One-hour glucose is an earlier marker of dysglycemia than two-hour glucose

Abstract

Aims:

The timing of increase in 1-hour PG and its utility as an earlier predictor of both prediabetes (PreDM) and type 2 diabetes (T2D) compared to 2-hour PG (2h-PG) is unknown. To evaluate the timing of crossing of the 1h-PG ≥ 155 mg/dl (8.6 mmol/L) for PreDM and 209 mg/dl (11.6 mmol/L) for T2D and respective current 2h-PG thresholds of 140 mg/dl (7.8 mmol/L) and 200 mg/dl (11.1 mmol/L).

Methods:

Secondary analysis of 201 Southwest Native Americans who were followed longitudinally for 6-10 years and had at least 3 OGTTs.

Results:

We identified a subset of 43 individuals who first developed PreDM by both 1h-PG and 2h-PG criteria during the study. For most (32/43,74%), 1h-PG ≥155 mg/dl was observed before 2h-PG reached 140 mg/dl (median [IQR]: 1.7 [−0.25, 4.59] y; mean±SEM: 5.3±1.9 y). We also identified a subset of 33 individuals who first developed T2D during the study. For most (25/33, 75%), 1h-PG reached 209 mg/dl earlier (median 1.0 [−0.56, 2.02] y; mean±SEM: 1.6±0.8 y) than 2h-PG reached 200 mg/dl, diagnostic of T2D.

Conclusions:

1h-PG ≥ 155 mg/dl is an earlier marker of elevated risk for PreDM and T2D than 2h-PG ≥ 140 mg/dl.

Introduction

Type 2 diabetes (T2D) is a progressive disease that can only be reversed or delayed by early intensive interventions, such as extreme caloric restriction or bariatric surgery (1–3). These treatment strategies can be effective for achieving partial or full remission and are more successful when implemented early in the disease course (3,4). On the other hand, preventing or substantially delaying T2D onset is attainable with moderate lifestyle intervention (diet and exercise) or medication (5,6). Therefore, early detection of individuals at increased risk for future diabetes is a cost-effective approach to reducing T2D incidence (7,8).

The current definitions for prediabetes (PreDM), using ADA criteria, were designed to optimize diabetes risk prediction and primary prevention interventions. However, identifying individuals at an early stage of dysglycemia but who are high risk for T2D progression remains a challenge and alternate OGTT metrics have been proposed (9). The 1h-PG value is promising as an ideal biomarker because its predictive ability is superior or equal to the current 2h-PG level threshold for PreDM and, therefore should be used to detect dysglycemia in high-risk individuals (10–12).

Abdul-Ghani et al. (13) previously suggested a 1h-PG level of ≥155 mg/dl (8.6 mmol/L) as a cut-point for predicting future T2D. Multiple subsequent studies have shown that individuals classified as having normal glucose tolerance (NGT) by current standards with 1h-PG ≥155 mg/dl (8.6 mmol/L) are at greater risk for developing T2D than those with NGT having a 1h-PG < 155mg/dl (8.6 mmol/L) (11,14). Some studies have shown that 1h-PG ≥155 mg/dl (8.6 mmol/L) is also a more sensitive predictor for the development of micro- and macrovascular complications and mortality than the 2h-PG (15–18). An elevated 1h-PG is also associated with an increased risk for hepatic fibrosis (19). Furthermore, the combination of an elevated 1h-PG and IGT confers an even greater risk for progression to T2D, complications, and mortality (20). Based on these studies, a petition was published recommending that a 1h-PG ≥ 155 mg/dl (8.6 mmol/L) be considered as an alternate criterion for PreDM (10).

However, previous studies have only considered the 1h-PG as a baseline predictor for the development of T2D and have not evaluated whether 1h-PG could be an earlier diabetes risk marker compared to 2h-PG. As intermediate glucose levels may be elevated during the OGTT when fasting plasma glucose (FPG) and 2h-PG levels are normal, the present study investigated the relative times when the 1h-PG and 2h-PG crossed their respective thresholds for PreDM and T2D. These observations are supported by simulations (Fig. S1) with a longitudinal mathematical model (21,22), indicating that the 1h-PG crosses the proposed threshold of 155 mg/dl (8.6 mmol/L) before the 2h-PG crosses the threshold of 140 mg/dl (7.8 mmol/L) defining IGT. Similarly, the model suggested that the 1h-PG may cross the proposed threshold of 209 mg/dl (11.6 mmol/L) to diagnose T2D before the 2h-PG crosses the threshold of 200 mg/dl (11.1 mmol/L) diagnostic of T2D by classic criteria (12). To test the hypotheses that the 1h-PG is an earlier marker of PreDM and T2D than the 2h-PG, we analyzed a longitudinal dataset of OGTTs collected during studies of Southwestern Native American (SWNA) over several decades (23,24). To estimate individual glucose trajectories for diabetes risk prediction over time, we used the linear mixed effects model to: (1) determine the order of elevation of 1h-PG vs. 2h-PG for predicting PreDM; and (2) evaluate the timing of elevation of 1h-PG vs. 2h-PG for prediction of PreDM and T2DM.

Subjects, Materials and Methods

Study design

A subset of participants in a longitudinal epidemiological study conducted in a SWNA community in Arizona, USA, were included in the present study (24). In that long-running longitudinal study, which began in 1965 and ended in 2007, individuals aged 5 years and older were invited for outpatient research examinations approximately every 2 years. These included an OGTT, consisting of post-load venous plasma glucose measurements at 1- and 2-h (24).

As previously described, adult community members without T2D were invited to participate in an inpatient metabolic study to assess the determinants of T2D (23). These participants were admitted to the Clinical Research Unit in Phoenix, AZ, USA, and found to be otherwise healthy based on a complete medical history and physical examination, including routine laboratory tests, and were not taking medications one month prior to each measurement known to affect glucose or insulin metabolism. After admission, volunteers were fed a weight-maintaining diet (energy distribution: 50% carbohydrates, 30% fat, 20% protein) and abstained from strenuous activity. After at least 3 days on the weight-maintaining diet, a series of tests were conducted including an OGTT, which routinely involved 1h-PG and evaluation of BMI.

We identified 201 individuals who had at least 3 OGTTs (see Table 1). For analysis of the onset of PreDM, we identified a subset of 43 individuals with a median of 8 OGTTs (range 3 – 12) performed 1.8±1.0 years apart during 8.7±3.5 years of follow-up, who had baseline FPG < 100 mg/dl (5.6 mmol/L), 1h-PG < 155 mg/dl (8.6 mmol/L), and 2h-PG < 140 mg/dl (7.8 mmol/L) and eventually crossed these thresholds (Data Set 1, Table 1; flow diagram, Fig. S2). Participants were young adults at baseline (age, 24.0±5.5 years) with obesity (BMI, 34.4±6.9 kg/m²). For the development of PreDM, we estimated the time points when 1h-PG and 2h-PG levels crossed proposed and established thresholds,155 mg/dl (8.6 mmol/L) and 140 mg/dl (7.8 mmol/L), respectively, by fitting a linear mixed effect model to trajectories of individual participants.

Table 1.

Baseline Characteristics of the data sets

Characteristics	All N=201	Data Set 1 N=43	Data Set 2 N=33	Data Set 3 N=120	Data Set 4 N=30
Age	25.7 (5.7)	24.0 (5.5)*	26.6 (6.1)	24.5 (5.5)*	24.8 (5.7)
Sex Male	117(57%)	28 (65%)	9 (27.3%)*	83 (69.2%)*	20 (66.7%)
BMI, kg/m2	35.1 (7.7)	34.4 (6.9)	36.6 (6.4)	33.6 (7.4)*	36.1 (5.5)
OGTT measures
FPG (mg/dl)	93.8 (16.2)	88.0 (8.1)*	91.2 (11.1)	87.1 (7.6)*	90 (5.7)*
1h-PG (mg/dl)	156.0 (45.1)	127.7 (17.9)*	165 (26.4)	135 (25.2)*	129 (17.5)*
2h-PG (mg/dl)	132.4 (46.8)	112.5 (17.3)*	138.2 (28.1)	110 (18.3)*	114 (14.7)*
FPI (μU/ml)	42.4 (22.6)	35.9 (16.4)*	47.4 (18.5)	33.8 (16.5)*	40.1 (15.6)
1h-PI (μU/ml)	255.5 (176.6)	200.0 (116.2)*	314.7 (213.9)	211.4* (140.4)	187.6* (76.1)
2h-PI (μU/ml)	219.9 (188.2)	157.1 (101.7)*	286.1 (227.7)	148.7* (113.3)	153.4* (76.2)
HOMA-IR	10.1 (6.6)	7.8 (3.8)*	10.7 (4.3)	7.3 (3.7)*	8.9 (3.6)
HOMA-β	572.5 (531.2)	518.5 (627.2)	651.1 (864.2)	561.3 (559.1)	556.8 (219.4)
Matsuda Index	1.4 (0.8)	1.7 (0.9)*	1.1 (0.5)*	1.7 (0.9)*	1.4 (0.5)
IGI	4.9 (11.2)	3.9 (6.9)	4.4 (2.7)	4.2 (4.9)	4.6 (2.5)
oDI	5.2 (17.0)	4.2 (30.1)	4.1 (2.5)	6.0 (19.1)	6.4 (4.7)
Follow-up
Duration (years)	7.2 (3.5)	8.7 (3.5)*	7.4 (2.9)	8.0 (3.5)*	8.1 (3.6)
Number of OGTTs	6.8 (2.6)	6.5 (2.5)*	6.1 (2.4)*	5.6 (2.4)*	6.7 (2.5)*
OGTT interval (years)	1.9 (1.0)	1.8 (1.0)	1.9 (1.0)	1.9 (1.0)*	1.9 (1.0)*

Variables are shown as number(%) or mean(SD).

^*indicates statistically significant difference vs. the complement in “All”.

BMI, body mass index; HOMA-IR, homeostatic Model Assessment for Insulin Resistance; HOMA-β, homeostatic Model Assessment for beta-cell function; IGI, Insulinogenic Index, FPG (fasting plasma glucose), 1h-PG (one-hour plasma glucose), 2h-PG (two-hour plasma glucose), FPI (fasting plasma insulin), 1h-PI (one-hour plasma insulin), 2h-PI (two-hour plasma insulin).

Data Set 1: NGT (FPG<100, 1h-PG<155, 2h-PG<140 mg/dl) at baseline and ever progressed to PreDM (1h-PG ≥155, 2h-PG ≥140 mg/dl) during the study. Threshold crossing for this and the other data sets did not have to occur at the same visit.

Data Set 2: No diabetes at baseline (FPG < 126, 1h-PG < 209, 2h-PG < 200 mg/dl) and ever progressed to T2D (1h-PG ≥ 209 and 2h-PG ≥ 200) during the study.

Data Set 3: NGT (FPG < 100, 2h-PG < 140 mg/dl) at baseline.

Data Set 4: NGT (FPG < 100, 1h-PG < 155, 2h-PG < 140 mg/dl) at baseline and ever exceeded both FPG = 100 mg/dl and 1h-PG =155 mg/dl during the study.

To analyze the onset of T2D, we identified a different subset of 33 individuals without diabetes at baseline (FPG < 126 mg/dl (7 mmol/L), 1h-PG <209 mg/dl (11.6 mmol/L), and 2h-PG < 200 mg/dl (11.1 mmol/L) but developed diabetes during the study (Data Set 2, Table 1; flow diagram, Fig. S3). This subset had at least three OGTTs (median 7, range 3 – 11) during 7.4±2.9 years of follow-up). We used the fitted mixed-effect model linear trajectories to estimate the time points when their 1h-PG and 2h-PG levels crossed the proposed and established thresholds, 209 mg/dl (11.6 mmol/L) and 200 mg/dl (11.1 mmol/L), respectively.

For analysis of the probability of progression to T2D from various starting conditions, we identified a subset of 120 participants (Data Set 3) with NGT (FPG < 100 mg/dl [5.6 mmol/L], 2h-PG < 140 mg/dl [7.8 mmol/L]) at baseline (Table 1; flow diagram, Fig. S4).

For analysis of the order of crossing FPG = 100 mg/dl (5.6 mmol/L) and 1h-PG =155 mg/dl (8.6 mmol/L), we identified a subset of 30 participants with FPG < 100 mg/dl (5.6 mmol/L), 1h-PG < 155 mg/dl (8.6 mmol/L), and 2h-PG < 140 mg/dl (7.8 mmol/L) at baseline (Data Set 4,Table 1; flow diagram, Fig. S5)

Statistics

Statistics were calculated using R, ver. 4.2.2 (25). Because individual trajectories exhibited marked fluctuations, we used a linear mixed effect model to smooth the data, correlate the 1h-PG with 2h-PG, and determine when these measures crossed their respective thresholds. Linear mixed-effect (LME) model fits were done using the R function lmer in the lme4 library. R² for the fits was evaluated using the MuMIn package (26), and P-values were determined using the R function anova to compare models with and without the fixed effects. The statistical model for LME in Wilkinson notation (27) was $y\sim x+\left(x|subject\right)$, where y is either 1h-PG or 2h-PG, and x is either 2h-PG or age. In other words, the fixed effects were slope and intercept, and each individual had their own random intercept and slope. A P-value < 0.05 was considered statistically significant.

Analytical Calculations

The insulinogenic index (IGI) was calculated at the ratio of the incremental change of insulin and glucose from 0 to 30 minutes of the OGTT (28–30).

The Matsuda insulin sensitivity index (ISI) was calculated as (31)

$$\frac{10000}{\sqrt{\overline{I}\overline{G}{I}_o{G}_0}}$$

where $I_0$ and $G_0$ are the fasting insulin (μU/mL) and glucose (mg/dl) and $\overline{I}$ and $\overline{G}$ are the average insulin and glucose over the OGTT calculated as AUC/120 min by the trapezoidal rule.

Using the same definitions and units, HOMA IR (homeostatic model assessment – insulin resistance), was calculated as (32):

$$\frac{I_o{G}_0}{405}$$

and HOMA Beta (homeostatic model assessment – beta-cell function) was calculated as (32):

$$\frac{360I_0}{G_0-63}$$

Ethics Statement

Written informed consent was obtained for all participants, and this study was approved by the institutional review board of the National Institute of Diabetes and Digestive and Kidney Diseases (NCT00340132).

Supplementary Material

The supplementary figures and legends are available at https://doi.org/10.6084/m9.figshare.22647535.

Results

Order of Elevated 1h-PG and 2h-PG in Detection of PreDM

First, we established that 1h-PG and 2h-PG were highly correlated by carrying out linear mixed effect (LME) fitting of 2h-PG vs. 1h-PG (Fig. S6) for the 43 participants whose 1h-PG crossed 155 mg/dl (8.6 mmol/L) and 2h-PG crossed 140 mg/dl (7.8 mmol/L) at some point during the study; in this and subsequent analyses, threshold crossings were not required to be concurrent (i.e., did not have to occur at the same visit). In this and subsequent LME figures, the solid red lines represent the overall population fit, which is the same for all individuals, and the dashed red lines represent the mixed-model fit for each individual. The dashed lines are distinct from the standard linear regression of each individual and represent a compromise between individual and group tendencies. Although not every individual showed a clear linear trend, the overall fit was very good (R = 0.88, P < 0.0001).

Figure 1 shows four individuals selected to illustrate more clearly the relationships between population and individual fits and how the dashed lines were used to determine whether 1h-PG or 2h-PG crossed their respective thresholds first. Although time is not explicitly represented in these plots, we can compare the intercepts of the fitted lines with the black vertical line [1h-PG threshold of 155 mg/dl (8.6 mmol/L)] and the black horizontal line [2h-PG threshold of 140 mg/dl (7.8 mmol/L)]. For the population line (solid red), the y-intercept is < 140 mg/dl (7.8 mmol/L) and the x-intercept is > 155mg/dl (8.6 mmol/L). If we assume, as occurred in the model simulations in Fig. S1, that the overall trend of glucose increases from lower left to upper right (this is not strictly true for each individual as discussed below), this implies that the 1h-PG threshold is crossed first for each of the four selected individuals.

Figure 1:

Order of prediabetes threshold crossings. To illustrate the method, four individuals were selected from the subset of 43 who had NGT at baseline and crossed the proposed 1h-PG threshold (155 mg/dl, 8.6 mmol/L; vertical black line) and the ADA 2h-PG threshold (200 mg/dl, 7.8 mmol/L; horizontal black line) for prediabetes. The 1h-PG and 2h-PG readings are plotted against each other (black dots) and fit with a linear mixed effect model (see Methods). In each panel, the solid red line represents the population fit, the dotted red line the individual mixed-model fit. The crossings of the thresholds by the fitted lines suggest that all the individuals shown in (A) – (C) crossed the vertical line first, but the individual shown in (D) crossed both lines at about the same time. For the full subset, see Fig. S6. Baseline characteristics are listed in Table 1, Data Set 1.

Each individual’s mixed-model fit (dashed red lines) has its own x- and y-intercepts. In Fig. 1A, B, the y-intercept < 140 and the x-intercept > 155, but in panel C, the y-intercept > 140 and the x-intercept < 155, and in panel D, the x- and y-intercepts approximate 140 and 155, respectively. The x- and y-intercepts for all 43 individuals whose trajectories are shown in Fig. S6 are shown in Fig. 2A. The points in the lower-right (4^th) quadrant of Fig. 2A represent those (34 of 43; 79%) who crossed the 1h-PG threshold first, while the points in the upper-left (2^nd) quadrant represent those who crossed the 2h-PG threshold first (9 of 43, 21%). The data are summarized in the box plot in Fig. 2B, demonstrating that the mean fitted 2h-PG value was 133.0±1.4 (mean ± SEM) when the fitted 1h-PG value was 155 mg/dl (8.6 mmol/L) (P < 0.0001). Thus, the average fitted 2h-PG value was in the NGT range (2h-PG < 140 mg/dl [7.8 mmol/L]) when the average fitted 1h-PG value reached its proposed threshold of 155 mg/dl (8.6 mmol/L) (P < 0.001). Conversely, when the fitted 2h-PG value was 140 mg/dl (7.8 mmol/L), the mean 1h-PG value was 172.8 mg/dl (9.3 mmol/L), i.e., 1h-PG was well above its proposed threshold for prediabetes when 2h-PG reached its threshold.

Figure 2:

Intercepts of the 1h-PG and 2h-PG prediabetes threshold lines. The intercepts of the Data Set 1 subset illustrated in Figs. 1 and S6 and plotted in panel A suggest that 34 crossed the vertical (1h-PG) threshold first and 9 crossed the horizontal (2h-PG) threshold for prediabetes. (B) When 1h-PG was 155 mg/dl (8.6 mmol/L), mean 2h-PG was 133.0±1.4 mg/dl (7.4±0.08 mmol/L).

Timing of 1h-PG and 2h-PG in Detection of PreDM

While the results in Figs. 1 and 2 suggest that the 1h-PG threshold is crossed first in most cases (79%), they do not indicate how much earlier this occurs because time is not taken into account. In addition, the interpretation of earlier crossing depends on the assumption that glucose increases monotonically for each individual from lower left to upper right. Figures S7 and S8, which plot 1h-PG and 2h-PG, respectively, vs. the age of each individual, show that this is not strictly true. The measured values fluctuate, some decreasing and some increasing with time, while other individuals cross the thresholds several times. Indeed, it was these fluctuations that led us to use LME fitting to smooth the data.

To overcome these limitations, we again used LME fitting, this time for 1h-PG or 2h-PG vs. age, to estimate the times at which individuals crossed each threshold. Figure 3 shows the LME fits of 1h-PG vs. age (left column) and 2h-PG vs. age (right column) for the four individuals in Fig. 1. Each row in Fig. 3 shows the pair of trajectories for one individual as they progress from NGT to IGT, confirming that 1h-PG and 2h-PG show a parallel increase in time. Moreover, individuals for whom the slope of the fitted 1h-PG line in the left column is greater had larger slopes of the 2h-PG fitting in the right column.

Figure 3:

Regression of age 1h-PG and 2h-PG vs. age by linear mixed effect modeling. Solid red lines represent the population fit, and the dotted red lines represent the individual mixed-model fit in greater detail for four of the individiduals in Figs. S7, S8; they are the same individuals as in Fig. 2. Solid black lines represent the 1h-PG (left column) and 2h-PG (right column) prediabetes thresholds. are shown in greater detail.

The LME fits for each individual (dashed lines in Fig. 3) allow us to estimate the ages when 1h-PG and 2h-PG cross their respective thresholds. These times are plotted in Fig. 4. The red line has slope 1 and intercept 0, representing the case when 1h-PG and 2h-PG cross their thresholds simultaneously. Most individuals (32/43, 74%) lie above the red line, indicating that the 1h-PG crossed 155 mg/dl (8.6 mmol/L) before 2h-PG crossed 140 mg/dl (7.8 mmol/L); this is close to the estimate of 34/43 from a different method in Fig. 2. The median time difference was 1.7 years (interquartile range: [−0.25, 4.59] years; mean±SEM: 5.3±1.9 years). By paired t-test and Wilcox test, the time difference was highly statistically significant (P = 0.007 and P=0.0001, respectively).

Figure 4:

Timing of prediabetes threshold crossings. The estimated ages are shown at which each individual in Fig. S6 – S8 crossed the 1h-PG threshold of 155 mg/dl (8.6 mmol/L) and the 2h-PG threshold of 140 mg/dl (7.8 mmol/L). 32 of the 43 (74%) crossed the 1h-PG threshold first. The 1h-PG threshold was crossed earlier (median, 1.7y; mean, 5.3y).

Timing of 1h-PG and 2h-PG in Diagnosis of T2D

We next applied the same method to Data Set 2 (Table 1) to calculate the ages at which 1h-PG crossed the proposed threshold for T2D, 209 mg/dl (11.6 mmol/L) (12) and 2h-PG crossed the standard threshold of 200 mg/dl (11.1 mmol/L) in a subset of 33 individuals from the full longitudinal study who did not have diabetes at baseline but crossed 1h-PG = 209 mg/dl (11.6 mmol/L) and 2h-PG = 200 mg/dl (11.1 mmol/L) at some point during the study. We used the 1h-PG and 2h-PG fits vs. age (not shown) to determine at what ages 1h-PG crossed 209 mg/dl and 2h-PG crossed 200 mg/dl for each individual and plotted the ages in Fig. 5A. In most cases (25/33, 75%), the 1h-PG threshold was crossed 1 year earlier (median; interquartile range: [−0.56, 2.02] years) than the 2h-PG threshold (mean±SEM: 1.6±0.8 years). By paired t-test and Wilcox test the time difference was statistically significant (P = 0.04 and 0.02, respectively). Expressed another way, when 2h-PG was 200 mg/dl (11.1 mmol/L), mean 1h-PG was already 217.9 mg/dl (12.1 mmol/L; P < 0.001), above its proposed threshold of 209 mg/dl (11.6 mmol/L) (Fig. 5B).

Figure 5:

Timing of diabetes threshold crossings. The estimated ages at which each individual in Figs. S7 and S8 (Data Set 2, Table 1) crossed the 1h-PG threshold of 209 mg/dl (11.6 mmol/L) and the 2h-PG threshold of 200 mg/dl (11.0 mmol/L). 27 of the 38 crossed the 1h-PG threshold first. The 1h-PG threshold was crossed earlier (median, 1.0y, mean 1.6y).

Crossing of 1h-PG = 155 mg/dl (8.6 mmol/L) predicts PreDM and T2D

Next, we asked whether crossing 155 mg/dl (8.6 mmol/L) by 1h-PG predicts PreDM (crossing 140 mg/dl [7.8 mmol/L] by 2h-PG). We analyzed a subset of 120 participants from the full longitudinal study (Data Set 3, Table 1) who had NGT at baseline (FPG < 100 mg/dl [5.6 mmol/L] and 2h-PG <140 mg/dl [7.8 mmol/L]). Among that subset, 48 of the 97 (49%) who had 1h-PG < 155 at baseline (8.6 ± 3.6 years of follow-up), progressed to IGT compared to 18 of the 23 (78%) who had 1h-PG >155 at baseline (8.6 ± 3.6 years of follow-up). The difference in proportions was statistically significant (p = 0.02).

We asked the same question about progression to T2D. Of the 97 who had 1h-PG < 155 at baseline, 9 (9.3%) progressed to T2D during the study, compared to 7/23 (30.4%) of those who had 1h-PG ≥ 155 at baseline. The proportion of progressors was significantly higher for those who had 1h-PG ≥ 155 at baseline (P=0.02).

1h-PG crosses 155 mg/dl (8.6 mmol/L) before FPG crosses 100 mg/dl (5.6 mmol/L)

We also compared the timing of FPG and 1h-PG. There were 30 individuals (Data Set 4, Table 1) who had FPG < 100 mg/dl (5.6 mmol/L), 2h-PG < 140 mg/dl (7.8 mmol/L), and 1h-PG < 155 mg/dl (8.6 mmol/L) at baseline and crossed FPG=100 (5.6 mmol/L) and 1h-PG=155 (8.6 mmol/L) at any point during the study (8.1 ± 3.7 years of follow-up). In all 30 cases, the 1h-PG threshold was crossed before the FPG reached the impaired fasting glucose (IFG) threshold of 100 mg/dl (5.6 mmol/L).

Beta-cell function is already impaired and insulin sensitivity is reduced when 1h-PG crosses 155 mg/dl (8.6 mmol/L)

In further support of the benefit of intervening therapeutically when 1h-PG crosses 155 mg/dl (8.6 mmol/L), we used the entire data set of 201 participants (“All” in Table 1) to compare the insulinogenic index (IGI) (33), the Matsuda ISI index for insulin sensitivity (31), and the product of the two (oral disposition index, oDI) (34) for participants who had NGT by ADA standards but 1h-PG < 155 (referred to as low 1h-PG in the following), NGT but 1h-PG ≥ 155 mg/dl (high 1h-PG), IGT or T2D. As shown in Fig. S9A, IGI was reduced in high 1h-PG (3.4±2.3) compared to low 1h-PG (5.3±4.6), was not reduced further in IGT (4.2±7.8), but declined again in T2D (1.5±1.2) (variables logged because of skewing). Figure S9B shows that ISI decreased progressively through all four glucose tolerance states (2.2±1.0, 1.7±0.7, 1.2±1.6, 0.8±0.3), and Fig. S9C shows that oDI follows the pattern of ISI (10.6±12.3, 4.6±2.0, 4.4±6.8, 1.1±0.8).

Discussion

In this longitudinal cohort study of Southwest Native Americans who had multiple OGTTs over many years, we employed a novel method of following individual trajectories smoothed by LME model regression to demonstrate that 1h-PG crosses the 155 mg/dl (8.6 mmol/L) threshold earlier than the 2h-PG crosses the threshold of 140 mg/dl (7.8 mmol/L), the standard ADA criterion for PreDM (impaired glucose tolerance or IGT). The median additional lead time of 1.7 years (interquartile range [−0.25, 4.59] years; mean±SEM: 5.3±1.9 years) is potentially clinically significant as this would permit earlier lifestyle intervention to prevent progression to T2D. In addition, we estimated that the 1h-PG crosses the proposed threshold for T2D of 209 mg/dl (11.6 mmol/L) 1 year earlier (median; interquartile range [−0.56, 2.02] years; mean±SEM: 1.6±0.8 years) than the 2h-PG crosses 200 mg/dl (11.1 mmol/L), the current ADA threshold for the diagnosis of T2D. Timing would vary if alternate thresholds, e.g., population specific thresholds, were used. However, we have followed longstanding ADA practice of applying a common set of thresholds to all populations (35).

Our novel findings complement multiple previous studies that have shown that people with 1h-PG ≥ 155 mg/dl (8.6 mmol/L) are at increased risk for T2D, and that the 1h-PG is a more sensitive predictor than the standard 2h-PG criterion. Further, people with 1h-PG ≥155 mg/dl (8.6 mmol/L) have a worse cardiovascular disease profile, increased mortality and risk for more severe hepatic fibrosis compared with 1h-PG < 155 mg/dl (8.6 mmol/L) (19).

Our results are also largely consistent with another longitudinal study that found a subset of individuals who progressed from NGT with 1h-PG ≥ 155 mg/dl (8.6 mmol/L) at baseline to IGT, implying that high 1h-PG occurred first (36). The risk of progressing to IGT was higher in those with high 1h-PG than those with low 1h-PG. That study also found elevated risk of progressing from NGT, high 1h-PG to IFG. The present study confirms the latter findings as those with NGT, high 1h-PG invariably preceded IFG.

We found that most individuals with 1h-PG ≥155 mg/dl (8.6 mmol/L) at some time also had 2h-PG ≥140 mg/dl (7.8 mmol/L) at some time (135/167, 81%). Among those for whom we could establish the time sequence, most (32/43; 74%) crossed the 1h-PG threshold first (Fig. 4). Thus, screening with the 1h-PG would include those who would be referred for preventive intervention using current criteria but earlier. We found that insulin sensitivity and beta-cell function were already impaired in the NGT, high 1h-PG state (Fig. S9), so there is no clinical justification for delaying earlier identification and intervention. A previous cross-sectional study suggested that earlier intervention may in fact be more effective, as beta-cell function relative to insulin resistance (the disposition index), though reduced, was found to be higher in the high 1h-PG group than in the IGT group (37). We also found that no one would be missed if only the 1h-PG were used for screening in place of the 2h-PG. Thus, sensitivity would be enhanced without sacrificing specificity, providing strong justification for using a shorter 1-h OGTT. Furthermore, identifying high-risk individuals before crossing the threshold for IGT is also clinically significant aside from preventing progression toT2D as earlier intervention, when the 1h-PG is ≥ 155 mg/dl (8.6 mmol/L), could reduce the risk for the development of micro- and macrovascular complications as well as mortality (14,15). An elevated 1-h PG has also recently been found to be associated with severity of hepatic fibrosis risk (19). Finally, it should be noted that individuals having both an elevated 1h-PG and IGT are at an even greater risk for complications and mortality, providing an additional important rationale for early screening (38).

Strengths:

The current study utilized a novel LME approach to estimate glucose trajectories in high risk, well-phenotyped individuals who had multiple OGTTs over a long duration. This method is is complementary to the static predictive ROC analyses employed in previous studies evaluating 1h-PG. The current study tracked the trajectories of individuals, made possible by the availability of multiple longitudinal OGTTs for each individual. In addition, individuals were followed sufficently long to be informative as 55% (66/120) of those who were not initially diagnosed with IGT progressed to IGT and more than 13% (16/120 to T2D.

Limitations:

The study was limited to a cohort of Southwest Native Americans and further studies are required to determine if the results are generalizable to other populations. Because this study was initiated almost 40 years ago, hemoglobin A_1c was not available for enough of the participants to compare with 1h-PG. The study was not designed to assess differences in biological sex, an important question that should be addressed in future studies.

Conclusions:

We demonstrated in this longitudinal cohort study of Southwest Native Americans that the proposed 1h-PG threshold of 155 mg/dl (8.6 mmol/L) for defining prediabetes was crossed 1.6 years earlier (median; mean, 5.3 years) than crossing the 2h-PG threshold of 140 mg/dl (7.8 mmol/L), the current diagnostic threshold for IGT. These findings posit that applying this threshold could result in earlier detection and intervention in individuals at high risk for PreDM and T2D. We conjecture that in other populations as well, a high 1h-PG is an intermediate state between NGT and IGT during the lengthy trajectory to T2D, which has recently been confirmed in a European population (36). Specific data on relative timing are needed for other populations to inform clinical decision making. Finally, we found that the proposed 1h-PG cut-point of 209 mg/dl (11.6 mmol/L) for diagnosing T2D was crossed 1 year earlier (median; mean 1.6 years) than the standard 2h-PG threshold of 200 mg/dl (11.1 mmol/L), which may also facilitate earlier initiation of therapy when reversal is more likely to be achieved.