Acute effect of clonidine on gastric emptying in patients with diabetic gastropathy and controls


Dr K. H. Soergel, Division of Gastroenterology/Hepatology, Froedtert Memorial Lutheran Hospital, 9200 West Wisconsin Avenue, Milwaukee, WI 53226, USA. E-mail:



The treatments available for diabetic gastropathy are frequently ineffective. Clinical observations suggest that clonidine, an a-2 adrenergic agonist, may improve diabetic gastropathy symptoms.


To establish whether a single oral dose of clonidine alters the gastric emptying of a solid meal in 10 patients with diabetic gastropathy and their matched controls. A secondary goal was to compare two methods of analysis of the data from gastric emptying studies.


Clonidine, 0.3 mg, or a matched placebo were administered orally in a double-blind fashion.


Only three of the 10 patients showed an increased gastric residual volume. Gastric emptying rates were comparable in patients and controls. Clonidine had no significant effect on gastric emptying in the controls but increased t1/2 values in the patient group. This effect just reached statistical significance only when calculated by the power exponential method (P=0.05 but not by the linear component model.


Delayed gastric emptying is not an invariable characteristic of symptomatic diabetic gastropathy. Clonidine, given as a single dose of 0.3 mg orally, has no gastric prokinetic effects. It may act on gastric afferent innervation or, more likely, at a central site to reduce nausea and vomiting. The analysis of gastric emptying data by the power exponential and the two linear component methods yields equivalent results.


Diabetes mellitus, particularly when it is long-standing, poorly controlled and accompanied by peripheral and autonomic neuropathy, is commonly associated with a variety of gastrointestinal complaints, including constipation, diarrhoea, nausea and vomiting.1, 2 Gastroparesis diabeticorum was originally described as radiologically detected gastric retention in asymptomatic diabetic patients.3 The term diabetic gastroparesis is widely used to describe diabetic patients with symptoms of nausea and vomiting, present for at least 3 months, in the absence of other local or systemic causes of these symptoms. Lately, the term has become restricted to patients with a demonstrated delay in gastric emptying, particularly in studies of therapeutic modalities. Increased gastric residual volume, as observed in the original description of the syndrome,3 is rarely mentioned. However, it has been noted that the presence of nausea and vomiting does not correlate with either delayed scintigraphic gastric emptying of solid meals4 or with abnormalities on gastric manometry.5 In fact, gastric emptying is abnormally rapid in some symptomatic diabetic patients.6 Furthermore, the symptomatic response to prokinetic agents appears to be independent of their effects on gastric emptying.4, 7 We designate chronic nausea and vomiting in diabetic patients with no alternate cause for their symptoms as diabetic gastropathy. The treatment goal is the elimination of symptoms, regardless of any abnormalities or induced changes in gastric emptying parameters.

In 1990, it was reported that some patients with diabetic diarrhoea treated with clonidine experienced coincidental relief from nausea and vomiting; these symptoms reappeared when clonidine was stopped and resolved again when therapy was resumed.8 A subsequent open-label pilot study9 of patients with diabetic gastropathy showed that clonidine decreased the symptom score in all patients, with four of six patients becoming symptom free. The half-time of gastric emptying of a liquid meal declined markedly in this group. We have since observed that the oral administration of clonidine, 0.1–0.3 mg b.d., caused long-term symptomatic improvement in the great majority of 26 patients with diabetic gastropathy who had failed standard therapy with metoclopramide, cisapride, erythromycin and bethanechol (K. H. Soergel 2001, unpublished observations). Clonidine, an α-2 adrenergic agonist with central and peripheral actions, has established utility in treating diabetic diarrhoea,10 and its peripheral mechanism of action in experimental and diabetic diarrhoea has been elucidated.11 In view of the general inhibitory actions of this drug on gastrointestinal tone and perception,12, 13 we postulated that any anti-emetic effects would not be due to the acceleration of gastric emptying.

We report a randomized, double-blind, crossover study of the effects of a single oral dose of clonidine, compared to placebo, on the gastric emptying of a labelled solid test meal in 10 patients with diabetic gastropathy and their matched controls. The results were compared with the presence or absence of autonomic neuropathy and other extra-intestinal complications of diabetes mellitus. A secondary goal was to compare the results of two different methods of analysis of the scintigraphic data: the power exponential14 and the two linear component4 models.


We recruited 10 adult patients with type I or II diabetes mellitus who had complained of nausea and vomiting for between 0.5 and 10 years. Upper gastrointestinal series or oesophago-gastro-duodenoscopy and physical examination had excluded other conditions to account for their symptoms. Additional exclusion criteria included systolic blood pressure of < 100 mmHg, diastolic blood pressure of < 65 mmHg, resting pulse rate of < 60/min, previous operations on the gastrointestinal tract and pregnancy. All had received two or more standard medications (metoclopramide, cisapride, erythromycin) to which they had either been unresponsive or developed intolerable side-effects or tachyphylaxis. Two patients were asymptomatic while receiving oral clonidine, but nausea and vomiting resumed during the 3-day drug-free period before testing. A medical history and physical examination were performed and the medical charts were reviewed. Autonomic function testing was performed according to Ewing & Clarke15 and Smith16 to assess two sympathetic and three parasympathetic cardiovascular reflexes. The Valsalva manoeuvre was omitted in patients with retinopathy. Ten age- (± 5 years) and sex-matched non-diabetic healthy control subjects were recruited with the same exclusion criteria applied. Informed written consent was obtained. The study protocol was approved by the Medical College of Wisconsin Institutional and Research and Human Research Review Committees.

Gastric emptying studies

All medications known or suspected to affect gastric emptying were discontinued 3 days before each gastric emptying study. The sequence of oral medication, placebo tablet or clonidine, 0.3 mg, was randomly chosen by the closed envelope method, administered by the pharmacy of Froedtert Memorial Lutheran Hospital, Milwaukee, WI, USA. All subjects were fasted overnight. Patients had their fasting gastric residual volume determined by aspiration through a nasogastric tube. This was repeated before the second emptying study if the aspirated volume was greater than 99 mL. This procedure ensured comparable intragastric volumes at the start of each emptying study. Standing and supine blood pressure and pulse rate were recorded every 30 min at the onset and during the study. Venous blood glucose concentration was maintained at 80–250 mg/dL by intravenous administration of 5% dextrose in water or regular insulin. The study medication was administered 2 h before the standard 225 kcal test meal, which consisted of beef stew and 1 cm3 cubes of chicken liver, injected in vitro with 300 mCi of 99Tc-sulphur colloid.17 With the subject standing before a wide-field gamma-camera (GE Maxi camera 300; General Electric Company, Pewaukee, WI, USA), 1-min anterior and posterior counts were obtained every 10 min until 60% of the radioactivity had left the stomach (region of interest (ROI)) or for 3 h. The geometric means of the anterior and posterior counts were used as data points, corrected for radioactive decay. The data were analysed by the power exponential method14 and the two linear component method previously described by our group.4

As the gastric emptying results were not normally distributed, they are presented as median values and ranges. The Mann–Whitney rank sum test was used to test differences between the groups, with P < 0.05 considered to be significant.


Six of the 10 patients and six of the 10 controls were male. Their ages (mean, range) of 47.6 years(31–64 years) and 46.2 years (27–63 years) were closely matched. The duration of diabetes was 15.8 years (4–23 years) and the onset of nausea and vomiting was 4.0 years (0.5–10.0 years) ago, with no significant difference between the six patients withtype I and the four patients with type II diabetes. Nine of the 10 patients were using insulin. Retinopathy was present in three patients, peripheral neuropathy in eight, nephropathy in none and symptomatic orthostatic hypotension in four.

The fasting gastric residual volume was normal (< 100 mL) in seven patients and was increased at 140, 100 and 220 mL in three. On repeat testing before the second gastric emptying study, the volumes obtained were 40, 20 and 25 mL, respectively.

The results (median; range) of the solid meal gastric emptying studies are presented in Table 1. During the placebo studies, there was no significant difference in the t1/2 and β values calculated by the power exponential method, or in the t1/2 values calculated by the two linear component method, between patients and controls. Among the patients, clonidine resulted in the prolongation of the half-emptying time which reached statistical significance only with the power exponential method, while the value of the parameter β decreased insignificantly, indicating a trend towards more rapid initial emptying. Clonidine had no significant effect on the gastric emptying parameters in the control subjects.

Table 1.  . Solid meal gastric emptying results calculated according to the power exponential14 and the two linear component4 models Thumbnail image of

As shown at the bottom of Table 1, the t1/2 values obtained by the two methods of calculation were numerically similar and did not differ significantly within the four groups of gastric emptying studies.

The two linear component analysis yields the additional parameters of the slopes of the initial (b1) and second (b2) linear components and the duration (C) of the initial emptying phase4 (Figure 1) (data not presented). None of these parameters differed between patients and control subjects, or between placebo and clonidine studies in either group.

Figure 1.

. Gastric emptying data obtained by the two-component linear method: disintegrations per minute (dpm) within the region of interest (ROI) are fitted to linear components by least-squares regression analysis until the value C, yielding the smallest sum of residual mean squares, is found. b1, b2, slopes (%/min) of the two lines; C, their intercept (min); t1/2, half-emptying time (min). See Loo et al.4 for details.

While gastric emptying rates did not differ significantly between control and patient groups, three of the 10 gastropathy patients had prolonged t1/2 values and one had an abnormally short t1/2 value when compared with the gender-specific control ranges (mean ± 2s.d.) previously established with an identical test protocol in the Section of Nuclear Medicine. On individual comparision, the effect of clonidine on gastric emptying parametrs did not differ between the 4 patients with abnormal t1/2 values vs. the 6 patients with normal t1/2 values.

Autonomic function testing

The three tests assessing parasympathetic cardiovascular function (variations in the heart rate with the Valsalva manoeuvre, deep respiration and on standing) yielded abnormal results in five of six, nine of nine and six of nine patients, respectively. The two tests of sympathetic function (blood pressure response to standing and with sustained hand grip) were abnormal in seven of nine and six of 10 patients tested. Two or more of the five test components were abnormal in all nine patients in whom two or more components could be assessed. There was no correlation between the results of autonomic function testing or the presence of extra-intestinal complications of diabetes and the results of the gastric emptying studies.

Three-month follow-up

Two patients again became asymptomatic immediately after resuming oral clonidine (0.1 mg b.d.) which they had been taking for 1 and 3 years, respectively. Nausea and vomiting had recurred during the first day of each of the two drug interruptions required by the study protocol. Four patients accepted the offer of oral clonidine therapy, 0.1–0.3 mg b.d.; three of these four were asymptomatic at the 3-month follow-up.


Of the available prokinetic agents, metoclopramide, domperidone and cisapride all act as agonists for 5-hydroxytryptamine-4 (5-HT4) receptors on cholinergic motor neurones, resulting in enhanced acetylcholine release at the nerve terminals which, in turn, activates muscarinic M3 receptors on smooth muscle. Metoclopramide and domperidone differ from cisapride in two major ways: first, they block dopamine-2 (D2) receptors on motor neurones, causing increased acetylcholine release; second, they cross the blood–brain barrier (metoclopramide > domperidone), gaining access to the vomiting centre where they exert anti-emetic actions directly by their antagonistic action on D2 and 5-HT3 receptors. Erythromycin acts peripherally by occupying motilin receptors.18

A recent meta-analysis concluded that, in patients with diabetic gastropathy, the above-mentioned four agents decreased the symptom score by 50% or less in placebo-controlled studies. For domperidone and metoclopramide, the two drugs with access to the vomiting centre, the improvement in symptom score was twice their effect of decreasing the t1/2 value of gastric emptying.7 We wished to determine whether any beneficial effects of clonidine on gastropathy symptoms occurred in the absence of prokinetic activity.

Clonidine, given by mouth, is > 99% bioavailable; the peak plasma concentration is reached at about 60 min and levels remain elevated for greater than 4 h after dosing.13 It is a centrally acting α-2 adrenergic agonist which decreases vascular reactivity, possibly by suppressing central noradrenaline (norepinephrine) release.19 Peripheral actions in humans include increased compliance and decreased visceral sensation and pain perception in the stomach and colorectum.12, 13 However, perceptions of bloating and nausea during gastric distension are not affected. These actions are believed to be mediated by the activation of pre-junctional adrenergic α-2 receptors on cholinergic nerve terminals of enteric neurones, leading to decreased acetylcholine release and to the inhibition of the afferent pathway of visceral nociception.12 Thumshirn et al. reported no effect on the emptying of a solid meal with single oral doses up to 0.1 mg,12 while Rosa-e-Silva et al. observed accelerated emptying of a liquid nutrient meal in six gastropathy patients after prolonged treatment with clonidine.9 Our results with a single dose of 0.3 mg are partly consistent with these reports. There was no effect on the emptying of a solid meal in healthy controls, while emptying appeared to be slowed in symptomatic diabetic gastropathy patients, although this effect was statistically significant with only one of the two methods of calibration. We confirm that diabetic gastropathy is not characterized by uniformly delayed gastric emptying.4 Thus, the term diabetic gastroparesis refers only to a sub-group of diabetic patients with chronic nausea and vomiting who have documented delays in gastric emptying and/or an increased gastric residual volume. Only five of our 10 patients met either one or both of these criteria.

Clonidine induces vomiting in dogs via α-2 adrenergic receptors in the chemoreceptor trigger zone, located outside the blood–brain barrier. This effect is blocked by yohimbine and scopolamine, but it is separate from the emetic actions of other chemoreceptor trigger zone stimulants, such as apomorphine, and direct irritation of the upper gastrointestinal tract. In fact, clonidine delays the onset of vomiting due to other stimuli.20, 21 By contrast, clonidine is an effective anti-emetic in several clinical settings. It reduces post-operative vomiting after eye and knee surgery,22, 23 and decreases anticipatory vomiting and nausea with chemotherapy.24

In summary, the therapeutic effect of clonidine in diabetic gastropathy, while not yet established in controlled studies, is probably caused by its action on the α-2 adrenergic receptor in the chemoreceptor trigger zone rather than by effects on gastric motor function or the gastric sensory threshold.12 Our study design did not address the questions of whether chronic clonidine administration accelerates gastric emptying, or whether a prokinetic effect is present in subgroups of diabetic gastropathy patients, such as those with abnormally rapid or slow, or with normal gastric emptying rates. The evidence available favours neither of these possibilities. Clonidine causes increased compliance in the human gastrointestinal tract;12,13 the acute effect of clonidine shows a trend toward delayed emptying in diabetic gastropathy patients (Table 1); and our two diabetic gastropathy patients who had been on long-term clonidine theraphy did not respond to the test dose with a decrease in the t1/2 value of gastric emptying. With regard to the second question, the response of our diabetic gastropathy patients to clonidine, 0.3 mg, was essentially the same, regardless of their gastric emptying rate during the placebo study. However, this observation is based on a post-hoc analysis of a small number of observations.

Elashoff et al.14 listed several desirable characteristics of methods for the analysis of gastric emptying data, including the following: the entire emptying curve should be included in the analysis; the curve should start at 100% at the time of meal ingestion; and the parameters calculated should have clear graphic interpretation. These authors proposed a power exponential, f=2−(t/t½)β, where the factor β varies with the slope of the initial emptying phase, with values greater than 1.0 indicating a slow initial phase. However, the duration of this initial emptying phase is not apparent (Figure 2). Further, this model incorporates symmetry of the emptying curve around an inflection point, an assumption not experimentally verified. The two linear component method of analysis, as described by Loo et al.4 yields parameters that directly describe the emptying process, and whether one or two different phases of emptying are present (Figure 1). Like the power exponential model, this method provides a better data fit than single linear and simple exponential models.4 As shown in the present study, the t1/2 values obtained by either method yield comparable results.

Figure 2.

. Two gastric emptying curves obtained by the power exponential, f=2−(t/t½)β.14 Parameter β, describing the curve shape, is 2.0 in both instances. t1/2=60 min (full line) and 10 min (dotted line). f is the fraction of disintegrations per minute recorded from the region of interest at time t.


The study was supported by a grant from the Juvenile Diabetes Foundation International. We thank Mark Kern for advice and the performance of statistical calculations.