at 90 min) but also Prior studies have focused mainly on NGT/1h-normal group. Conversely, the prognostic ability (performing better than the ability of 1-h glucose measurements IGT/1h-high subgroup contained individuals glucose measurements at 20 and 40 min) to correctly subdivide subjects with NGT at very high risk of future type 2 diabetes. (17). The proposed cut point for 1-h blood into those with low(er) and high(er) risk of Therefore, virtually all individuals with true glucose identifies a substantially larger future type 2 diabetes and focused less on IGT at 2 h who progressed to manifest type 2 proportion of subjects at high risk com- the direct comparison between 1-h and diabetes were captured by an elevated 1-h pared with conventional IGT (10,12,13), and 2-h OGTT from a clinical standpoint (10,12– glucose level. its use could lead to more widespread 15). Previous results from the MalmoPre-¨ The ability of 1-h blood glucose levels to preventive efforts. However, since subjects ventive Project have further suggested predict diabetes complications and mor- for whom screening for type 2 diabetes is that the 1-h blood glucose level also tality is less well studied. Elevated glucose recommended already have an adverse risk adds significant predictive value to clinical levels at 1 h are associated with adverse profile (7,8), using a 1-h OGTT should lead risk factors (16,17). However, such strat- metabolic and cardiovascular changes, re- to a reduced burden of both diabetes and egies do not allow for the omission of the flected by body composition, cholesterol its complications, without an excess risk 2-h OGTT or preclude the use of complex levels, and subclinical target organ damage, of harm. multivariable clinical prediction models, including arterial stiffness, carotid intima- The pathophysiological significance of for which clear cutoff values and benefits media thickness, and left ventricular elevated 1-h blood glucose levels is not 176 One-Hour Oral Glucose Tolerance Test Diabetes Care Volume 41, January 2018

fully understood. Proposedly, these indi- had a slightly greater impact on the 2-h measurements, especially with longer dura- viduals may be at an intermediate stage level than the 1-h level, improving the rel- tion of follow-up. The 1-h blood glucose level between NGT and IGT or represent a phe- ative sensitivity of the 1-h measurement, is further associated with diabetes complica- notype distinct from that of individuals with but significant differences have not been tions and mortality. Finally, simple logistics IGT (10,19,20). In subjects with NGT, an shown for 75 g versus 100 g, and it is not favor the use of a 1-h versus a 2-h OGTT. elevated 1-h glucose level is associated likely that important differences would ex- Therefore, 1-h blood glucose should be con- with insulin resistance to a degree similar ist for the glucose load used in our study sidered as a replacement for 2-h blood glu- to that seen among individuals with IGT. compared with the standard 75-g load cose as the preferred marker of IGT. Insulin secretion is also affected, albeit to (43). In contrast with the well-established a lesser extent (19,20). Furthermore, the partitions for fasting glucose and 2-h risk of future type 2 diabetes associated OGTT, the cut point of 8.6 mmol/L for Acknowledgments. The authors sincerely with elevated 1-h blood glucose levels 1-h OGTT has been less thoroughly inves- thank data managers Anders Dahlin and Hakan˚ was temporally persistent but greater in tigated and was derived from the San Andersson at the Department of Clinical Sciences, the IGT/1h-high group than in the NGT/ Antonio Heart Study (10). However, use of Lund University, Skane˚ University Hospital, for 1h-normal group, particularly at shorter a cut point derived from our own cohort help with retrieving data. Funding andDuality ofInterest. This study was follow-up. These findings, in addition to would have inflated its utility compared funded by the Danish Diabetes Academy, sup- those stated above, largely support the first with 2-h OGTT. An additional limitation ported by the Novo Nordisk Foundation. D.L.B. theory and further support the concept that might be attributed to the use of two discloses the following relationships: Advisory worsening b-cell function, not insulin resis- different approaches for measurement board, Cardax, Elsevier Practice Update Cardiol- ogy, Medscape Cardiology, and Regado Biosci- tance, is the main culprit involved in pro- of glucose levels, although the glucose ences; Board of directors, Boston VA Research gressive glucometabolic deterioration (13). oxidase and hexokinase methods have Institute and Society of Cardiovascular Patient Our results highlight the potential ben- been shown to deliver well-correlated Care; Chair, American Heart Association Quality efit of targeting individuals with an ele- results (44). The generalizability of our re- OversightCommittee;Data-monitoringcommit- vated 1-h blood glucose level and provide a sults beyond white men may be limited, tees, Cleveland Clinic, Duke Clinical Research Institute, Harvard Clinical Research Institute, strong rationale for an interventional study but there is no evidence to suggest sex- or Mayo Clinic, Mount Sinai School of Medicine, in which subjects are selected based on this race-related differences in the prognostic and Population Health Research Institute; hon- abnormality. Should preventive efforts ability of 1-h versus 2-h OGTT. However, oraria, American College of Cardiology (Senior prove beneficial in this relatively large women may have slightly lower mean con- Associate Editor, Clinical Trials and News,acc. group, it would become prudent to concentrations of 1-h glucose, with a slower org), Belvoir Publications (Editor in Chief, Har- fi vard Heart Letter), Duke Clinical Research Insti- sider 1-h blood glucoseasareplacement return to baseline levels (45). This ts well tute (clinical trial steering committees), Harvard for 2-h blood glucose. with the observation that traditionally de- Clinical Research Institute (clinical trial steering A few limitations deserve mention. The fined IGT is more prevalent in females committee), HMP Communications (Editor in definition of type 2 diabetes has changed than in males (9). Most women were re- Chief, Journal of Invasive Cardiology), Journal of over the past decades, notably with the cruited later during the Malmo¨ Preventive the American College of Cardiology (Guest Editor, Associate Editor), Population Health Research Insti- lowering of the fasting glucose threshold Project, when routine performance of the tute (clinical trial steering committee), Slack Publi- in 1997 and the introduction of HbA1c- comprehensive OGTT had been omitted cations (Chief Medical Editor, Cardiology Today’s defined diabetes in 2011 (7). However, owing to financial restrictions. Thus, very Intervention), Society of Cardiovascular Patient despite there being only a partial overlap few women underwent such testing and Care (Secretary/Treasurer), and WebMD (CME steering committees); other, Clinical Cardiology between subjects with prediabetes and the results obtained from this subgroup (Deputy Editor), NCDR-ACTION Registry Steering diabetes defined according to these dif- should be interpreted very cautiously, Committee (Chair), and VA CART Research and ferent measures, the predictive capability given the minuscule number of events. Publications Committee (Chair); research funding, of clinical risk factors remains comparable The isolated IGT group was small, and al- Amarin,Amgen,AstraZeneca,Bristol-Myers 2 Squibb, Chiesi, Eisai, Ethicon, Forest Laboratories, (7,39). The 30 g/m glucose load was the though results were consistent, CIs were fi fi Ironwood, Ischemix, Lilly, Medtronic, P zer, standard procedure at the initiation of wide. Still, this nding by itself is valuable, Roche, Sanofi, and The Medicines Company; roy- the study, and based on estimates of as it indicates a minimal loss by not cap- alties, Elsevier (Editor, Cardiovascular Intervention: body surface area, this resulted in an av- turing these individuals. Finally, although A Companion to Braunwald’s Heart Disease); site erage glucose load 23% lower than with we excluded the individuals who under- co-investigator, Biotronik, Boston Scientific, and the use of 75-g glucose. Prior studies have went lifestyle intervention to enhance the St. Jude Medical (now Abbott); trustee, American College of Cardiology; and unfunded research, suggested that a larger glucose load re- interpretability of our study, this might FlowCo, Merck, PLx Pharma, and Takeda. M.H.O. sults in greater differences in glucose have resulted in removal of the subjects discloses receipt of a part-time clinical research concentrations at 2 h than at 1 h. In indi- who had the highest a priori risk for type 2 grant from the Novo Nordisk Foundation. No viduals without IGT or diabetes, the glu- diabetes and associated complications, other potential conflicts of interest relevant to cose levels at 1 h are virtually identical underestimating the true incidence of this article were reported. Author Contributions. M.P. conceptualized the when comparing a 50-g with a 100-g glu- outcomes, particularly in the subgroups hypothesis, designed the work, analyzed and inter- cose load. Even at 2 h, the difference ap- with abnormal baseline glycemic status. preted data, drafted the manuscript, and revised the pears to be small, with highly correlated In conclusion, the 1-h blood glucose manuscript critically for important intellectual content. values (40–43). Discrepancies between level is a powerful predictor of future D.L.B. and M.H.O. conceptualized the hypothesis, fi designed the work, interpreted data, and revised glucose levels are larger among subjects type 2 diabetes, with net reclassi cation the manuscript critically for important intellectual with IGT at both 1 h and 2 h (42). Theoret- improvement and detection rates signifi- content. M.L.N., R.J., K.-F.E., P.M.N., and M.B. con- ically, the lower glucose dose might have cantly greater than 2-h blood glucose ceptualized the hypothesis, interpreted data, and care.diabetesjournals.org Pareek and Associates 177

revised the manuscript critically for important in- future type 2 diabetes: results from the Botnia but early carotid atherosclerosis. Atherosclerosis tellectual content. All authors read and approved Study. Diabetes Care 2009;32:281–286 2009;207:245–249 the final version of the manuscript and agree to be 14. Priya M, Anjana RM, Chiwanga FS, 30. Sciacqua A, Miceli S, Carullo G, et al. One- accountable for all aspects of the work in ensuring Gokulakrishnan K, Deepa M, Mohan V. 1-hour hour postload plasma glucose levels and left ven- that questions related to the accuracy or integrity of venous plasma glucose and incident prediabetes tricular mass in hypertensive patients. Diabetes any part of the article are appropriately investigated and diabetes in Asian indians. Diabetes Technol Care 2011;34:1406–1411 and resolved. M.P. and M.H.O. are the guarantors of Ther 2013;15:497–502 31. Sciacqua A, Maio R, Miceli S, et al. Association this work and, as such, had full access to all the data in 15. Fiorentino TV, Marini MA, Andreozzi F, et al. between one-hour post-load plasma glucose lev- the study and take responsibility for the integrity of One-hour postload hyperglycemia is a stronger els and vascular stiffness in essential hyperten- the data and the accuracy of the data analysis. predictor of type 2 diabetes than impaired fasting sion. PLoS One 2012;7:e44470 Prior Presentation. Parts of this study were pre- glucose. J Clin Endocrinol Metab 2015;100:3744– 32. Orencia AJ, Daviglus ML, Dyer AR, Walsh M, sentedinabstractformatthe52ndAnnualMeeting of 3751 Greenland P, Stamler J. One-hour postload plasma the European Association for the Study of Diabetes, 16. Alyass A, Almgren P, Akerlund M, et al. Mod- glucose and risks of fatal coronary heart disease and Munich,Germany,12–16 September 2016. elling of OGTT curve identifies 1 h plasma glucose stroke among nondiabetic men and women: the Chi- level as a strong predictor of incident type 2 di- cago Heart Association Detection Project in Industry abetes: results from two prospective cohorts. Di- (CHA) Study. J Clin Epidemiol 1997;50:1369–1376 References abetologia 2015;58:87–97 33. Meisinger C, Wolke¨ G, Brasche S, Strube G, 1. American Diabetes Association. Economic 17. Nielsen ML, Pareek M, Leosd´ ottir´ M, et al. Heinrich J. Postload plasma glucose and 30-year costs of diabetes in the U.S. in 2012. Diabetes Follow-up duration influences the relative impor- mortality among nondiabetic middle-aged men Care 2013;36:1033–1046 tance of OGTT and optimal timing of glucose mea- from the general population: the ERFORT Study. 2. Cavender MA, Steg PG, Smith SC Jr, et al.; surements for predicting future type 2 diabetes. Ann Epidemiol 2006;16:534–539 REACH Registry Investigators. Impact of diabetes Eur J Endocrinol 2016;174:591–600 34. Strandberg TE, Pienimaki¨ T, Strandberg AY, mellitus on hospitalization for heart failure, car- 18. Bergman M, Chetrit A, Roth J, Jagannathan R, et al. One-hour glucose, mortality, and risk of di- diovascular events, and death: outcomes at Sevick M, Dankner R. One-hour post-load plasma abetes: a 44-year prospective study in men. Arch 4 years from the Reduction of Atherothrombosis glucose level during the OGTT predicts dysglycemia: Intern Med 2011;171:941–943 for Continued Health (REACH) registry. Circulation observations from the 24year follow-up of the Israel 35. Bergman M, Chetrit A, Roth J, Dankner R. One- 2015;132:923–931 Study of Glucose Intolerance, Obesity and Hyperten- hour post-load plasma glucose level during the 3. Tuomilehto J, Lindstrom¨ J, Eriksson JG, et al.; sion. Diabetes Res Clin Pract 2016;120:221–228 OGTT predicts mortality: observations from the Is- Finnish Diabetes Prevention Study Group. Preven- 19. Marini MA, Succurro E, Frontoni S, et al. In- rael Study of Glucose Intolerance, Obesity and Hy- tion of type 2 diabetes mellitus by changes in sulin sensitivity, b-cell function, and incretin effect pertension. Diabet Med 2016;33:1060–1066 lifestyle among subjects with impaired glucose in individuals with elevated 1-hour postload plasma 36. Olson DE, Rhee MK, Herrick K, Ziemer DC, tolerance. N Engl J Med 2001;344:1343–1350 glucose levels. Diabetes Care 2012;35:868–872 Twombly JG, Phillips LS. Screening for diabetes and 4. Knowler WC, Barrett-Connor E, Fowler SE, 20. Bianchi C, Miccoli R, Trombetta M, et al.; pre-diabetes with proposed A1C-based diagnostic et al.; Diabetes Prevention Program Research GENFIEV Investigators. Elevated 1-hour postload criteria. Diabetes Care 2010;33:2184–2189 Group. Reduction in the incidence of type 2 di- plasma glucose levels identify subjects with nor- 37. Barry E, Roberts S, Oke J, Vijayaraghavan S, abetes with lifestyle intervention or metformin. mal glucose tolerance but impaired b-cell func- Normansell R, Greenhalgh T. Efficacy and effec- N Engl J Med 2002;346:393–403 tion, insulin resistance, and worse cardiovascular tiveness of screen and treat policies in prevention 5. Chiasson JL, Josse RG, Gomis R, Hanefeld M, risk profile: the GENFIEV study. J Clin Endocrinol of type 2 diabetes: systematic review and meta- Karasik A, Laakso M; STOP-NIDDM Trial Research Metab 2013;98:2100–2105 analysis of screening tests and interventions. BMJ Group. Acarbose for prevention of type 2 diabetes 21. Eriksson KF, Lindgarde¨ F. Prevention of type 2 2017;356:i6538 mellitus: the STOP-NIDDM randomised trial. Lan- (non-insulin-dependent) diabetes mellitus by diet 38. Maki DG. Review: HbA1c has low accuracy cet 2002;359:2072–2077 and physical exercise. The 6-year Malmo¨ Feasibil- for prediabetes; lifestyle programs and metfor- 6. Saito T, Watanabe M, Nishida J, et al.; ity Study. Diabetologia 1991;34:891–898 min reduce progression to T2DM. Ann Intern Zensharen Study for Prevention of Lifestyle Dis- 22. Eriksson KF, Lindgarde¨ F. No excess 12-year Med 2017;166:JC41 eases Group. Lifestyle modification and prevention mortality in men with impaired glucose tolerance 39. Balkau B, Soulimane S, Lange C, Gautier A, of type 2 diabetes in overweight Japanese with im- who participated in the MalmoPreventiveTrial¨ Tichet J, Vol S; DESIR Study Group. Are the same paired fasting glucose levels: a randomized con- with diet and exercise. Diabetologia 1998;41: clinical risk factors relevant for incident diabetes trolled trial. Arch Intern Med 2011;171:1352–1360 1010–1016 defined by treatment, fasting plasma glucose, and 7. Inzucchi SE. Clinical practice. Diagnosis of di- 23. Berglund G, Nilsson P, Eriksson KF, et al. Long- HbA1c? Diabetes Care 2011;34:957–959 abetes. N Engl J Med 2012;367:542–550 term outcome of the Malmo¨ Preventive Project: 40. Sisk CW, Burnham CE, Stewart J, McDonald 8. American Diabetes Association. Standards of mortality and cardiovascular morbidity. J Intern GW. Comparison of the 50 and 100 gram oral glucose Medical Care in Diabetesd2016.DiabetesCare Med 2000;247:19–29 tolerance test. Diabetes 1970;19:852–862 2016;39(Suppl. 1):S1–S112 24. World Health Organization. Diabetes melli- 41. National Center for Health Statistics. The one- 9. Unwin N, Shaw J, Zimmet P, Alberti KG. Im- tus: report of a WHO study group. World Health hour oral glucose tolerance test. Response of paired glucose tolerance and impaired fasting gly- Organ Tech Rep Ser 1985;727:1–113 middle-aged men to 100-gram and 50-gram doses caemia: the current status on definition and 25. Eriksson KF, Lindgarde¨ F. Impaired glucose tol- of glucose given fasting and 1, 2, and 3 hours after intervention. Diabet Med 2002;19:708–723 erance in a middle-aged male urban population: a meal. Vital Health Stat 1973;2:1–34 10. Abdul-Ghani MA, Williams K, DeFronzo RA, new approach for identifying high-risk cases. Diabe- 42. de Nobel E, van’t Laar A. The size of the load- Stern M. What is the best predictor of future type 2 tologia 1990;33:526–531 ing dose as an important determinant of the re- diabetes? Diabetes Care 2007;30:1544–1548 26. Ludvigsson JF, Andersson E, Ekbom A, et al. Ex- sults of the oral glucose tolerance test: a study in 11. Dankner R, Abdul-Ghani MA, Gerber Y, ternal review and validation of the Swedish National subjects with slightly impaired glucose tolerance. Chetrit A, Wainstein J, Raz I. Predicting the Inpatient Register. BMC Public Health 2011;11:450 Diabetes 1978;27:42–48 20-year diabetes incidence rate. Diabetes Metab 27. Harrell FE Jr, Califf RM, Pryor DB, Lee KL, 43. National Diabetes Data Group. Classification and Res Rev 2007;23:551–558 Rosati RA. Evaluating the yield of medical tests. diagnosis of diabetes mellitus and other categories of 12. Abdul-Ghani MA, Abdul-Ghani T, Ali N, JAMA 1982;247:2543–2546 glucose intolerance. Diabetes 1979;28:1039–1057 Defronzo RA. One-hour plasma glucose concen- 28. Pencina MJ, D’Agostino RBS Sr, Steyerberg 44. Carroll JJ, Smith N, Babson AL. A colorimetric tration and the metabolic syndrome identify sub- EW. Extensions of net reclassification improve- serum glucose determination using hexokinase jects at high risk for future type 2 diabetes. Diabetes ment calculations to measure usefulness of new and glucose-6-phosphate dehydrogenase. Bio- Care 2008;31:1650–1655 biomarkers. Stat Med 2011;30:11–21 chem Med 1970;4:171–180 13. Abdul-Ghani MA, Lyssenko V, Tuomi T, 29. Succurro E, Marini MA, Arturi F, et al. Ele- 45. Sayetta RB, Murphy RS. Summary of current DeFronzo RA, Groop L. Fasting versus postload vated one-hour post-load plasma glucose levels diabetes-related data from the National Center for plasma glucose concentration and the risk for identifies subjects with normal glucose tolerance Health Statistics. Diabetes Care 1979;2:105–119