Epidemiology of diabetic gastroparesis

Gastroparesis is increasingly recognised as a significant health problem. The number of hospitalisations for gastroparesis increased by nearly 158% from 1995 to 2004.4 Patients admitted with gastroparesis require more procedures, have a longer hospital stay and incur higher charges than the mean.4 Diabetic gastroparesis was initially described by Kassander in 1958 as ‘gastroparesis diabeticorum’ in patients with type 1 diabetes with gastric retention. Though diabetic gastroparesis has traditionally been associated with advanced type 1 diabetes with poor glycaemic control, it is increasingly being recognised in patients with type 2 diabetes. In a tertiary referral study of diabetic patients undergoing gastric scintigraphy for upper gastrointestinal symptoms, approximately equal numbers of patients had type 1 and type 2 diabetes.5 There are several population-based studies which now show an increased occurrence of symptoms suggestive of upper gastrointestinal dysmotility in patients with type 2 diabetes compared with non-diabetic controls.3 6 7 In a population-based survey of 423 patients with diabetes (94.8% type 2 diabetes), a significantly higher incidence of upper gastrointestinal symptoms in patients with diabetes were reported.3

Given the definition of gastroparesis, its diagnosis requires a test to measure the rate of gastric emptying. The prevalence of delayed gastric emptying in patients with diabetes has been reported to be between 28% and 65%.8 9 However, with the increased availability of gastric emptying tests, it is now well established that there are subsets of patients with delayed gastric emptying and no symptoms,10 accelerated gastric emptying and identical symptoms to patients with delayed gastric emptying, and symptoms with normal gastric emptying.5 From a clinical perspective, the term diabetic gastroparesis is therefore better limited to the combination of a delay in gastric emptying of solids in the absence of obstruction and upper gastrointestinal symptoms including nausea, vomiting, bloating and early satiety.11 Pain is often overlooked, but can be a predominant symptom in a subset of patients with gastroparesis.12 A recent community-based study from Olmsted County in the USA using delayed gastric emptying and typical symptoms as criteria for diagnosis showed a cumulative incidence of 4.8% in type 1 diabetes and 1% in type 2 diabetes as compared with 0.1% in controls.13

The demographics of diabetic gastroparesis, a disease affecting predominantly young females of childbearing age, makes diabetic gastroparesis a disease associated with considerable morbidity and societal impact.13 14 In a single-centre study of patients with gastroparesis seen over 6 years, the mean age of onset was 34 years and 82% of the patients were women.11 The reasons for the female preponderance of gastroparesis remain largely unknown. Gastric emptying is slower overall in females with diabetes as compared with males with diabetes.9 Some studies have shown that gastric emptying is also slower in normal females.15 This raises the possibility that females are closer to the threshold where a decrease in gastric emptying becomes problematic. However, a large randomised controlled study failed to show any alteration in gastric transit in postmenopausal females receiving oestrogen and progesterone supplementation.16 Recently, differences in neuronal nitric oxide synthase (nNOS) dimerisation between females and males have been proposed as an alternative reason for the striking gender difference.17 While diabetic gastroparesis is associated with considerable morbidity, increased mortality on the other hand is usually not due to diabetic gastroparesis but is rather related to cardiovascular and renal disease that often co-exist.18

Pathogenesis

Our understanding, at a cellular level, of the pathogenesis of diabetic gastroparesis has taken a marked step forward in the past 5 years. Additionally, at the patient level, increased knowledge has resulted in the need to re-evaluate long held tenets. An increasingly controversial area is the relationship between symptoms and gastric emptying. Most recent studies show that the correlation between symptoms and gastric emptying is poor.5 8 9 19 These data have been used to suggest that gastric emptying should no longer be used to define diabetic gastroparesis. However, given the increased understanding of the complexity of gastric emptying and of the association between defined changes at a cellular level and changes in gastric emptying, it is possible that the poor correlation reflects the non-specific nature of the symptoms and the inability of current tests to measure different aspects of gastric emptying. Gastric emptying requires coordinated actions of the proximal stomach, which serves as a reservoir that delivers food to the gastric body and antrum, which in turn triturate and mix food, the pylorus, which regulates the size of the food particles that leave the stomach, the proximal small intestine, which feeds back to regulate gastric emptying, and hormones and peptides that are released during normal digestion.20 This complex process requires the interplay of the extrinsic nervous system, enteric nervous system, smooth muscle, interstitial cells of Cajal (ICCs) and immune cells (figure 1A). As new tests are introduced that can separate out segmental as well as global function, it is likely that more will be learned about which, if any, symptom, is associated with which aspect of gastric emptying.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

(A) Anatomy of the human stomach as it relates to normal gastric physiology in the postprandial state. Normal gastric function requires interplay of several cell types including extrinsic nerves, enteric nervous system, immune cells, interstitial cells of Cajal (ICCs) and smooth muscle. The fundus is the receptive organ in the stomach and accommodates a meal without increasing intragastric pressure. Vagal efferents through release of nitric oxide (NO) and other neurotransmitters are primarily responsible for receptive relaxation of the fundus. ICCs generate slow waves which are rapidly propagated to smooth muscle circumferentially and slowly propagated longitudinally, resulting in circumferential contractions that sweep the body and antrum. Smooth muscle cells secrete insulin-like growth factor (IGF-1) which exerts protective effects on ICCs. Macrophages are normally resident in the stomach. In conditions associated with increased oxidative stress, a subclass of macrophages express antioxidant enzymes such as haem oxygenase-1. Haem oxygenase-1 protects against the deleterious effects of increased oxidative stress through generation of products such as carbon monoxide (CO). Antral contractions triturate the food particles against a closed pylorus till they reach a size <3 mm, at which point they pass through the pylorus into the duodenum. Gastric emptying is also regulated by feedback from the duodenum through enterogastric reflexes and release of hormones such as cholecystokinin. (B) Pathophysiological changes in diabetic gastroparesis. Autonomic neuropathy leads to disordered motility and function of the stomach and may underlie the pain experienced by patients. In diabetes, advanced glycation end-products formed due to increased reactive oxygen species can lead to loss of neuronal nitric oxide synthase (nNOS), impaired neurotransmission and delayed gastric emptying. Increased oxidative stress seen in diabetes is usually offset by upregulation of antioxidant enzymes in macrophages such as haem oxygenase-1. Loss of upregulation leads to damage and loss of ICCs, resulting in delayed gastric emptying. In addition, smooth muscle atrophy leads to loss of IGF-1, a survival factor for ICCs.

Another controversial area is the role of hyperglycaemia in gastric emptying. Acute changes in blood glucose are well documented to alter gastric emptying.21 An acute increase in blood glucose decreases fundic tone, decreases the contractility of the mid and distal stomach and also alters small bowel contractile activity. Hyperglycaemia has been shown to stimulate localised pyloric contraction and inhibit antral contraction, resulting in delayed gastric emptying.22 Acute hyperglycaemia can induce gastric myoelectric disturbances particularly tachygastria.23 In addition, hyperglycaemia can also attenuate the effect of prokinetics, reducing their efficacy.24 25 Induction of acute hypoglycaemia accelerates gastric emptying.26 In contrast to these relatively clear effects, the effect of chronic hyperglycaemia on gastric emptying is much less clear. There appear to be differences between chronic hyperglycaemia and gastric emptying in those with type 1 and type 2 diabetes, with varied effects of hyperglycaemia on gastric emptying reported for type 2 diabetes.27 28 The presence or absence of autonomic neuropathy can also markedly alter the effect of hyperglycaemia on gastric emptying.29 30

Changes in gastric emptying also play an important role in blood glucose homeostasis. The rate of gastric emptying is a major determinant of the initial postprandial glycaemic response both in healthy individuals31 and in patients with type 1 or type 2 diabetes,32 and delayed gastric emptying can cause postprandial hypoglycaemia in insulin-treated individuals.33 Hence gastric emptying is an important consideration when developing new treatments to improve glycaemia control in patients with diabetes. Another consideration is the absorption kinetics of drugs, which can be influenced by changes in gastric emptying.

Autonomic neuropathy was one of the first abnormalities associated with diabetic gastroparesis. The sham feeding test, used to evaluate the integrity of the vagus nerve, shows a blunted pancreatic polypeptide response as well as reduced gastric secretion in response to sham feeding in patients with diabetic gastroparesis.34 Vagus nerve dysfunction (figure 1B) is also thought to mediate some of the acute effects of hyperglycaemia such as reduced pyloric relaxation, as a similar effect can be induced by subdiaphragmatic vagotomy.35 Morphologically, both myelinated and unmyelinated nerve fibres of the vagus nerve have been noted to be smaller in the Bio Breeding rat model of spontaneous diabetes.36 37 Similar to the parasympathetic nervous system, changes have also been described in the axons and dendrites within the prevertebral sympathetic ganglia.38 39 Pain, which is a predominant symptom in a subset of patients with diabetic gastroparesis, is often attributed to neuropathy, but better studies are needed to test if this is indeed the case.

The enteric nervous system consists of nearly one hundred million neurons distributed in the myenteric and submucosal plexi and can operate autonomously. Normal gut function requires a balance between the release of excitatory neurotransmitters, such as acetylcholine and substance P,40 and inhibitory transmitters such as nitric oxide (NO)41 and vasoactive intestinal peptide. NO and the enzyme responsible for its generation, nNOS, have been consistently noted to be decreased in patients and animal models of diabetic gastroparesis and have therefore been proposed to be central to the development of delayed gastric emptying (figure 1B). Animal studies, both in a rat model of spontaneous diabetes and in streptozotocin (STZ)-induced diabetes, have shown impaired gastric relaxation and a decrease in nNOS expression and activity in the myenteric plexus.42–45 The loss of nNOS expression cannot simply be explained by neuronal loss as it appears to occur predominantly without accompanying loss of gastric enteric neurons,46 although some animal studies and case reports in humans do also show neuronal cell death which would contribute to the decrease in nNOS expression.47–49 The relative preservation of enteric nerves offers a therapeutic opportunity to target the remaining nNOS-expressing neurons, and/or increase nNOS expression, or target dimerisation of nNOS as a treatment for diabetic gastroparesis. The regulation of nNOS is complex and can be influenced by glucose, insulin or insulin-like growth factor (IGF).50 51 Loss of nNOS expression seen in diabetes may represent inhibition of nNOS by advanced glycation end-products which can bind to myenteric neurons and inhibit nNOS.52 A more recent study shows that gastric relaxation correlates better with the dimerised form of nNOS rather than absolute nNOS levels as seen in female diabetic rats, suggesting that post-translational modification is important.17

While a decrease in nNOS expression is now well established in diabetic gastroparesis, it is unclear if loss of nNOS expression on its own leads to development of delayed gastric emptying. Previous efforts to increase NO using nitroglycerine or using sildenafil to exploit the downstream effector pathway of NO have not shown any clinical benefit in humans.53 54 In the non-obese diabetic (NOD) mouse model of type 1 diabetes, loss of nNOS expression is one of the earliest events associated with development of hyperglycaemia.55 However, the loss of nNOS by itself was not predictive of developing delayed gastric emptying55 and was present irrespective of whether gastric emptying was normal or delayed.55 These data suggest that nNOS may act as a cofactor in the pathogenesis of diabetic gastroparesis or that the form of nNOS may be more relevant than the total amount of the protein. Better methods to deliver NO to the stomach and studies on dimerised versus non-dimerised nNOS are needed to clarify the role of NO in the development of delayed gastric emptying in diabetes as well as its therapeutic potential.

ICCs serve multiple functions in the gastrointestinal tract. ICCs generate slow waves which are then transmitted to the smooth muscles, are involved in aspects of neurotransmission, set the smooth muscle membrane potential gradient and are involved in mechanotransduction.56 Given these multiple functions, it is not surprising that alterations in ICC networks have been looked for in diabetic gastroparesis. Indeed, loss of ICCs is one of the most consistent histological finding in diabetic gastroparesis (figure 1B). Loss of ICCs in the stomach has been well established both in animal models of diabetes and in humans with diabetic gastroparesis.47 48 57–61 The NOD mouse model is the best studied model of type 1 diabetes, and several studies have reported loss of ICC networks in both the corpus and antrum, as examined by Kit expression.55 Kit is a receptor tyrosine kinase expressed in ICCs and mast cells in the gastrointestinal tract. More recently Ano-1 has been shown to be a marker for ICCs throughout the gastrointestinal tract.62 Ano-1 is a calcium-activated chloride channel expressed in ICCs. In the NOD mouse model, loss of Kit expression tightly correlated with development of delayed gastric emptying.55 In that study, all mice that developed delayed gastric emptying had loss of Kit expression while all mice that were resistant to development of delayed gastric emptying had normal Kit expression. Myenteric and intramuscular ICCs are decreased in the stomach of db/db mice (leptin receptor mutant), a model of type 2 diabetes.63 In STZ-induced diabetic rats, significant loss of ICCs was observed in the gastric antrum by 12 weeks.64 Therefore, there is a strong body of literature linking ICC loss to gastroparesis in animal models of diabetes.

Human studies are limited, given the difficulty in obtaining full-thickness biopsies and the difficulty in obtaining control tissue from anatomically corresponding regions. This has resulted in the use of paraffin-fixed human tissue that is not optimal for examination of ICC networks. With these limitations in mind, retrospective human studies have reported significant loss of ICCs in patients with diabetes with gastroparesis. McCallum et al reported profound loss of ICCs in the antrum in 9 out of 23 patients with refractory diabetic gastroparesis.65 Similarly, Iwasaki et al reported loss of intramuscular ICCs in eight patients with severe diabetes, along with loss of nNOS-positive neurons.61 To address the current limitations, the National Institute of Health in the USA has established a Gastroparesis Clinical Research Consortium (GpCRC). Patients enrolled in this study represent the largest cohort of gastroparesis patients who are being followed longitudinally. In a subset of these patients, full-thickness biopsies are collected prospectively and stored appropriately. Preliminary data from this consortium suggest that ∼50% of patients with diabetic gastroparesis have a significant (>50%) decrease in ICCs.46

Loss of ICCs has been shown to result in impaired gastric function. Loss of ICCs observed in diabetic gastroparesis is associated with disruption of the generation and propagation of electrical slow waves, resulting in gastric dysrhythmias. Both bradygastria and tachygastria have been seen in patients with diabetes, with symptoms related to meals.66 Gastric dysrhythmias predict abnormal gastric emptying.67 Both tachygastria and bradygastria may result from ICC loss. A patchy disruption of ICC networks may result in re-entrant tachyarrhythmias as well as loss of generation of the slow waves, resulting in bradyarrhythmias. A recent study reported severe ICC loss in 12 out of 34 patients with refractory diabetic gastroparesis and correlated loss of ICCs with an abnormal electrogastrogram (EGG) showing tachygastria.59 Similar findings have also been reported in idiopathic gastroparesis with loss of ICCs, suggesting that abnormal function is a result of loss of ICCs rather than an effect of diabetes. In animal models of diabetes, abnormal slow wave activity has been reported in STZ-induced diabetic rats, Otsuka Long-Evans Tokushima Fatty rats (model of type 2 diabetes)68 and NOD mice (model of type 1 diabetes).

The mechanism for ICC loss in diabetic gastroparesis has been the focus of recent studies. Loss of ICCs results from imbalance between the processes that injure ICC networks and processes that generate and maintain ICCs.56 One factor is smooth muscle atrophy as a result of relative insulopenia and IGF-1 deficiency in diabetes. This leads to depletion of smooth muscle cell-produced stem cell factor, an important ICC survival factor.58 Smooth muscle degeneration and fibrosis (figure 1B) have been described in diabetes.69 Also, in a rodent model of diabetes, loss of protein kinase C activation and smooth muscle dysfunction have been reported.70 Diabetes is a high oxidative stress state. Recent data obtained using the NOD model of type 1 diabetes suggest that critical to the development of diabetic gastroparesis is the loss of mechanisms that normally counteract increased oxidative stress, such as upregulation of macrophage haem oxygenase-1. This leads to loss of ICCs and development of delayed gastric emptying.55 Upregulation of haem oxygenase-1 by haemin increases expression of Kit and nNOS, and completely reverses the delay in gastric emptying. This finding has clinical implications as haemin can also increase haem oxygenase activity in humans.71 A preliminary study suggests that carbon monoxide mediates the protective effects of haem oxygenase-1 in vivo.72 The studies on IGF-1 and on the haem oxygenase/carbon monoxide pathways provide new avenues for development of treatments for diabetic gastroparesis based on the underlying pathogenesis.

Immune cells have recently been identified as potentially playing a role in development of diabetic gastroparesis. Upregulation of the antioxidant haem oxygenase-1 protects ICCs from damage.55 Upregulation of haem oxygenase-1 occurs in CD206-positive cytoprotective M2 macrophages (figure 1B), while loss of expression of haem oxygenase-1 in these cells leads to development of delayed gastric emptying.73 In a preliminary study from the gastroparesis consortium, abnormal CD45 (a protein tyrosine phosphatase found on leucocytes and other cells of haematological origin) staining was seen in 9 out of 20 patients with diabetic gastroparesis,46 suggesting that immune dysregulation may be an underappreciated feature of the condition. Future studies will need to address whether the changes in macrophages seen in mice also occur in humans and whether immune cell changes are associated with defined changes in cytokines that may alter gastric function.

Diagnosis

Despite the poor correlation between currently available methods to assess global gastric emptying and symptoms, documentation of delayed gastric emptying to solids is required before a diagnosis of diabetic gastroparesis can be made. Abnormal gastric emptying still remains the only objective marker of an underlying defect in the neuromuscular apparatus of the stomach.

Scintigraphy is the most common and widely available modality for measuring gastric emptying. It is, however, relatively expensive, associated with some radiation exposure and, despite position papers published on the topic, still not standardised across medical centres. The standard technique involves scintigraphic determination of emptying of a solid meal. The American Neurogastroenterology and Motility society recommends use of a ^99mTc (technetium) sulfur colloid-labelled egg sandwich as a test meal.74 It is required that scintigraphic measurement continues to 4 h as this has been shown to improve the accuracy of the test.74 However, several centres continue to extrapolate gastric emptying data from 90–120 min readings. The results of tests shorter than 4 h should not be used to make a diagnosis of gastroparesis. In spite of these limitations, scintigraphy remains the test of choice and is considered the gold standard for comparison of newer diagnostic modalities. Imaging should be obtained in relaxed environments and can be obtained sitting or standing as long as the same posture is maintained throughout the test. Hyperglycaemia will delay gastric emptying and therefore needs to be tested for and corrected before carrying out the test. Recent advances allow some separation of distal and proximal gastric function.75 Single photon emission CT (SPECT) imaging following labelling of the gastric mucosa with ^99mTc is helpful in determining gastric volume and can be combined with scintigraphy to measure gastric emptying and accommodation concurrently.76 77

Gastric emptying tests that utilise non-radioactive forms of carbon incorporated in safely ingestible food or liquid products, such as octanoic acid and acetate, correlate well with scintigraphy78 and offer the advantage of being able to be carried out in office settings. The gastric emptying breath test using ¹³C, a stable (non-radioactive) isotope, has been well studied in both human and animal models.79 80 The test relies on the fact that the rate-limiting step in metabolism of ¹³C substrate is emptying of the substrate from the stomach to the intestine. The substrate is rapidly absorbed once it is in the intestine as it does not need digestion; it is catabolised in the liver and excreted as ¹³C in the breath. The ratio of ¹³C to ¹²C provides a reliable measure of the gastric emptying rate. This test has been validated in subjects with diabetes.81 Given the non-radioactive nature of the test, it can be used to measure changes in gastric emptying over time and has a sensitivity and specificity of 75% and 86%, respectively.82 A variation of the ¹³C octanoic acid technique is the ¹³C Spirulina platensis breath test. A recent study validated a gastric emptying breath test using a shelf-stable 238 kcal meal consisting of freeze-dried egg mix, saltine crackers and 100 mg of ¹³C Spirulina platensis. The test was 89% sensitive and 80% specific in identifying delayed gastric emptying using breath samples at 150 and 180 min when compared with gastric scintigraphy.83

Gastroduodenal manometry is invasive, expensive, uncomfortable and of very limited availability. However, when available, it offers the ability to assess the frequency and strength of antral and proximal intestinal contractility, antroduodenal coordination and the presence or absence of phase III of the migratory motor complex. This information can help differentiate between predominantly neuropathic versus non-neuropathic processes, as well as predict tolerability of gastric or small intestinal tube feedings.84

Transabdominal ultrasonography represents a simple non-invasive technique to evaluate gastric function. However, studies are still limited and the technique requires considerable technical expertise. When carried out correctly, transabdominal ultrasonography provides information on global gastric emptying, with high correlation with scintigraphy, and also of accommodation and movement of intragastric contents. Two-dimensional ultrasound can indirectly measure gastric emptying by quantifying changes in antral area over time, and studies have shown increased antral area both in the fasting state and after meals in diabetes.85 Three-dimensional ultrasound provides better information of gastric pathophysiology by allowing assessment of intragastric meal distribution and gastric volume,86 but is time consuming and requires an even more experienced operator. A particular problem with transabdominal ultrasonography is that it is harder in obese patients, which can be an issue in patients with obesity-related type 2 diabetes.

MRI of the stomach also correlates well with scintigraphy.87 It has the additional advantage over ultrasound in that it can differentiate solid and liquid components of intragastric content, secretion and air. The utility of the test is currently limited by the speed with which images can be acquired using most of the current machines, the procedure cost and the time required for interpretation. Newer, faster, higher Tesla machines and better software should make MRI a more attractive option in the future.

Recently, a non-digestible capsule that records pH, pressure and temperature as it travels through the gastrointestinal tract has been introduced as another option to measure gastric emptying. The change in pH between the distal stomach and proximal small intestine allows documentation of egress of the capsule from the stomach and therefore documentation of the time from ingestion to arrival in the small intestine.88 The human pylorus prevents movement of particles bigger than 2–3 mm from the stomach to the small intestine; therefore, emptying of the capsule probably coincides with onset of the phase III migrating motor complex. It can discriminate between normal and delayed gastric emptying with a sensitivity of 87% and specificity of 92% compared with radiopaque markers.89

A challenge to the development of new treatments is the lack of acceptance of tools in clinical trials to assess improvement on treatment by the agencies that approve new drugs. Non-radioactive-based gastric emptying techniques are one option, as is the use of patient-reported outcome measures such as the Diabetes Bowel Symptom Questionnaire, a useful measure of gastrointestinal symptoms and glycaemic control in patients with both type 1 and type 2 diabetes,90 and the Gastroparesis Cardinal Symptom index (GCSI), a patient-reported outcome measure of symptoms of gastroparesis consisting of nine commonly reported symptoms.91 A further modification of the initial questionnaire, the GCSI-DD, has been developed, where patients record symptoms every day, allowing clinicians to capture the daily variability of patient symptoms.92 Validation of the GCSI to meet approval agencies' requirements is currently underway.

Treatment

The past 5 years have seen significant advances in our understanding of the cellular changes that give rise to diabetic gastroparesis and the introduction of several new diagnostic modalities. In contrast, there are fewer options available to treat diabetic gastroparesis than were available 5 years ago. The available treatment options include nutritional support, improvement of gastric emptying using prokinetics, symptom control and, in refractory cases, use of a gastric electric stimulator, although the use of the latter is still controversial. The paucity of approved drugs to treat diabetic gastroparesis is somewhat mitigated by an increasing number of drugs under development with different mechanisms of action.

Nutritional support is often overlooked in patients with diabetic gastroparesis and there is a lack of randomised controlled trials assessing the effect of nutritional intervention on outcome. Patients are often advised to eat small frequent meals, chew their food well, avoid fibre and consume a diet low in fat as studies have shown fat can slow gastric emptying in healthy volunteers.93 This advice makes physiological sense and should be given. However, there are few data to show how these nutritional interventions compare with other treatment modalities for gastroparesis or if they affect the natural history and outcome of patients with diabetic gastroparesis.

In the absence of drugs that target the underlying mechanisms, the aims of treatment of diabetic gastroparesis should be to tighten glycaemic control, treat symptoms and optimise nutritional intake. The effect of erratic gastric emptying on glycaemic control is becoming increasingly clear. Therefore, there may be advantages to the use of prokinetics even in the absence of significant symptom relief. Care must be taken, however, to exclude rapid gastric emptying which is being increasingly recognised as occurring in a subset of patients with diabetes.5

It is important to asses the nutritional status of patients with diabetic gastroparesis as they can have unintentional weight loss, dehydration and electrolyte imbalance secondary to vomiting. Malnutrition can often be missed due to the higher starting weight of patients with gastroparesis and type 2 diabetes.

In milder cases of malnutrition, oral nutrient drinks can be offered, but in malnourished patients with >5% weight loss over 3 months one should consider enteral feeding to bypass the dysfunctional stomach. A nasojejunal feeding tube trial should be carried out for several days prior to placement of an endoscopic or surgical feeding tube as this may unmask any underlying small bowel dysmotility that would result in failure of tube feedings. Gastroduodenal manometry, when available, can be used to assess the small bowel, as potentially can the Smart Pill. It is helpful to keep patients fasting during a trial as it is often difficult to differentiate intolerance to oral feeds and enteral feeds when they continue to eat. Venting gastrotomy may be helpful in some patients and can prevent repeated hospitalisation, but the data to support its benefit are limited. Total parenteral nutrition should be reserved only for patients who fail enteral feeding trials. With gastroenterology, endocrinology and nutritional services working together, this should be a very rare occurrence.

The mainstay of treatment in patients with diabetic gastroparesis has been prokinetic medications. Both metoclopramide and domperidone are dopamine-2 receptor antagonists and equally effective in reducing symptoms of nausea and vomiting in patients with diabetic gastroparesis.94 Metoclopramide is also a weak 5-HT₃ receptor antagonist and 5-HT₄ receptor agonist. Domperidone does not cross the blood–brain barrier and is associated with fewer central nervous system effects. Domperidone is widely available, except in the USA where it requires Institutional Review Board approval and a Food and Drug Administration (FDA) exemption. Metoclopramide remains the primary prokinetic in the USA and is available in liquid and oral dissolving tablets which may be tolerated better. It has also been shown to be effective by the subcutaneous route, bypassing issues associated with delayed gastric emptying.95 However, a well recognised complication of metoclopramide is tardive dyskinesia96 and, since February 2009, the drug carries a box warning in the USA alerting patients about the risks of long-term and high dose use. This has limited its use. A recent study in rodents shows that dopamine-3 receptor agonists inhibit stimulated pyloric relaxation and gastric emptying,97 suggesting that dopamine-3 receptor antagonists given together with dopamine-2 receptor antagonists may be more effective in treating symptoms, although further studies are needed to test this hypothesis.

Erythromycin is a potent prokinetic agent which acts by activating the motilin receptor, probably on the cholinergic neurons. It is a useful agent for short-term treatment of patients in the hospital; however its long-term benefit is limited due to development of tachyphylaxis. Several other motilin agonists have been developed to avoid tachyphylaxis. Mitemcinal (GM-611), a macrolide motilin receptor agonist with acid resistance, has been shown to improve gastric emptying in patients with diabetic gastroparesis.98 GSK962040 is a recently identified small molecule motilin receptor agonist which selectively activates the motilin receptor in humans and is being evaluated to determine safety and tolerability in humans.99 An issue with most motilin agonists is that they also increase gastric tone and therefore can make symptoms worse even when gastric emptying improves. They, of course, should not be used in the subset of patients with rapid gastric emptying. Several other agents have been evaluated for their prokinetic effect, including SK-951, a benzofuran derivative which improves gastric emptying in STZ-induced diabetic dogs,100 and epalrestat, an aldose reductase inhibitor which increases the amplitude of three cycles per minute waves on EGG in patients with diabetic gastroparesis.101 The 5-HT₄ agonists prucalopride and TD-5108 accelerate gastric emptying but have not been tested in gastroparesis.102 103 Phase III clinical trials will be required for all these drugs before their potential role in the treatment of diabetic gastroparesis is established.

Pylorospasm has been reported in diabetes104 and it has been hypothesised to act as a resistance to emptying of gastric content. This has led to the use of intrapyloric injection of botulinum toxin injection to open the pylorus. While initial open-label pilot studies were encouraging, recent randomised controlled trials have failed to show improvement in symptoms.105 It therefore does not appear to have a place in the management of diabetic gastroparesis.

Ghrelin is an endogenous ligand for the growth hormone secretagogue receptor expressed on the vagal afferent neurons and enteric neurons in the stomach.106 Ghrelin has been shown to improve gastric emptying in patients with diabetic gastroparesis in a placebo-controlled study; however, the mechanism by which it exerts the prokinetic effect is unclear and it has a short half-life.107 A ghrelin receptor agonist (TZP-101) has been evaluated in humans, and appears to be well tolerated in patients with diabetic gastroparesis, improving gastric emptying in these patients.108

New drugs have recently been introduced to control postprandial hyperglycaemia, such as amylin analogues and glucagon like peptide-1 (GLP-1) receptor analogues and agonists. These drugs can delay gastric emptying. The amylin (normally co-secreted with insulin by β cells) analogue pramlintide delays gastric emptying in both patients with type 1 and type 2 diabetes.109 The effects of pramlintide on gastric emptying appear to be mediated by the vagal nerve supply to the stomach. Therefore, in patients with vagal dysfunction, the delay in gastric emptying induced by pramlintide may be reduced. Exenatide is a GLP-1 mimetic used in type 2 diabetes and, like the short-acting native GLP-1 hormone, can delay gastric emptying, though this effect may be less pronounced in patients in whom gastric emptying is already delayed.110 We are still learning about the effect of these newer classes of drugs on the gastrointestinal tract and on symptoms in patients with diabetic gastroparesis. Our current understanding is that the benefit of the drugs on glycaemic control outweighs their effects on gastric emptying, but these drugs need to be kept in mind when assessing an abnormal gastric emptying study.

Nausea and vomiting are often the most debilitating symptoms for patients with diabetic gastroparesis. Phenothiazines such as prochlorperazine are one of the most commonly utilised antiemetics, exerting a central effect by acting at the dopaminergic and cholinergic receptors, but they carry a risk of extrapyramidal side effects such as tardive dyskinesia. Other antiemetics such as the 5-HT₃ antagonists odansetron and granisetron primarily developed for chemotherapy-induced nausea and vomiting, cannabinoids, opioid agonists, benzodiazepines and H₁ receptor agonists, such as diphenhydramine, can be used for symptomatic control of nausea and vomiting, though they have not been evaluated in diabetic gastroparesis. Low dose tricyclic antidepressants have been shown to be effective in improving symptoms of nausea and vomiting in some patients with delayed gastric emptying.111 The neurokinin receptor antagonist aprepitant has recently been shown to control vomiting effectively in a patient with refractory diabetic gastroparesis.112 Similarly, the antidepressant mirtazapine, which acts on the 5-HT₃ receptor, has also been shown to be effective in a patient with refractory gastroparesis.113 However, controlled studies are needed to further evaluate the benefit of these medications in improving symptoms in patients with diabetic gastroparesis. Among alternative treatments, ginger114 and acupuncture have been evaluated. Acupuncture has shown benefit in patients with diabetic gastroparesis.115

Gastric electrical stimulation is being increasingly used for patients with diabetic gastroparesis with refractory nausea and vomiting. Initial work in the field focused on pacing the stomach to increase gastric emptying. In contrast, currently available gastric electrical stimulation uses low energy, high frequency stimulation with no substantial effect on gastric emptying.116 The Enterra system, a low energy high frequency gastric electric stimulation device, was approved by the FDA under the humanitarian use device designation in 2000 for treatment of patients with refractory nausea and vomiting due to gastroparesis. A review of the literature published to date suggests that symptoms of nausea and vomiting improve with implantation of the device, as do quality of life and nutritional status; however, most of the studies are open-label single-centre studies. The mechanism of action of gastric electrical stimulation is still not known; the data suggest that afferent neural mechanisms and perhaps modulation of gastric biomechanical activity play a role.117 Gastric electrical stimulation is invasive and there are surgical complications associated with its use. It therefore should be considered only in refractory diabetic gastroparesis after exhaustion of other therapeutic options and only when the main symptoms are nausea and vomiting and not pain. Temporary gastric electrical stimulation may help predict who will respond to gastric electrical stimulation.118 There are currently several alternatives being studied that include long pulse high energy stimulation, as well as single and multiple channel gastric electric stimulation with the goal of achieving sustainable pacing. Preliminary data presented in abstract form show benefit in patients with severe diabetic gastroparesis using two-channel gastric pacing at 1.1 times the intrinsic frequency which normalised gastric dysrhythmia and helped entrain gastric slow waves.119

Pain management is often challenging in patients with gastroparesis as there is a lack of clear understanding of the physiological defect causing pain in these patients. While patients may respond to prokinetics and other conventional treatments used to improve gastric function, some patients require additional medications to help control pain. Medications commonly used for chronic abdominal pain such as tricyclic and tetracyclic antidepressants, gabapentin and pregabalin can be used to treat pain, though there have been no specific studies evaluating their role in gastroparesis. Pain management is best undertaken using a multidisciplinary approach targeting the underlying motility disorder as well as peripheral and central circuits. Opiates should be used very sparingly and only in refractory cases. If needed, a weaker opiate such as tramadol or a κ agonist, asimadoline,120 should be considered.

Future directions

The past 5 years have seen significant advances in our understanding of the pathophysiology of diabetic gastroparesis and the development of several new diagnostic modalities. While treatment modalities are currently severely limited, there do appear to be several promising new drugs in the pharmaceutical pipeline. Overall, the outlook is optimistic. Future treatments will be guided by our increased understanding of the pathophysiology of the disease and we propose can be hastened by the following.

1. Development of non-surgical approaches to full-thickness biopsies of the stomach. While animal models have proven to be very useful, we ultimately need human tissue to determine if what we have learned in animal models also applies to humans. Also, the non-specific nature of the symptoms of diabetic gastroparesis, often indistinguishable from those of patients with diabetes and normal gastric emptying, make a tissue diagnosis attractive as long as it can be carried out without the need for surgery. Recent reports have described non-invasive endoscopic technique to obtain full-thickness gastric biopsies.121 122 Human trials will be needed to determine the viability of these approaches and whether the size of the biopsy obtained is sufficient to examine the key gastric wall components.

2. New treatment based on our increased understanding of the pathophysiology of the disease. Drugs that target the IGF-1 pathway, target reduction in oxidative stress, target ICCs, target dimerisation of nNOS and target immune cells; all may be of therapeutic utility. Of particular interest is the potential use of drugs such as haemin that are already approved for use in humans and target expression of haem oxygenase-1 or drugs that replace its products such as carbon monoxide. In animal models, this approach has been shown to reverse changes in ICCs, nNOS expression, and macrophages.55

3. New prokinetics. While the relationship between current measures of gastric emptying and symptoms is weak, control of gastric emptying is still important to regulate delivery of nutrients to the small intestine. Testing of a variety of ways to better deliver energy to allow true gastric pacing may also serve to achieve the same goals.

4. Studies aimed at better understanding the correlation between electrophysiology, global and segmental gastric function and symptoms.

5. Prospective follow-up studies to determine the long-term outcome of diabetic gastroparesis as well as the long-term outcome of treatments.

6. Development of predictors of which patients will go on to develop diabetic gastroparesis. While the duration of diabetes is well known to be associated with development of diabetic gastroparesis, many patients with diabetes never develop gastrointestinal complications of diabetes while others develop diabetic gastroparesis within a few years of onset of diabetes.

7. Development and validation of better animal models, especially of type 2 diabetes.

8. Studies on the potential use of stem cell-based treatments. Given that we now have a better understanding of the cellular defects, recent advances in regenerative medicine also hold promise for future treatment options in patients with diabetic gastroparesis by restoring tissue integrity. Studies in this area are needed to determine the feasibility of this approach.