Delayed Gastric Emptying and Gastric Autoimmunity in Type 1 Diabetes

We studied 42 euthyroidic type 1 diabetic patients, comprising 15 PCA+ and 27 PCA− subjects (male/female 29/13; mean age 40 ± 15 years; mean diabetes duration 18 ± 9 years; mean HbA_1c 7.8 ± 0.9%). The patients were consecutively recruited, regardless of gastrointestinal symptoms, from 69 PCA+ and 100 PCA− patients attending the Antwerp University Diabetes Clinic, and fulfilled criteria for type 1 diabetes (29). Patients with a history of gastric surgery or gastric outlet obstruction or who were taking motility-altering drugs, anti-inflammatory drugs, antacids, or proton pump inhibitors were excluded. Smoking, drinking beverages containing alcohol or caffeine, and strenuous exercise were prohibited in the 24 h before the gastric emptying tests. The local ethics committee approved the study, and each patient gave written informed consent.

PCAs were assayed by indirect immunofluorescence (Medical Diagnostics California, Carlsbad, CA; Nl <1/20 dilution) (23). Gastrin was measured by radioimmunoassay (Euro-Diagnostics, Malmö, Sweden; Nl <110 ng/l). Thyroid peroxidase antibodies were assayed by radiobinding assay (Henningtest, Brahms, Germany; Nl <100 units/ml). Thyroid function was estimated by assay of thyroid-stimulating hormone (TSH) (Nl 0.47–4.7 mU/l) and free T4 levels (fT4) (Nl 10.8–21.9 pmol/l) (Vitros; Ortho Clinical Diagnostics, Amersham, Amersham, U.K.).

Gastrointestinal symptoms were assessed by questionnaire. Anorexia, nausea, early satiety, abdominal bloating/fullness, vomiting, abdominal pain, dysphagia, heartburn, and acid regurgitation were graded as 0 = none, 1 = mild (the symptom could be ignored), 2 = moderate (the symptom could not be ignored, but did not influence daily activities), and 3 = severe (the symptom influenced daily activities) (3). Reproducibility of this questionnaire was measured by a test-retest procedure, and validity was documented by comparing questionnaire data with an independent interview. The reproducibility and validity were high in this study (coefficients of variation [CVs] 4.8 and 9.5%, respectively).

Autonomic nerve function was documented by standardized cardiovascular reflex tests (30). Parasympathetic function was evaluated by the variation of heart rate (R−R interval) during deep breathing and the immediate heart rate response to standing (30:15 ratio) and the Valsalva maneuver. Sympathetic function was assessed by the fall in systolic blood pressure in response to standing. The results were classified as follows: 0 = normal; 1 = borderline cardiovascular autonomic neuropathy (CAN; one abnormal heart rate test), 2 = definite CAN (≥2 abnormal heart rate tests), and 3 = severe CAN (all abnormal heart rate and blood pressure tests). The reproducibility, expressed as the CV for autonomic neuropathy test, was 2.4% for heart rate variability, 9.5% for the deep breathing test and Valsalva maneuver, and 7.1% for the heart rate response to standing and for the postural blood pressure test.

HP IgG antibodies were assayed by enzyme-linked immunosorbent assay (Roche Diagnostics, Brussels, Belgium). Urea breath tests, using 75 g ¹³C-urea, were performed after a 12-h overnight fast (31). Antrum, corpus, and fundus biopsies were examined for HP colonization by a modified Giemsa stain and/or immunostaining. HP infection was diagnosed if any test was positive. The period between the urea breath test and gastroscopy was ≤14 days.

At upper gastrointestinal endoscopy (Olympus Videoscope GIF/Q140; Melville, NY), at least two biopsies from fundus, corpus, antrum, and descending duodenum were taken. Then, two investigators evaluated 5-μm sections, unaware of the patient’s clinical and laboratory findings. The visual analog scale of the updated Sydney system was used to evaluate inflammation, activity, atrophy, intestinal metaplasia, and HP colonization (32). HP immunostaining was performed by standard procedures using a polyclonal rabbit HP IgG antibody (Dako B0471; Dako, Glostrup, Denmark; dilution 1/200).

Gastric acid secretion was studied in euglycemic conditions after a 12-h overnight fast using pentagastrin-stimulated acid output (6 μg/kg body wt., s.c.) (Pentagastrin Injection BP; Cambridge Laboratories, Newcastle Upon Tyne, UK). Gastric juice was collected for 1 h before and 1 h after pentagastrin administration in 15-min aliquots. Hypochlorhydria was defined as a maximal acid output <15 mmol H⁺/h.

To eliminate the effect of acute hyperglycemia on gastric emptying, patients’ blood glucose levels were stabilized in the euglycemic range. Patients received their standard morning insulin dose at least 30 min before the test meal. At 15-min intervals, glycemia was tested by fingerstick (One Touch blood glucose meter; Lifescan, Milpitas, CA) to maintain glycemia at 4.5–6 mmol/l for at least 30 min before and at 4.5–8 mmol/l after the test meal. One intravenous cannula was inserted into antecubital vein for infusion of glucose 5% buffered with 10 units short-acting insulin at 50 ml/h (Human Actrapid; Novo Nordisk, Copenhagen, Denmark) to maintain euglycemia. If glycemia exceeded ≥ 7 mmol/l, intravenous short-acting insulin was given: 1 unit for glycemia ≥7 mmol/l, 2 units for glycemia ≥8 mmol/l, and 3 units for glycemia ≥9 mmol/l. If glycemia was <3.5 mmol/l, 20% glucose in 10-ml aliquots was given to normalize levels: 1 ampulla of 10 ml glucose 20% for glycemia <3.5 mmol/l and 2 ampullae for glycemia <2.5 mmol/l. To avoid ¹⁴C, we used two ¹³C breath tests, which were performed within an 8- to 24-day period.

The test was performed after a 12-h overnight fast. Patients avoided ¹³C-rich meals 48 h before the breath test to increase its accuracy. Baseline breath ¹³CO₂ before eating the test meal was measured. The test meal consisted of a beaten egg with the yolk doped with 100 mg of ¹³C-octanoic acid (Isotec, Miamisburg, OH), two slices of white bread, and 5 g margarine (33,34). The yolk was baked separately from the egg white to ensure firm labeling. The meal, with a caloric value of 250 kcal (42% carbohydrates, 18% protein, and 40% fat), was ingested in 10 min, and patients were allowed to drink 150 ml water in this period. Breath samples were collected over a 4-h period at 15-min intervals. ¹³CO₂ measurements were performed with an isotope ratio mass spectrometer (Finnigan MAT-250; Finnigan, Bremen, Germany). Mathematical analysis of the excretion rate of ¹³CO₂ allowed the definition of the gastric-emptying coefficient (GEC) and the gastric half-emptying time (t_1/2). With a cutoff value for t_{1/2breathsolids} of 75 min and a cutoff value for GEC_breathsolids of 3.1 min, excellent sensitivity, specificity, and positive and negative predictive values were obtained to separate normal subjects from patients with delayed gastric emptying (33,34).

We chose 100 mg of ¹³C-glycine as the marker of the liquid phase because it is easily solubilized in water and is rapidly absorbed and converted into ¹³CO₂ after it enters the small intestine. Cutoff values for the GEC_{breathliquids} and t_{1/2breathliquids} were 3 and 55′, respectively (34).

Good-to-excellent correlations between the GEC and the scintigraphic half-emptying time (r = 0.74 for liquids, r = 0.88 for solids) and between the t_1/2 determined by breath test and scintigraphy (r = 0.91 for liquids, r = 0.92 for solids) have been previously described (33,34).

Results were statistically analyzed using SPSS software (SPSS, Chicago, IL). The unpaired t test or Mann-Whitney U test was used to determine differences between groups. Differences in distributions of categorical data were investigated by χ² or Fisher’s exact test. The Pearson or Spearman rank correlation test was used. Stepwise forward logistic regression analysis was used to assess the strength and independence of associations. A two-tailed P < 0.05 was considered significant.

Gastrointestinal symptoms were reported by 43% of patients, but the majority of patients had no or only mild symptoms, with a median symptom score of 4. Early satiety and abdominal bloating were present in 17 patients (33%), nausea and/or vomiting in 7 patients (17%), abdominal pain in 6 patients, and regurgitation and/or heartburn in 5 patients. No associations were noted between presence or score of symptoms and indexes of solid and liquid emptying (GEC and t_1/2). Autonomic neuropathy was present in seven patients (three severe, two definite, and two borderline).

Sex, age, diabetes duration, HbA_1c, and presence and severity of gastrointestinal symptoms were similar in PCA+ and PCA− diabetic subjects (Table 1). PCA+ patients were more prone to having hypochlorhydria (odds ratio 12.1 [2.7–54.3], P = 0.0008) and higher gastrin levels (P = 0.003) than PCA− subjects. However, no differences were found in the prevalence of autonomic neuropathy, gastrointestinal complaints, HP infection, or delayed solid and liquid gastric emptying between PCA+ and PCA− subjects (Table 1).

Solid gastric emptying was delayed (GEC_breathsolids <3.1, t_{1/2breathsolids} >75′) in 40% of type 1 diabetic patients. The mean glycemia during solid emptying was 119 ± 37 mg/dl. In this range, glycemia did not correlate with indexes of solid emptying (GEC and t_1/2). However, patients with delayed solid emptying had higher HbA_1c levels than those with normal solid emptying (P = 0.005) (Table 2). Autonomic neuropathy was equally common in both groups, and no association was observed between CAN score and rate of solid emptying. No differences were found in gastric histology (i.e., absence/presence of autoimmune gastritis), acid secretion parameters (basal, maximal, or peak acid output), or gastrin levels between those with and without gastroparesis. HbA_1c level (β = 1.34;P = 0.011) was the only independent risk factor for delayed solid gastric emptying in a logistic regression model testing HbA_1c, autonomic nerve function, PCA, HP status, gastrin, TSH, age, sex, and diabetes duration. The t_{1/2breathsolid} correlated strongly with GEC_breathsolid (r = −0.81, P < 0.0001). Patients with delayed solid emptying tended to have a delay in liquid emptying (P = 0.1); furthermore, a correlation between t_{1/2breathsolid} and t_{1/2breathliquid} (r = 0.44, P = 0.037) was found. Solid and/or liquid emptying was delayed in 23 patients (55%).

Liquid gastric emptying was delayed (GEC_{breathliquids} <3, t_{1/2breathliquids} >55′) in 36% of type 1 diabetic patients. The mean glycemia during liquid emptying was 129 ± 34 mg/dl. No correlations were observed between the level of glycemia and indexes of liquid emptying (GEC and t_1/2). Although each subject was euthyroidic, patients with delayed liquid emptying had higher TSH levels (P = 0.02) than those with normal liquid emptying (Table 3). Further, a strong correlation was noted between t_{1/2breathliquid} on one hand and TSH levels (r = 0.83, P < 0.0001), fT4 levels (r = −0.65, P = 0.021), and CAN score (r = −0.79, P = 0.001) on the other. However, logistic regression analysis could not identify a single independent risk factor for liquid gastric emptying when HbA_1c, autonomic nerve function, PCA and HP infection status, gastrin and TSH levels, age, sex, and duration of diabetes were tested.

This study investigated the putative role of PCA, present in 15–20% of type 1 diabetic patients (23), in the etiology of diabetic gastroparesis, using breath tests for liquid and solid emptying in euglycemic conditions. More than half of our type 1 diabetic patients (55%) had delayed gastric emptying (solids in 40% and liquids in 36%), in line with previous data (1–3). HbA_1c level was identified as a risk factor for delayed solid emptying. Autonomic nerve function, PCA or HP status, gastric histology, acid secretion, or gastrin or TSH level did not seem to affect solid emptying. On the other hand, TSH level and autonomic neuropathy score correlated with t_{1/2breathliquid}. Age, sex, and diabetes duration did not correlate with gastric emptying rate, supporting results from other studies (13,35,36). Breath tests have several advantages over scintigraphy, with which they correlate very well (see RESEARCH DESIGN AND METHODS). They are easy to perform, noninvasive, have good reproducibility, and avoid radiation burden, thereby facilitating emptying studies in fertile women and children (13,27,28,33,34). Moreover, many tests can be done simultaneously because ¹³CO₂ can be analyzed overnight. Unlike a number of studies, ours controlled for many variables that may affect gastrointestinal symptoms and emptying: glucose level at the time of gastric emptying, gastric mucosal status, HP status, and autonomic nerve function.

Most patients had no or only minor complaints, reflecting the fact that they were not selected on the basis of symptoms. We observed no association between gastric motility and symptoms, confirming some (7,37), but opposing other reports (4,13,38). In line with recent data (38), bloating was the most common symptom. Abnormalities in visceral sensation (39) or a variable degree of vagal afferent and efferent dysfunction (1) might explain the poor link between symptoms and gastric motility. Glycemic control may also influence the prevalence or perception of gastrointestinal symptoms (15,40), although this was not supported by our results.

Gastric neuromuscular dysfunction may account for poor metabolic control, resulting from the dyssynchrony of insulin administration and emptying of nutrients (2,3). All our patients had good metabolic control. Nevertheless, we found that HbA_1c levels were associated with delayed solid emptying, contrary to findings in other studies (12,13). It is not clear whether worse metabolic control determines or is determined by delayed gastric emptying. Prospective follow-up studies designed to see whether improving HbA_1c would also improve gastric motility may help to clarify this interesting finding. Acute hyperglycemia may slow down gastric emptying (14,15) by evoking pyloric contractions (41) or suppressing antral contractions (42). In contrast, others found no link between glycemia and gastric motility (4,6). In euglycemic conditions, we did not see an effect of glucose levels on gastric emptying.

Autonomic neuropathy with loss of vagal tone and increased sympathetic nervous system activity has been associated with gastric dysrhythmias (5,43). Years ago, Kassander (5) noted the similarity between clinical and radiographic findings of gastroparesis and gastric atony observed after surgical vagotomy. In addition, some (44), but not all (45), researchers have found histologic abnormalities in vagal nerve tissue of diabetic patients. We did not observe a link between CAN scores and rates of solid emptying, confirming results from some studies (4,27,37), but opposing results from other studies, describing a weak association (6,12,13). However, a good correlation was found with t_{1/2breathliquid}. Perhaps these conflicting results are caused by differences in neuropathic involvement in solid and liquid emptying (4) or by differences in accuracy between classical CAN tests and the new ones using spectral and vector analysis (46). Our approach in which breath tests were performed on 2 days to avoid the use of ¹⁴C could also have introduced a source of variability. However, all tests were done in euglycemic conditions using standard procedures. Despite the poor relationship between emptying of solids and liquids (6,13), suggesting that predictors of delayed solid emptying differ from those of liquid emptying (38), we observed a weak correlation between t_{1/2breathsolid} and t_{1/2breathliquid}; in addition, patients with delayed solid emptying tended to have a delay in liquid emptying.

Cyclic variation in gastric acid secretion, synchronous with the various phases of the migrating motor complex, have been shown in duodenal ulcer disease (21). Moreover, one group observed a delay in gastric emptying in patients with chronic atrophic gastritis and hypo- or achlorhydria (20). However, others have reported that interdigestive antroduodenal motility in achlorhydria is not different from that in normal acid-secreting control subjects (21). Our results support the observation that acid secretion is not essential in the regulation of gastric motility (47). Moreover, we did not observe a link between gastric motility and PCA positivity, the latter being associated with hypochlorhydria, or gastric histology. Furthermore, HP infection did not influence gastric emptying or the presence or severity of gastrointestinal symptoms (22). This is an important observation, because abnormal gastric emptying could affect urea breath test results. We and others have found no association between urea breath test and gastric emptying breath test results (22).

Despite the fact that all patients were euthyroidic, we observed a strong correlation between t_{1/2breathliquid} and TSH levels. Thus subclinical hypothyroidism may affect gastric motor activity, as does overt hypothyroidism (19). The pathogenesis of gastric motor hypoactivity is not precisely known. Autonomic neuropathy, disturbance of impulse conduction at the myoneural junction, changes in gastrointestinal peptide hormone metabolism, and autoimmune factors might explain this dysfunction.

In conclusion, more than half of type 1 diabetic patients studied had a delay in gastric emptying that did not correlate with symptoms. Gastric autoimmunity did not contribute to diabetic gastroparesis. HbA_1c level was the only independent risk factor for delayed solid emptying. Autonomic nerve function, PCA or HP infection status, age, sex, diabetes duration, gastric histology, gastrin levels, and acid secretion did not seem to affect solid emptying. On the other hand, TSH level and autonomic neuropathy score were correlated with t_{1/2breathliquid}. It is important to report the lack of contribution of gastric autoimmunity to delayed gastric emptying and to propose an etiological role of HbA_1c and subclinical hypothyroidism. It is hoped this work will stimulate new hypotheses and encourage the use of breath tests.

This study was supported by a grant from the European Foundation for the Study of Diabetes.

The technique of the breath tests was introduced to the Department of Gastroenterology by Prof. Dr. Y. Ghoos and Dr. B. Geypens (Department of Gastroenterology, University Hospital Gasthuisberg, Leuven, Belgium).

We gratefully acknowledge Dr. K. Van Acker and Dr. F. Peiffer (Department of Endocrinology-Diabetology) for their help in patient recruitment, Dr. C. Van Campenhout and K. Van Cotthem (Laboratory of Immunology and Protein Chemistry), Prof. Dr. E. Van Marck and Prof. Dr. J. Bogers (Department of Pathology), Prof. Dr. V. Van Hoof (Laboratory of Biochemistry), Apr. M. Martin (Laboratory of Hormonology), Prof. Dr. M. Ieven (Laboratory of Microbiology), B. Raeymaeckers and R. Borrie (Department of Gastroenterology), and Dr. B. Manuel y Keenoy, J. Vertommen, P. Aerts, S. Schrans, and M. Vinckx (Laboratory of Endocrinology-Diabetology) of the University of Antwerp.

Parts of this study were presented at the 61st Scientific Sessions of the American Diabetes Association, Philadelphia, PA, June 2001.