Day-to-Day Variations in Fasting Plasma Glucose Do Not Influence Gastric Emptying in Subjects With Type 1 Diabetes

Disorders of gastric emptying can occur in patients with diabetes as the result of autonomous neuropathy affecting the gastrointestinal tract (1–6). A moderate delay in gastric emptying has been described to be common in patients with type 2 diabetes (5,7). More extreme forms associated with clinical symptoms usually are the consequence of long-standing, insufficiently controlled type 1 diabetes, most often associated with other severe diabetic complications (e.g., blindness and end-stage renal disease). This phenomenon is called gastroparesis (1) and, because of early satiety, abdominal fullness, nausea, vomiting, and other symptoms, may lead to reduced calorie intake, causing difficulties in maintaining a healthy body weight (3,8–10). Less symptomatic deceleration of gastric emptying may nevertheless require clinical attention, because it still slows the entry of nutrient-derived carbohydrates, thus attenuating the postprandial rise of the blood glucose level (4,11,12). In patients with slowed gastric emptying, the kinetics of insulin replacement may need to be different from that in patients with normal gastric emptying (13). For these reasons, there is an interest in robust, noninvasive measurement of gastric emptying. Besides scintigraphy, which is considered to be the “gold standard” (14–16), breath tests have been used frequently. Breath tests profit from the rapid absorption, metabolism, and pulmonary excretion of short-chain fatty acids following gastric emptying (17–21). The stable isotopes used in the acetate (liquid meal) or octanoate (solid meal) breath tests avoid exposure to radioactivity. Standardized mixed-meal ¹³C-CO₂ octanoate breath tests have been validated against scintigraphy (17–20). ¹³C-CO₂ octanoate breath tests analyzed according to the Wagner-Nelson method, which accounts for the velocity of elimination of ¹³C-CO₂, deliver results numerically similar to scintigraphy (22).

Acute variations in blood glucose levels by hyperglycemic clamp experiments have shown significant deceleration of gastric emptying of liquid and solid meals in healthy subjects and in patients with type 1 and type 2 diabetes, even at moderate hyperglycemia like 8 mmol/L (144 mg/dL) (23–27). Conversely, acute hypoglycemia accelerated gastric emptying at glucose levels of 4 mmol/L (72 mg/dL) (28–30), even in long-standing type 1 diabetes (31). While this mechanism could help to maintain glucose homeostasis, the intrinsic variations in blood glucose levels typically found in type 1 diabetes may consequently induce large variations in gastric emptying patterns in this population (32,33). Assuming such a marked impact of glycemic fluctuations on gastric emptying, glucose control would be difficult to achieve, because the (variable) time course of carbohydrate entry into the circulation would not always match insulin profiles with a constant regimen of insulin replacement. Also, assessing the velocity of gastric emptying (for example, to diagnose gastroparesis) may be difficult if a single measurement were subject to erratic changes, due to variations in blood glucose levels or other reasons.

Thus, the purpose of the present studies was to analyze gastric emptying (standardized, mixed-meal, ¹³C-CO₂ octanoate breath tests, and Wagner-Nelson analysis [22,34]) in relation to spontaneous variations in fasting blood glucose. To address this, a cohort of patients with type 1 diabetes and matched healthy control subjects were studied prospectively to relate gastric emptying to fasting plasma glucose (FPG) measured on three occasions and grouped into low, intermediate, and high FPG categories. In addition, a retrospective analysis of patients with type 1 diabetes undergoing gastric emptying tests for clinical reasons, grouped into tertiles of FPG, was performed.

The protocol of the prospective observational study was approved by the ethics committee of the medical faculty of the Georg-August University of Göttingen (Göttingen, Germany) prior to the study on 11 September 2009 (registration number 35/7/09). Written informed consent was obtained from all participants in the prospective study cohort. For the retrospective cohort, written consent to scientific analysis of their clinical data was obtained as part of the treatment contract signed with the Diabeteszentrum Bad Lauterberg. Data obtained by chart review were processed in a pseudonymized manner, so that no information could be traced to any individual patient, in accordance with local data protection rules.

Twenty patients with type 1 diabetes and 10 sex-, age-, and weight-matched healthy control subjects were studied. Their anthropometric and clinical characteristics are presented in Table 1. Inclusion criteria were type 1 diabetes on intensified insulin regimens or insulin pump treatment, age between 18 and 65 years, BMI between 20 and 30 kg/m² (both included), and diabetes duration ≥2 years. Exclusion criteria were any concomitant treatments likely to affect either glycemic control (e.g., systemic glucocorticoids) or gastric emptying (e.g., anticholinergic drugs), uncontrolled arterial hypertension (≥160/95 mmHg), liver or kidney disease with an associated functional impairment (e.g., estimated glomerular filtration rate <60 mL/min per 1.73 m² body surface, calculated according to the Chronic Kidney Disease Epidemiology Collaboration equation), current alcohol or substance abuse, participation in any clinical trial during the previous 3 months, and any other severe illnesses or reason precluding successful completion of the study protocol.

Data from all gastric emptying examinations using a standardized protocol for a mixed-meal ¹³C-CO₂ octanoate breath test performed between January 2005 and August 2008 were retrieved from primary source documents at the Diabeteszentrum Bad Lauterberg. Only data from patients with type 1 diabetes were used. Patients with type 2 diabetes, chronic pancreatitis, and other forms of diabetes were excluded. In patients undergoing repeated testing, only the most recent results were used. A total of 258 gastric emptying tests were retrieved. One patient had to be excluded from the analysis because of a missing FPG, and three patients were excluded because of conventional insulin (premixed insulin) treatment, leaving 255 patients for the analysis. Characteristics including clinical indications for the gastric emptying study are summarized in Table 1.

In the prospective study cohort, standardized, mixed-meal, ¹³C-CO₂ octanoate breath tests were performed three times during an inpatient treatment aimed at improving glycemic control. At least 1 day without any study-related activity was required between any two studies of gastric emptying. Gastric emptying data were analyzed comparing three conditions, based on the FPG concentration measured before initiating the gastric emptying study. The lowest value determined within the same subject defined the category FPG “low.” Similarly, the intermediate and highest value led to the classification as FPG “intermediate” and “high,” respectively. Variations in FPG concentrations were not the result of specific protocol-mandated measures for inducing such fluctuations. Specifically, changes to basal insulin dosage were not allowed for the period when gastric emptying tests were performed.

In the retrospective, cross-sectional analysis cohort, gastric emptying data were grouped into tertiles based on low, intermediate, and high FPG concentrations measured on the day when gastric emptying tests were performed. Parameters characterizing gastric emptying were compared among these three groups to detect a potential influence of spontaneous variations in ambient FPG concentrations. As a secondary analysis, we also related gastric emptying to current HbA_1c as a marker of chronic glycemia.

All ¹³C-CO₂ octanoate breath tests were performed in the morning after an overnight fast. A mixed test meal containing ¹³C-octanoic acid as a nonradioactive marker to allow the measurement of gastric emptying by analyzing the appearance of ¹³C-CO₂ in exhaled air was served. In subjects with type 1 diabetes, the short-acting insulin was injected 10–30 min prior to the meal, depending on the respective type of insulin or insulin analog used as empirically determined for each subject. The recommended individual insulin dose was calculated for a meal containing 25 g carbohydrate, based on the prior experience of prevalent hypoglycemic episodes 3–5 h after meal ingestion, if insulin was provided for the full carbohydrate content of the standardized meal (see below). Meal-related insulin doses were corrected for current plasma glucose concentrations determined by hand-held glucometers using algorithms empirically optimized for each patient.

The mixed meal contained one scrambled egg, a slice of ham, 10 g of butter, two slices of toast, 20 g strawberry jam, and 200 mL of unsweetened tea and represents a typical European breakfast. The total caloric load of this test meal was 541 kcal (53.5% as carbohydrate, 22.4% as protein, and 24.1% as fat). ¹³C-octanoic acid (∼110 µL = 100 mg) was mixed with the egg yolk before preparing scrambled eggs for labeling the solid component of the test meal. Because the liquid served with the meal did not contain nutrients, substrates, or any ¹³C-labeled material, the gastric emptying rates reported are those for the solid component of the meal.

Capillary blood was sampled for the determination of plasma glucose concentrations (glucose oxidase method; EBIOS; Eppendorf, Hamburg, Germany; coefficient of variation <1.5%) every 60 min for 6 h. Additional sampling was mandatory when symptoms of hypoglycemia occurred.

Every 20 min for the first 2 h, every 30 min for the following 2 h, and every 60 min for the remainder of the registration, breath specimens were sampled into gas-tight plastic bags holding ∼400 mL. Within 24 h, the ¹³CO₂ content of these samples was determined using near-infrared absorptiometry (Wagner Analysen Technik, Bremen, Germany). Total CO₂ and ¹³C isotope concentrations were determined (20,21). The proportion (percentage) of the orally administered dose of ¹³C exhaled as CO₂ was quantified for each sampling interval (expressed per hour) and used for curve fitting to determine the coefficient a and the gastric emptying coefficient (= ln[a]) as previously described (21). The proportion of the orally administered dose of ¹³C recovered as CO₂ was summed up to yield the cumulated percentage dose over time. The analysis of gastric emptying based on exhaled ¹³C-CO₂ was performed as previously described (21), applying the equations for the Wagner-Nelson approach, which better takes into consideration delays in the appearance of ¹³C-CO₂ in exhaled breath (22,34). Specifically, an elimination coefficient of 0.55 was used (22). T_lag represents the time point after which 10% of the initial gastric content has left the stomach and signals potential delays in initial gastric emptying. T_1/2 represents the time until 50% of the initial gastric content had been emptied.

All other laboratory values were determined by standard clinical chemistry methods.

Patient characteristics are reported as numbers, proportions, or percentages (categorical variables) or means ± SD. Results are reported as mean ± SEM. Differences (Δ) are presented as means and 95% CI.

Statistical calculations were carried out using Statistica version 13.0 (software system for data analysis) (TIBCO Software Inc., https://www.tibco.com) and GraphPad Prism version 8.4.0 for Windows (GraphPad Software, San Diego, CA, https://www.graphpad.com). In the prospective study, the fasting blood glucose tertile (fixed effects) and the subject (random effects) were independent variables, and parameters characterizing gastric emptying were dependent variables. In the retrospective, cross-sectional analysis, fasting blood glucose tertile was the independent variable (fixed effects), and parameters characterizing gastric emptying were dependent variables. Gastric half-emptying time (primary end point) was analyzed using one-way ANOVA with Duncan post hoc test to determine significance of differences between any of the three experimental conditions (low, intermediate, or high glucose in the prospective study; tertiles 1–3 of FPG in the cross-sectional analysis). Time courses of gastric emptying and plasma glucose concentrations were analyzed using general linear models with the same input variables as described for ANOVA but individual values at all time points as dependent variables. This analysis provided P values for effects of different experimental conditions (A), for changes over the time course (B), and for their interaction (AB). If a significant effect of the experimental condition or a significant interaction of experimental condition and time was documented, values at single time points were compared by one-way ANOVA for each time point and (for significant results) were followed by Duncan post hoc test. Gastric emptying parameters were further related to fasting blood glucose and to other gastric emptying parameters by standard parametric regression analysis. Spearman correlation coefficient and the regression coefficient r² were calculated as well as corresponding P values. A P value <0.05 based on a two-sided analysis was taken to indicate significant differences.

Based on a previous study using the same methodology, which determined T_1/2 to be 66.1 ± 16.1 min (35), 20 subjects in the prospective study (paired comparison) would provide a power of >95% to detect a difference of 20% (T_1/2 changing from 59.5 to 72.7 min) around this mean value, and 255 patients (unpaired comparison) would provide >95% power to detect the same difference with an α error of <0.05.

FPG concentrations in patients with type 1 diabetes varied substantially from 5.1 ± 0.3 (low) to 7.3 ± 0.6 (intermediate) and 10.0 ± 1.1 mmol/L (high) and differed by 4.8 (95% CI 3.4; 6.2) mmol/L between the high and low FPG conditions (P < 0.0001) (Fig. 1).

In the 20 subjects with type 1 diabetes participating in the prospective, observational study, the coefficient of variation of their FPG concentrations was 34.1 ± 4.2% (Supplementary Table 1). Despite these marked variations in glycemia, gastric emptying T_1/2 in patients with type 1 diabetes did not differ depending on FPG concentrations with a Δ T_1/2 between highest and lowest FPG of 1 min (95% CI −35; 37) (P = 0.90). The comparison among the low, intermediate, and high blood glucose conditions is presented in Table 2. In line with the gastric emptying T_1/2 (Table 2), the time course of gastric emptying did not differ significantly among the three FPG conditions (Fig. 1A). Plasma glucose concentrations remained significantly different over a period of 180 min, with the high FPG conditions remaining significantly higher than the low FPG conditions for 180 min and than the intermediate FPG condition for 120 min (Fig. 1D).

The number of subjects with an FPG concentration of ≤70 mg/dL (3.9 mmol/L) were 24 (9.4%) in the cross-sectional study, 6 (10.0%) in the prospective cohort with type 1 diabetes, and 1 (3.3%) in the control subjects, respectively. For time points 60–360 min, plasma glucose concentration of ≤70 mg/dL (3.9 mmol/L) were observed in 9.6% (cross-sectional study), 9.2% (prospective study, subjects with type 1 diabetes), and 0.6% (healthy subjects).

The number of subjects with extreme hyperglycemia (>270 mg/dL or 15.0 mmol/L) in the fasting state was 3.5% in the cross-sectional study, 1.6% in the prospective study in subjects with type 1 diabetes, and none in healthy subjects. Extreme hyperglycemia was observed from 60 to 360 min in 4.6% (cross-sectional study), 2.5% (type 1 diabetes, prospective study), and none (healthy control subjects) of the samples.

When including the order of tests in the statistical analysis relating gastric emptying to low, intermediate, or high FPG concentrations, the effect on gastric emptying half-time and gastric emptying lag time remained nonsignificant (details not shown).

The coefficient of variation for T_1/2 determined three times was 21.0 ± 3.3% in patients with type 1 diabetes and 14.5 ± 2.7% in subjects without diabetes. Coefficients of variation for other gastric emptying parameters are found in Supplementary Table 1. Coefficients of variation for FPG were 34.1 ± 4.2% in patients with type 1 diabetes and 5.4 ± 1.3% in subjects without diabetes (P < 0.0001).

For FPG in healthy subjects, similar results at much lower variations (4.6 ± 0.2 [low FPG] to 4.7 ± 0.2 [intermediate FPG] and 5.1 ± 0.6 mmol/L [high FPG]) were found (Fig. 1B and E).

Gastric emptying did not differ between FPG levels with Δ T_1/2 between the highest and the lowest FPG of 7 min (95% CI −10; 23) (P < 0.0001). The comparison of tertiles 1–3 is presented in Table 2. FPG varied from 4.4 ± 0.1 (low, tertile 1) to 7.2 ± 0.1 (intermediate, tertile 2) and 11.7 ± 0.3 mmol/L (high, tertile 3) (i.e., by 7.2 [95% CI 6.7; 7.8] mmol/L between tertiles 1 and 3) (Fig. 1F).

In the low FPG tertile, 24 (28.2%) values were ≤3.9 mmol/L (i.e., in the hypoglycemic range). Under high FPG conditions, 56 (66.7%) values were >10 mmol/L (i.e., in the hyperglycemic range).

Plasma glucose concentrations remained significantly different over a 240-min period, with the high FPG tertile remaining significantly higher than the low and intermediate fasting plasma tertile for 240 and 120 min, respectively, and the intermediate fasting plasma tertile remaining higher than the low FPG tertile for 120 min (Fig. 1F).

Like gastric half-emptying time, gastric emptying lag times or gastric emptying coefficients did not differ with varying FPG categories (Table 2).

Neither in the prospective study nor in the cross-sectional study did FPG concentrations significantly correlate with HbA_1c (Fig. 2A–C), indicating that differences in FPG were the result of day-to-day variations, which are typical for patients with type 1 diabetes (32,33).

When gastric half-emptying and lag times were related to HbA_1c rather than FPG in the cross-sectional study, T_1/2 (P = 0.0025) and T_lag (P = 0.0029) were significantly prolonged by 28% and 31%, respectively, in the tertile with the highest (vs. lowest) HbA_1c concentrations (Supplementary Table 2). A similar analysis was not deemed meaningful in the prospective, observational study. In the cross-sectional study, there was a significant correlation between HbA_1c and T_1/2 as well as T_lag (Supplementary Figs. 1 and 2). Similar associations were described in the prospective, observational study (Supplementary Figs. 1 and 2).

Supplementary Figure 3 shows the distribution of various parameters characterizing gastric emptying from the prospective (patients with type 1 diabetes and healthy subjects) and the retrospective, cross-sectional studies. Prolonged gastric emptying T_1/2 and T_lag occurred mainly in subjects with type 1 diabetes, while mean values for all parameters were similar between patients with type 1 diabetes and healthy subjects. These results were not different between the prospective and retrospective, cross-sectional studies (details not shown).

Classifying subjects based on mean plasma glucose concentrations at baseline and after 60 and 120 min (to test for a potential influence of postmeal rises in glycemia) rather than based on FPG alone did not reveal any significant differences in gastric half-emptying time related to this combined measure of fasting and postmeal glycemia (details not shown).

The overall conclusions would not change had they been based on the conventional Ghoos et al. (21) methodology for analyzing ¹³C octanoate breath tests.

The results of the current study suggest that the velocity of gastric emptying does not depend on the variations in ambient FPG levels in patients with type 1 diabetes. This was uniformly found in both a prospective observational study and a retrospective, cross-sectional analysis of a large database from a single diabetes center (Fig. 1 and Table 2).

Our study confirms that large day-to-day variations in fasting glycemia occur in patients with insulin-treated type 1 diabetes. These variations are even present during a hospital stay aiming for improved glycemic control, a setting that would favor a reduced variability in exogenous factors possibly contributing to glycemic variability. Since there was no significant relationship between FPG and HbA_1c in both the prospective and the cross-sectional studies (Supplementary Fig. 1), and since HbA_1c was not different between the three tertiles of FPG in the cross-sectional study (Table 1), it is obvious that the FPG concentrations must be the result of short-term variations. The fact that neither in the prospective observational study nor in the cross-sectional study FPG concentrations significantly correlated with HbA_1c values (Supplementary Fig. 1) attests to the substantial intraindividual variations in FPG concentrations in type 1 diabetes under the conditions of the current study. The lack of a significant relationship between FPG and HbA_1c in our study may also be explained by the fact that we studied inpatients aiming for improvements in glycemic control. In addition, blood samples for FPG and HbA_1c were not obtained simultaneously.

Bearing in mind that the three gastric emptying studies in the prospective study were performed within approximately a week, this is further confirmed by the similarity of the spread in FPG concentrations in the prospective and cross-sectional studies (Table 1 and Fig. 1). Due to the much larger patient number, more extreme glucose concentrations (including unequivocal hypoglycemia or hyperglycemia) were observed in the cross-sectional study (Table 2) than in the prospective study. Fluctuations of FPG can be assumed to be even larger in outpatients or in patients not currently striving for improved glycemic control with the help of a specialized health professional team. Therefore, the conditions of our study reflect clinical practice and may have important clinical implications.

While studying T_1/2 (Table 2) or the full time course (Fig. 1) clearly supports the conclusion that the velocity of gastric emptying is independent from ambient FPG, relating individual FPG or HbA_1c to gastric emptying T_1/2 or T_lag in our prospective analysis suggested some influence of glycemic status (FPG and HbA_1c) on gastric emptying (Supplementary Figs. 1 and 2). Along these lines, in patients with type 1 diabetes with the highest HbA_1c values (tertile 3), gastric emptying was found to be significantly slower in the cross-sectional study (Fig. 2 and Supplementary Table 2). This is confirmed by the finding that in the cross-sectional analysis, only HbA_1c was significantly correlated to gastric emptying T_1/2 or T_lag, while FPG was not (Supplementary Figs. 1 and 2).

Previous work relating markers of glycemic control to the velocity of gastric emptying in type 1 diabetes has described reduced postprandial glycemic excursions to be associated with delayed gastric emptying, while overall glycemic control from 24-h plasma glucose profiles was worse at the same time (36). While our study focused on short-term relationships among FPG, HbA_1c, and the velocity of gastric emptying, long-term follow-up within the Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) study revealed hyperglycemia over 27 years to be predictive of delayed gastric emptying (37). We now find a similar, but weaker, significant relationship between HbA_1c and gastric half-emptying or lag time (Fig. 2 and Supplementary Table 2). The difference most likely is explained by the availability of a single HbA_1c measurement versus long-term trajectories of HbA_1c in the DCCT/EDIC study (37). Overall, this points to slowed gastric emptying being part of autonomic neuropathy, a complication of diabetes related to chronic hyperglycemia (38,39).

Nevertheless, these subtle differences detected by analyzing individual data did not translate into significant differences in the initial phase of breath test sampling for the group comparisons (Fig. 1). Perhaps more frequent sampling would be needed to pick up differences in early gastric emptying. In addition, the spread of individual FPG concentrations in the prospective study was much larger than the differences among mean concentrations within the low, intermediate, and high plasma glucose tertiles in the cross-sectional analysis (Figs. 1 and 2).

Our findings apparently contradict widely quoted publications demonstrating an acute influence of changing the level of glycemia using the hyper- (23–26) or hypoglycemic clamp (28–31) procedure. In these studies, hyperglycemia, even of moderate degree, slowed gastric emptying of both liquid and solid meal components (23,24,27), whereas hypoglycemia was associated with a dramatic acceleration of gastric emptying (28,29,31). This was the case in both healthy subjects and in patients with type 1 diabetes. Another study in patients with type 2 diabetes did not see a change in the velocity of gastric emptying before and after improving the management of glycemic control, with several weeks between the initial and final study (40). The latter study is more compatible with our results since the time course and intensity of changing glucose concentrations probably was more similar to that during our present study.

Thus, the acute influences of abruptly induced hypo- or hyperglycemia described by Oster-Jørgensen et al. (26), Schvarcz et al. (24,28,29), Samsom et al. (25), Vollmer et al. (27), Fraser et al. (23), and Russo et al. (31) (“square-wave” impulses) may not apply to the type of day-to-day variations in FPG observed under the conditions of the current study. We can only speculate on potential differences between abrupt changes toward hyper- and hypoglycemia induced by hyper- or hypoglycemic clamp experiments and changes in plasma glucose concentrations during everyday life, which may occur in response to variations in carbohydrate intake, physical activity, and associated changes in insulin therapy as well as counterregulatory reactions to episodes of hypoglycemia. Fluctuations of plasma glucose in everyday life most likely occur much more slowly, allowing counterregulatory or adaptive responses. Square-wave impulses as provided by glucose clamps may potentially acutely lead to imbalances between the sympathetic and parasympathetic autonomous nervous system translating into changes in the velocity of gastric emptying.

A potential influence of FPG on the velocity of gastric emptying (not observed under the conditions of the current study) may depend on other prerequisites, which have not been systematically measured in our study: the velocity of change before reaching the current plasma glucose concentration, long-term glycemic control (37), and the degree of autonomic neuropathy.

Like it applies to the relationship of hyperglycemia to the velocity of gastric emptying, low plasma glucose concentrations may more profoundly accelerate gastric emptying at more extreme levels of hypoglycemia (30). The distribution of FPG concentrations observed in our study may not be fully representative of such extreme hypoglycemia (Supplementary Figs. 1 and 2).

While our data do not support a significant influence of current (fasting) plasma glucose on the velocity of gastric emptying, they are compatible with previous reports suggesting that chronic hyperglycemia is associated with slowed gastric emptying (41,42). Looking at those patients with extremely low or high FPG (Supplementary Figs. 1 and 2) or those with extremely low or high HbA_1c (Fig. 2), their results are scattered around the regression line. Since our analysis does not support an influence even of extreme FPG concentrations on gastric emptying, current recommendations for deferring gastric emptying tests (if FPG is >180 mg/dL [10.0 mmol/L]) are not supported by our data. However, studies in the critically ill have found an influence of fasting glucose on gastric emptying (43).

Our results only reflect the emptying of solid meals, which was not found related to FPG concentrations. Hyperglycemia may have more prominent effects on the emptying of liquid rather than solid components (44,45).

Since the setting of this study was a specialized diabetes hospital and patients usually came to this institution because of problems concerning their glycemic control, this probably explains the selection of patients characterized by diabetes-related problems rather than by gastrointestinal symptoms (Table 1).

The present analysis has several strengths: our approach combines the advantages of a prospective observational design and a large cohort of well-characterized patients with type 1 diabetes studied in our cross-sectional analysis of a rather large number of well-characterized patients with type 1 diabetes, who are representative of patients for whom gastric emptying studies may be clinically required (Table 1). A total of 255 patients undergoing gastric emptying studies represent ∼12% of the patients with type 1 diabetes seen at the Diabeteszentrum Bad Lauterberg during the study period. They were selected based on gastrointestinal symptoms, residual gastric content detected after overnight fasts at gastroscopy, glycemic profiles alerting to possible gastroparesis (postprandial hypoglycemia etc.), known autonomic neuropathy, or peculiarities of their insulin regimens (e.g., relatively low proportion of prandial insulin relative to basal insulin). We have used a method for gastric emptying that does not lead to exposure to radioactive nuclides and that is not considered the gold standard. Nevertheless, this method has been validated against gastric emptying scintigraphy, especially when using Wagner-Nelson analysis (22,34). Finally, the prospective study and the retrospective, cross-sectional analysis led to qualitatively and quantitatively similar results.

Some weaknesses need to be addressed. Other evidence of autonomous neuropathy was not widely or even systematically studied in our patient cohorts. Also, we cannot describe results concerning gastric emptying of liquid meal components, which cannot simultaneously be measured by the methodology used. Scintigraphy may provide more accurate and more reproducible results compared with the ¹³C-octanoate breath tests used by us.

Some important conclusions can be drawn from the current study: a gastric emptying test indicated for clinical reasons can be performed at any time and at any level of glycemia. Our analysis does not support the common practice of postponing gastric emptying tests if the blood glucose concentrations do not fall into a narrow euglycemic range.

In conclusion, both a prospective, observational study and a retrospective, cross-sectional cohort analysis indicate the absence of a significant influence of ambient glycemia on the velocity of gastric emptying. This contrasts to published influences of acutely induced hyper- and hypoglycemia using the clamp methodology. We believe that the conditions of the current study better resemble conditions typically found in a clinical setting. Even if we do not find an acute effect of ambient FPG on gastric emptying, we confirm the influence of chronic hyperglycemic on gastric emptying.

Acknowledgments. The authors thank Sabine Schminkel, Heike Schulze, and Ute Buss (Diabetes Center Bad Lauterberg, Bad Lauterberg im Harz, Germany) for technical assistance.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. L.A., B.B., J.J.M., and M.A.N. were responsible for study concept and design. L.A., B.B., S.G., and M.A.N. were responsible for acquisition of data. L.A., B.B., J.J.M., and M.A.N. were responsible for the analysis and interpretation of data. L.A., B.B., and M.A.N. drafted the manuscript. S.G., D.R.Q., and J.J.M. performed critical revision of the manuscript for important intellectual content. L.A., B.B., and M.A.N. performed statistical analysis. M.A.N. was responsible for study supervision. M.A.N. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.