Diabetic Gastroparesis

Abstract

This review covers the epidemiology, pathophysiology, clinical features, diagnosis, and management of diabetic gastroparesis, and more broadly diabetic gastroenteropathy, which encompasses all the gastrointestinal manifestations of diabetes mellitus. Up to 50% of patients with type 1 and type 2 DM and suboptimal glycemic control have delayed gastric emptying (GE), which can be documented with scintigraphy, ¹³C breath tests, or a wireless motility capsule; the remainder have normal or rapid GE. Many patients with delayed GE are asymptomatic; others have dyspepsia (i.e., mild to moderate indigestion, with or without a mild delay in GE) or gastroparesis, which is a syndrome characterized by moderate to severe upper gastrointestinal symptoms and delayed GE that suggest, but are not accompanied by, gastric outlet obstruction. Gastroparesis can markedly impair quality of life, and up to 50% of patients have significant anxiety and/or depression. Often the distinction between dyspepsia and gastroparesis is based on clinical judgement rather than established criteria. Hyperglycemia, autonomic neuropathy, and enteric neuromuscular inflammation and injury are implicated in the pathogenesis of delayed GE. Alternatively, there are limited data to suggest that delayed GE may affect glycemic control. The management of diabetic gastroparesis is guided by the severity of symptoms, the magnitude of delayed GE, and the nutritional status. Initial options include dietary modifications, supplemental oral nutrition, and antiemetic and prokinetic medications. Patients with more severe symptoms may require a venting gastrostomy or jejunostomy and/or gastric electrical stimulation. Promising newer therapeutic approaches include ghrelin receptor agonists and selective 5-hydroxytryptamine receptor agonists.

Essential Points

• Up to 50% of patients with moderately controlled type 1 and type 2 diabetes mellitus (DM) have delayed gastric emptying (GE), which can be documented with scintigraphy, ¹³C-breath tests, or a wireless motility capsule

• Many patients with DM and delayed GE are asymptomatic; others have dyspepsia (i.e., mild to moderate indigestion, with or without a mild delay in GE) or gastroparesis, which is a syndrome characterized by moderate to severe upper gastrointestinal symptoms and delayed GE that suggest, but are not accompanied by, gastric outlet obstruction

• Hyperglycemia, autonomic neuropathy, and enteric neuromuscular inflammation and injury are implicated in the pathogenesis of delayed GE in DM

• The management of diabetic gastroparesis is guided by the severity of symptoms, the magnitude of delayed GE, and the nutritional status

• The initial options include dietary modifications, supplemental oral nutrition, and antiemetic and prokinetic medications; patients with more severe symptoms may require a venting gastrostomy or jejunostomy and/or gastric electrical stimulation

• Promising newer therapeutic approaches include ghrelin receptor agonists and selective 5-hydroxytryptamine receptor agonists

By tradition, diabetic gastroparesis has been used to describe the upper gastrointestinal (GI) manifestations of diabetes mellitus (DM). Gastroparesis is a syndrome characterized by delayed gastric emptying (GE) and upper GI symptoms that suggest, but are not associated with, gastric outlet obstruction. However, other patients with DM have GI symptoms with normal GE or abnormal GE with no or mild symptoms. Also, these GI symptoms may originate not only from the stomach but also from the small intestine (1). Hence, diabetic gastroenteropathy is a more inclusive term to describe the GI manifestations of DM. In addition to gastroparesis, diabetic gastroenteropathy also includes diabetic dyspepsia (i.e., indigestion), which is characterized by upper GI symptoms and normal, rapid, or slightly delayed GE and paucisymptomatic or asymptomatic delayed GE. The latter comprises up to 40% of patients with DM and delayed GE (2). In some patients, discriminating between dyspepsia and gastroparesis can be challenging because the symptom- and emptying-based boundaries are poorly defined (3).

Consistent with clinical practice and research studies, the studies cited in this review defined delayed GE by an objective assessment of GE with scintigraphy, a capsule, or a gastric emptying breath test (GEBT). Among these, scintigraphy is the criterion standard and the most widely used. A few studies also considered the presence of retained food in the stomach at endoscopy or during an upper GI radiograph to suggest delayed GE (4).

Although GE is the primary and most widely used objective assessment in patients with diabetic gastroenteropathy, the condition is also associated with impaired gastric accommodation and increased or reduced visceral sensation, which may coexist with delayed GE. A smaller proportion of DM patients have rapid GE.

The Epidemiology of Diabetic Gastroparesis

Prevalence and incidence

In the 20th century, that is before the routine use of intensive insulin therapy for type 1 DM, up to 60% of patients with long-standing type 1 DM and GI symptoms had diabetic gastroparesis (5). However, no studies have evaluated the prevalence of delayed GE among randomly selected people in a community, probably because scintigraphy requires specialized laboratories and entails radiation exposure. The only population-based studies on the epidemiology of diabetic and nondiabetic gastroparesis, both of which were conducted in Olmsted County, Minnesota, under the aegis of the Rochester Epidemiology Project, were based on data collected from patients who sought care rather than the population at large (4, 6) [Table 1 (6–13)]. The first study evaluated the prevalence of gastroparesis defined as definite (i.e., typical symptoms for >3 months and delayed GE documented with scintigraphy), probable (i.e., typical symptoms and food retention on endoscopy or upper GI study), and possible (i.e., typical symptoms alone or delayed GE by scintigraphy without GI symptoms) (4). The overall age-adjusted prevalence of definite gastroparesis per 100,000 persons was approximately fourfold higher in women (37.8; 95% CI, 23.3, 52.4) than in men (9.6; 95% CI, 1.8, 17.4). Twenty-one of 83 (25%) patients with definite, 39 of 127 (31%) patients with definite or probable, and 103 of 222 (46%) patients with gastroparesis by any of the three definitions had diabetic gastroparesis.

Table 1.

Community-Based Epidemiologic Studies of GI Symptoms in DM

Study	Respondents	Response Rate (No. of Respondents)	Key Findings
			Upper GI	Lower GI
Dyck et al., 1993 (7)	Residents of Rochester, MN, with DM	44% (102 DM1, 278 DM2)	Gastroparesis: 0% DM1, 1% DM2	Diarrhea: 1% DM1, 0.4% DM2
Janatuinen et al., 1993 (8)	All residents in a hospital district with DM and a randomly selected control group	92%–100% (89 DM1, 481 DM2, 635 controls)	Symptoms of nausea and vomiting were not different between cases with DM and controls without	Diarrhea not different from controls
Maleki et al., 2000 (9)	Samples of Olmsted County residents with DM1, DM2, and corresponding age- and sex-stratified controls	59% (138 DM1, 217 DM2, 388 controls)	No difference in proportions with stomach symptoms between DM and controls; less heartburn reported by DM1 patients	Constipation: 17% DM1 vs 14% controls; 10% DM2 vs 12% controls
				Diarrhea: 0% for all groups FI: 0.7% DM1 vs 1.2% controls; 4.6% DM2 vs 1.8% controls
Bytzer et al., 2001 (10), Hammer et al., 2003 (11)	Sex-stratified sample of 15,000 people in Sydney, Australia	60% for entire sample (423 of 8555 respondents had DM, 95% had DM2)	Small differences were detected with the highest adjusted OR for vomiting 1.7% vs 11% (OR, 2.71); when upper gut dysmotility symptoms were evaluated, the results were 18.2% vs 15.3% (OR, 1.75)	Diarrhea or constipation: 15.6% DM vs 10% controls
				Fecal incontinence: 2.6% DM vs 0.8% controls
Talley et al., 2002 (12)	Two surveys of patients with DM2 on mailing list of Diabetes Australia at 3-y interval	64% returned second survey (892 DM2 in first survey)	Not applicable	Similar symptom prevalence for first (second) surveys
				Abdominal pain: 7.6% (8.3%)
				Constipation: 25.7% (23.7%)
				Diarrhea: 2.6% (2.2%)
				FI: 7.2% (7.2%)
Choung et al., 2012 (6)	Follow-up of 1374 patients (269 with DM1, 409 DM2) and controls matched for age and sex (735) in Olmsted County, MN	89% (1226 patients) authorized review of medical records; questionnaires at interviews were not performed	During 10 years, gastroparesis developed in 5.2% (DM1), 1% (DM2), and 0.2% (controls); higher risk (HR, 4.4; 95% CI, 1.1, 17) in DM1 than DM2;
Aleppo et al., 2017 (13)	Multisite study of 7107 patients >26 y old with DM1 for >2 y		4.8% had a clinical diagnosis of gastroparesis; higher prevalence in women than in men (5.8% vs 3.5%)

All surveys, with the exception of Aleppo et al. (13), used a mailed questionnaire.

Abbreviations: DM1, type 1 DM; DM2, type 2 DM; FI, fecal incontinence; HR, hazards ratio.

Another community-based study from Olmsted County, Minnesota, evaluated the risk of developing gastroparesis among patients with DM in the community (6). In that study, gastroparesis was documented by physician diagnosis, scintigraphy, or symptoms, as well as by retained food at endoscopy. The cumulative incidence of symptoms and delayed GE during 10 years was 5% in type 1 DM [hazard ratio (HR), 33; 95% CI, 4.0, 274, adjusted for age and sex vs controls], 1% in type 2 DM (HR, 7.5; 95% CI, 0.8, 68, adjusted for age and sex vs controls), and 1% in controls. The risk of gastroparesis was greater in type 1 DM than in type 2 DM (HR, 4.4; 95% CI, 1.1, 17). Because the study only captured patients who presented for care, it does not include people who did not have a GE study. Hence, this study assessed the cumulative incidence of diabetic gastroparesis (during 10 years) rather than the prevalence of diabetic gastroparesis (6).

The T1D Exchange clinic registry is the largest registry of type 1 DM patients in the United States. Among the patients enrolled in this registry who were >26 years of age and had DM for at least 2 years, 4.8% received a clinical diagnosis of gastroparesis (13). The prevalence was higher in women (5.8%) than in men (3.5%). However, the criteria used to diagnose gastroparesis in this cohort are unclear.

Among studies that did not evaluate GE, the prevalence of GI symptoms is variable and not significantly different between people with diabetes and nondiabetic people. In an Olmsted County study of diabetic neuropathy, only 1% of patients with diabetes had symptoms of gastroparesis (7) (Table 1). In another Olmsted County study published 7 years later, the prevalence for nausea and/or vomiting or dyspepsia was not significantly different in patients with type 1 or type 2 DM relative to controls (9). In contrast, community-based studies from Australia, which were conducted in patients who predominantly had type 2 DM, observed that several upper GI symptoms, including abdominal pain or discomfort, early satiety, postprandial fullness, bloating, heartburn, nausea, vomiting, and dysphagia, were more common in type 2 DM than in controls (10).

Similar to functional GI disorders, there is turnover of GI symptoms in DM, reflecting appearance and disappearance of symptoms over time. Among patients with type 2 DM, GI symptoms improved in some patients but resolved in others during a 3-year period. Hence, the overall prevalence was comparable (12). Several factors other than glycemic control predicted symptom change. For example, diabetic foot problems, abnormal sweating, abdominal surgery, and depression predicted the turnover of abdominal pain. However, the specific predictive factors varied among symptoms. In another cohort of 139 patients with diabetes of whom ∼50% had type 1 DM and 5% had severe autonomic dysfunction, the turnover of symptoms ranged from 15% to 25%. This was not significantly different from controls (14). The turnover of symptoms was correlated with depression; appearance and disappearance of depression were associated with gain and loss of GI symptoms, respectively. Glycemic control and autonomic neuropathy did not influence the turnover of symptoms.

Hospitalization

Data from the Nationwide Inpatient Sample, which comprises a representative sample of 5 to 8 million hospitalizations in the United States, suggest that the incidence of gastroparesis did not change significantly between the periods 1996 to 2000 and 2001 to 2006 (15). In contrast, in another study conducted between 1995 and 2004, the hospitalizations related to gastroparesis increased by 138% and the number of patients in whom gastroparesis was the primary diagnosis for the hospitalization increased by 158% (16). The proportion of these patients who had DM increased from 21% in 1995% to 26.7% in 2004. This figure was greater than the corresponding changes in DM-related hospitalizations (+53%), all hospitalizations (+13%), and hospitalizations with four other GI conditions, that is, gastroesophageal reflux disease (GERD), gastric ulcer, gastritis, or nonspecific nausea/vomiting as the primary diagnosis (−3% to +76%). Several factors may explain the increased incidence in hospitalizations for gastroparesis. These include an increased prevalence of DM or gastroparesis, changes in diagnostic criteria, severity and/or treatment of gastroparesis, better recognition and/or diagnosis of this disorder, or a change in hospital coding practices. The temporal trends after 2000 may partially be explained by withdrawal of the prokinetic agent cisapride from the market and the humanitarian device exemption for gastric electric stimulation (GES) by the US Food and Drug Administration (FDA). For this device, patients are often hospitalized when the stimulator is placed (17). Compared with a primary diagnosis of four other GI conditions (i.e., GERD, gastric ulcer, gastritis, or nonspecific nausea/vomiting) the length of stay and total hospital charges were greater for hospitalizations in which gastroparesis was the primary ($20,573) or secondary ($24,965) diagnosis in 2004, except for gastric ulcer, for which the total charge was $23,259 (16). Similar comparative findings were observed in 1995 (16).

For the Nationwide Inpatient Sample study, data are not broken down separately for idiopathic and diabetic gastroparesis (15). In the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) consortium study, patients with gastroparesis due to type 1 DM needed more hospitalizations [i.e., 5.1 ± 6.4 per year (mean ± SD)], mostly for vomiting and dehydration, than did those with idiopathic (1.6 ± 3.0) or type 2 DM (2.7 ± 5.7) (18). In another study, among patients with DM and upper GI symptoms, patients with delayed GE spent more days in the hospital than did patients with normal GE (i.e., 25.5 vs 5.1 per 1000 patient-days) (19).

Impact on quality of life

The SF-36 physical and mental component scores are lower in patients with idiopathic and diabetic gastroparesis, indicating impaired quality of life (QOL) (20). It is uncertain whether there are differences in QOL between patients with idiopathic and diabetic gastroparesis. Overall, the impact of gastroparesis on QOL is comparable to active inflammatory bowel disease (21, 22). The physical QOL is comparable to rheumatoid arthritis, but the mental QOL is lower than in patients with rheumatoid arthritis (23). The impairment in QOL, anxiety, and depression were greater in patients with more severe abdominal pain (24). Although a cross-sectional study of 34 DM patients with GI symptoms demonstrated that QOL was reduced equally between those with normal and delayed GE (25), it is unclear whether the impaired QOL is solely explained by GI symptoms or also by comorbid conditions (e.g., depression) and/or medications (e.g., opiates).

Our understanding of the impact of gastroparesis on QOL in the community is limited. Among a cohort of people with predominantly type 2 DM in the community, the SF-36 physical and mental subscores were lower in patients with DM and GI symptoms compared with population norms (26). The reduction in QOL was more pronounced in patients who had more GI symptoms, but independent of age, sex, smoking, alcohol use, and type of DM. Surprisingly, the absolute SF-36 scores in this community-based study were approximately twofold higher than the NIDDK consortium study, which comprised patients at tertiary referral centers (24, 26). Gastroparesis is also variably associated with unemployment, lower household income, and work absenteeism (18, 27).

Mortality rates

In case series, estimated mortality rates in patients with gastroparesis range from 38% at 4 years to 34% at 25 years (28–33). Among patients with gastroparesis, in tertiary centers or the community, survival was significantly lower than in the overall population (4, 28). For example, the estimated 5-year survival adjusted for age and sex was 67% (95% CI, 60%, 75%) and 81% in gastroparesis and the overall community population, respectively (4). Survival was better in idiopathic than nonidiopathic gastroparesis (4); the latter category was predominantly comprised of patients with diabetic gastroparesis. Likewise, in a tertiary referral cohort, survival was better in postinfectious and idiopathic gastroparesis than in DM and postsurgical gastroparesis (28). The increased mortality in diabetic gastroparesis is partly explained by comorbidities associated with DM. However, the extent to which survival is affected by comorbid conditions or the severity of gastroparesis is unknown.

Pathogenesis and Pathophysiology

Several factors such as extrinsic (i.e., sympathetic and parasympathetic) and intrinsic (i.e., enteric) neuromuscular dysfunctions, hyperglycemia, and hormonal disturbances have been implicated to cause GI sensimotor dysfunctions in DM (2, 34). Neural dysfunctions have been attributed to several mechanisms (e.g., oxidative stress) described below and detailed elsewhere (35).

Normal GI motor functions

GI sensorimotor functions are primarily regulated by the intrinsic or enteric nervous system (i.e., the “little brain” in the digestive tract), which is modulated by big brain, via the extrinsic (i.e., parasympathetic and sympathetic) nervous system (Fig. 1) (36). The intrinsic and extrinsic controls are independent of each other. However, the prevertebral ganglia serve to integrate afferent and efferent impulses between the gut and the central nervous system and provide additional reflex control of the abdominal viscera. The parasympathetic nervous system inhibits the sphincters and excites nonsphincteric muscle. The sympathetic component exerts the opposite effects.

Figure 1.

Schematic representation of enteric and extrinsic mechanisms that control GI motility. Antegrade peristalsis, the motor pattern resulting in aborad propulsion, results from proximal contraction mediated by excitatory neurotransmitters, coordinated with distal relaxation mediated by inhibitory neurotransmitters. The sympathetic neural input, which is primarily mediated by presynaptic and prejunctional inhibition of acetylcholine release via adrenergic α₂ receptors, can tonically inhibit antegrade peristalsis in the viscus and stimulates tonic contraction of the sphincters (not shown).

The enteric nervous system contains 100 million neurons. It is organized in distinct ganglionated plexi, that is, the myenteric plexus, which regulates motility, and the submucous plexus, which primarily regulates absorption and secretion. The interstitial cells of Cajal (ICCs) serve as pacemakers and also convey messages from nerve to smooth muscle. DM can affect not only the gut's autonomic and enteric nervous system but also the somatic nerve supply (e.g., pudendal nerves). Derangements of the extrinsic nerves at any level may alter GI motility and secretion (37).

In the GI tract, digestion and absorption require motility, gastric and pancreatic secretion, and GI hormones, which modulate motor, secretory, and absorptive functions in the upper gut (38). These functions take place in three sequential but integrated phases, that is, cephalic, gastric, and intestinal. Liquids empty relatively quickly from the stomach in a nonlinear, generally monoexponential, manner. The GE of liquids is slower for caloric than for noncaloric liquids probably because caloric liquids evoke duodenogastric feedback mechanisms that inhibit GE (Fig. 2). By comparison to liquids, solids empty more slowly and in a biphasic manner (39, 40). During the initial lag period, which lasts between 15 and 90 minutes (average of 45 to 60 minutes), little if any solid material is emptied into the duodenum. In the lag phase the gastric antrum grinds solids into particles <2 mm in size so they can be emptied through the pylorus, whereas the fundus and body relaxes or accommodates, providing room for digestion to occur (38, 41). Thereafter, solids are emptied in a linear fashion. The GE of solids can be summarized in two phases (i.e., lag and linear) or by an exponential function (Fig. 2) (39). On average, GE is 50% and 100% complete in 2 and 4 hours, respectively (39, 40, 41).

Figure 2.

Assessment of GE by scintigraphy. Liquids empty from the stomach in an exponential manner whereas the emptying of solids is characterized by a biphasic profile that includes an initial lag phase, followed by a more rapid, linear emptying. Liquids with higher caloric content are more likely to evoke duodenogastric feedback mechanisms that inhibit GE and hence take longer to empty from the stomach than do noncaloric liquids. For solids, the lag period is prolonged by adding 10% or 25% dextrose to the meal. However, the slope of the linear phase is comparable for water and 10% and 25% dextrose. [Adapted with permission from Collins PJ, Horowitz M, Cook DJ, et al. Gastric emptying in normal subjects—a reproducible technique using a single scintillation camera and computer system. Gut 1983;24:1117–1125.]

Small intestinal digestion is accomplished by pancreatobiliary secretions and mechanical processes; absorption occurs thereafter. In the small intestine, solids and liquids are transported at approximately the same rate; the head of the column of liquid chyme reaches the cecum as early as 30 minutes after ingestion (38). Solids take longer to empty from the stomach. Hence, liquids typically arrive in the colon before solids. Assuming solids are presented in a triturated form to the small bowel, it takes ∼150 minutes for half the solid and liquid chyme of similar caloric density to traverse the small bowel (38).

Through a feedback loop, complex carbohydrates or fat in the distal small intestine inhibit proximal small intestinal motility (i.e., the small intestinal brake) (38). Chyme is transferred from the ileum to colon in intermittent boluses (42). Contents travel from the cecum to the rectum in an average time of 36 hours, with an upper limit of 65 hours (43–45). Hence, compared with the stomach and small intestine, colonic transit is relatively prolonged. This allows digestion (i.e., fiber) and absorption (i.e., of water and electrolytes) to be completed.

Autonomic neuropathy

Autonomic neuropathy has been implicated as the major mechanism for DM gastroenteropathy (46). More recently, there is emerging evidence for disturbances in the enteric nervous system (i.e., enteric neurons and ICCs) and smooth muscle myopathy.

In DM patients, and in spontaneously diabetic rats, there are fewer cells in motor vagal and sensory sympathetic ganglia and structural changes (e.g., segmental demyelination and axonal degeneration) of vagal nerve fibers both within myenteric and submucosal plexi and outside of the GI tract (46). The loss of nerve fibers is often multifocal, suggestive of ischemic injury.

In humans, autonomic nervous function can be detected by assessing GI (i.e., rise in plasma pancreatic polypeptide in response to sham feeding), cardiovascular, or sudomotor autonomic functions (47, 48). GI vagal dysfunction is manifested by a reduced increment in plasma pancreatic polypeptide levels during sham feeding (49, 50). Although a subnormal pancreatic polypeptide response is associated with cardiovascular vagal dysfunction, a threshold increase of 20 pg/mL was 100% specific but only 46% sensitive for cardiac vagal dysfunction (51, 52). Also, hyperglycemia is associated with greater fasting levels of plasma pancreatic polypeptide and a reduced response to sham feeding, which affect the interpretation of the test in DM. Alternative approaches to evaluate GI vagal function (e.g., effect of sham feeding on gastric acid secretion or insulin-induced hypoglycemia on plasma pancreatic polypeptide) are seldom used in clinical practice (53, 54).

Based on the premise that vagal neuropathy is generally a length-dependent axonopathy (55), impaired cardiovascular vagal responses, such as reduced heart rate responses to deep breathing and the Valsalva maneuver, are used as surrogate markers of GI vagal dysfunction. Diabetic gastroenteropathy is also associated with adrenergic and sudomotor dysfunction (56, 57). Autonomic dysfunctions are correlated with GI dysmotility (56) and cardiovascular vagal dysfunction is associated with delayed GE in some (58, 59), but not all, studies (1). Likewise, autonomic dysfunctions were associated with symptoms in some (1), but not all, studies (57).

Enteric neuromuscular disturbances

Enteric neuropathologic abnormalities have been studied in humans and in animal models of DM, which include genetically modified, streptozotocin-induced and high-fat diet–induced mouse models (46, 60). Some features of the peripheral neuropathologic injury in these animal models are comparable to diabetic neuropathy in humans (61).

In the enteric nervous system, the primary abnormalities, which vary in severity among the models and have also been described in humans, include loss of enteric neurons and/or ICCs; smooth muscle disturbances have also been described (46, 62–64). Several mechanisms, including apoptosis, oxidative stress, advanced glycation end products, and neuroimmune mechanisms may be responsible for neuronal loss and gut dysmotility (35). In the largest assessment of full-thickness gastric biopsies from humans with diabetic gastroparesis, there were three main findings: loss of ICCs, loss of enteric nerves, and an immune infiltrate (65). The reductions in enteric nerves and ICCs were correlated. The number of ICCs was correlated with faster GE at 4 hours. These findings are consistent with the hypothesis that loss of nerves and/or ICCs contributes to delayed GE. However, compared with controls (i.e., 5.6 ICC bodies per field), ICCs were also reduced in patients with chronic vomiting and normal GE (i.e., 3.5 bodies per field), albeit to a lesser extent than in patents with gastroparesis (2.3 bodies per field) (66). The mechanisms for ICC loss are unclear. In murine models, reduced insulin/IGF-1 signaling in DM may deplete ICCs via smooth-muscle atrophy and reduced stem cell factor production (63). Insulin and IGFs prevent the loss of ICCs in tissue culture. Because ICCs do not express receptors for insulin or IGF-1, these effects are perhaps mediated by insulin-mediated secretion of stem cell factor from smooth muscle, which is the most important growth factor for ICCs, rather than directly by insulin and IGFs (34).

In some models, there is a preferential loss of inhibitory neurons that express neuronal nitric oxide synthase. Conceptually, this reduction in nitrergic inhibitory innervation may contribute to impaired gastric accommodation and accelerated intestinal transit in DM (67). Because nitric oxide effects pyloric relaxation, loss of neuronal nitric oxide synthase may impair pyloric relaxation, resulting in pylorospasm in humans with diabetic gastroparesis (46, 68). Insulin also restored the expression of neuronal nitric oxide synthase and GE in animal models of DM (63, 69).

There is considerable evidence that interactions between the innate immune system, specifically macrophages, and ICCs and the neuromuscular apparatus may mediate gastroparesis in nonobese diabetic mice and in humans. Shortly after developing DM, nonobese diabetic mice, which are a model for type 1 DM, have rapid GE (70). Within 2 weeks, GE normalizes and, thereafter, a subset developed delayed GE. The gastric tunica muscularis in mice with normal GE is populated with alternatively activated M2 macrophages that express cytoprotective markers including heme oxygenase-1. In the mice that develop delayed GE, the muscularis propria macrophages are predominantly classically activated M1 macrophages, which produce cytokines that result in ICC loss. Normally, hyperglycemia-induced oxidative stress increases expression of heme oxygenase-1, which degrades heme to generate carbon monoxide. Carbon monoxide has cytoprotective actions and relaxes smooth muscle (71). In nonobese diabetic mice, failure to upregulate expression of heme oxygenase-1 leads to loss of ICCs and delayed GE whereas hemin administration induced heme oxygenase-1 and restored GE (70, 72). Hemin did not improve GE in humans with diabetic gastroenteropathy, possibly because the dose was inadequate or because these patients had irreversible disease (73). Further studies have determined that macrophages are actually required for the development of delayed GE in diabetic mice (74). In the streptozotocin-induced diabetic mouse model, the GE status correlates with the balance between conventionally activated macrophages, expressing proinflammatory cytokines, and alternatively activated macrophages, expressing M2 macrophage markers (75).

Pathophysiology of diabetic gastroenteropathy: in vivo studies in humans

In humans, most attention has centered on delayed GE and its relationship to acute hyperglycemia and autonomic neuropathy (2), which is often associated with other complications (i.e., peripheral neuropathy, retinopathy, and nephropathy) (76). Delayed GE may result from reduced antral contractility, which impairs the ability of the antrum to triturate or grind food into smaller particles that can be emptied from the stomach, and/or pylorospasm, hindering GE (56, 68, 77). Both disturbances (i.e., antral hypomotility and pylorospasm) may result from a vagal neuropathy or enteric neuromuscular dysfunction.

Other patients with DM have rapid GE. Earlier studies suggested that delayed GE, usually of solids, affected patients with longstanding complicated type 1 DM whereas rapid GE, typically for liquids, occurred in patients with early type 2 DM without autonomic dysfunction (78–82); however, these studies were restricted either to patients with rapid or slow GE. In a cohort of 129 consecutive patients with DM and upper GI symptoms who underwent scintigraphy, 55 (42%) had normal, 46 (36%) had delayed, and 28 (22%) patients had rapid GE (83). In each GE category, the number of patients with type 1 DM and type 2 DM was approximately equal. Although a neuropathy was associated with rapid GE, symptoms and other characteristics of the DM phenotype (i.e., type and duration of DM, glycosylated hemoglobin levels, and extraintestinal complications) did not discriminate between normal and delayed or rapid GE. Conceivably, impaired gastric accommodation, perhaps caused by a vagal neuropathy (84), may increase gastric pressure and predispose to accelerated GE of liquids. However, the relationship between rapid GE and impaired gastric accommodation has not been substantiated.

In a study of 18 DM patients with GI symptoms and 13 asymptomatic controls, the patients had impaired postprandial gastric accommodation after a liquid meal (67). Because the vagus nerve mediates postprandial accommodation, impaired accommodation may be due to vagal neuropathy. However, the relationship between vagal neuropathy and impaired postprandial accommodation is unclear because accommodation may be preserved even in patients with diabetes with vagal neuropathy (85), perhaps because nonvagal adaptive mechanisms, involving enteric neurons, restore accommodation (86).

Perception of GI distention may be reduced (87–89) or increased in diabetic gastroenteropathy (2). In a study that compared cerebral-evoked potentials after esophageal stimulation in 10 patients with diabetes with polyneuropathy and 10 controls, decreased GI perception was associated with an autonomic neuropathy (87). Conversely, in another study, painful diabetic neuropathy was associated with more severe autonomic dysfunction than painless neuropathy (91). Likewise, a comparison of 40 patients with diabetic gastroenteropathy and 24 controls observed that 78% of patients had a cardiovascular vagal or adrenergic neuropathy, which was associated with more severe symptoms during the GE study (60). These findings suggest that, similar to peripheral neuropathy, autonomic nervous dysfunction is associated not only with sensory loss (i.e., negative symptoms or “numbness” as in peripheral neuropathy) but also increased symptoms (e.g., pain) (92). Similar to peripheral nerves, improvement in GI symptoms may signify progression of neuropathy with worsening of sensory function. In addition to autonomic dysfunctions, several studies have observed altered brain processing and networks, as evidenced by electroencephalography or diffusion tensor imaging in the resting state (93, 94) and with measurement of evoked potentials in response to electrical stimulation of the esophagus (95) and anorectum (96). Likewise, studies with MRI demonstrate central changes in patients with painful diabetic neuropathy (97). It is unclear to what extent, if any, these changes are secondary to peripheral dysfunctions.

Some patients with DM and gastroparesis also have small intestinal dysmotility, which is generally characterized by reduced, rather than by increased, motility (98). Symptoms may originate from the small intestine or secondary to gastric stasis resulting from small bowel dysmotility (98). Approximately 10% of asymptomatic healthy controls reported severe GI symptoms during duodenal lipid infusion (300 kcal during 2 hours) (99). In comparison, 27% of DM patients with normal, 38% with delayed, and 20% with rapid GE reported severe symptoms during the intestinal lipid infusion (99). Hence, 31 of 40 (78%) patients had abnormal GE and/or increased perception based on symptoms during enteral nutrient infusion. Moreover, as detailed in the next section, the severity of symptoms during a GE study and during intestinal lipid perfusion predicted the severity of daily symptoms.

Relationship between GI dysfunctions and daily symptoms

Some studies have questioned the contribution of delayed GE to daily symptoms because many patients with delayed GE are asymptomatic (100) and because symptom profiles are similar among patients with DM and GI symptoms who exhibit normal or delayed GE (25). Also, in therapeutic trials of prokinetic agents for gastroparesis, the improvements in GE and symptoms were not correlated (101). However, these findings may be partly explained by limitations of the techniques used to measure GE in these studies. A meta-analysis of cross-sectional studies that evaluated the association between symptoms and delayed GE identified only two studies in DM patients in which GE was evaluated with optimal techniques (100). In that meta-analysis, early satiety and fullness were associated with delayed GE in DM (odds ratio [OR], 2.0; 95% CI, 1.0, 4.1); however, no other symptoms (i.e., nausea, vomiting, abdominal pain, or bloating) were associated with delayed GE. In comparison, among all patients (i.e., DM and non-DM), all of these symptoms were associated with delayed GE.

Why delayed GE in individuals with DM can be asymptomatic is unclear. One possible explanation is afferent vagal dysfunction (87). Supporting that hypothesis, hyperglycemia (i.e., a greater HbA_1c) was correlated with less severe symptoms during a GE study and during intestinal lipid perfusion (60, 87). Similar discrepancies between symptoms and objective findings exist in other diseases (e.g., Crohn disease) (102).

“The cardinal symptoms of gastroparesis include early satiety, postprandial fullness, nausea, vomiting, bloating, upper abdominal pain, and weight loss.”

In comparison with GE per se, the symptoms during a GE study are more strongly correlated with daily symptoms. In a cohort of 40 patients with DM gastroenteropathy and 24 controls, 55% of patients had normal, 33% had delayed, and 12% had rapid GE (60). Of patients with abnormal GE, 62% with delayed GE and 60% with rapid GE had severe symptoms during the GE study. The GE and symptoms during a GE study independently explained 40% of the variation in daily symptom severity and 32% of the variation in QOL among patients with DM. These findings support an earlier study in 388 dyspeptic patients, of whom 73 had DM, in which the severity of symptoms during a GE study was correlated with daily symptoms (103). Symptoms during a GE study were more useful for predicting daily symptoms than were the symptoms during duodenal lipid infusion (60).

Depression

Patients with functional GI disorders (e.g., irritable bowel syndrome, functional dyspepsia) often have anxiety and depression, which, presumably via effects on the gut–brain axis, can interact with biologic and social factors to shape these disorders and their symptoms (104). A systematic review of 42 studies observed that DM is associated with depression; the prevalence of depression among patients with DM (i.e., 9% to 27%) is approximately twofold greater than in individuals without DM (105). The prevalence of comorbid depression was significantly higher in women (28%) than in men (18%).

The severity of DM and depression go hand in hand. In a meta-analysis of 24 studies, the severity of depression was associated with the magnitude of hyperglycemia in type 1 and type 2 DM (106). Another meta-analysis of 27 studies observed that the severity of depression was correlated with the prevalence of complications of DM (i.e., retinopathy, nephropathy, neuropathy, macrovascular complications, and sexual dysfunction) (107). Among 200,936 patients with depression, 74,160 (36.9%) had DM, including 57,418 (28.6%) with complications (108). Among depressed patients, the incidence of serious psychiatric outcomes was greater in patients with DM (6.7%) than without DM (3.3%). Patients with depression with microvascular and macrovascular complications were more likely to experience serious outcomes than were those without complications of DM (OR, 2.2; 95% CI, 2.07, 2.34). In the NIDDK Gastroparesis Consortium study, nearly 50% of patients each had clinically significant anxiety and moderate to severe depression (109). A systematic review suggested a slightly lower prevalence of anxiety and depression in patients with gastroparesis but highlighted the numerous screening tools and cutoff values used to evaluate the multiple cohorts (20). However, moderate to severe depression was associated with more severe symptoms (20). After 48 weeks of treatment, GI symptoms were more likely to improve in patients treated with antidepressants and less likely to improve in patients with moderate to severe depression at baseline (110). Taken together, these findings underscore that depression is common, often severe, and may contribute to the expression of GI sensorimotor dysfunctions in DM.

Bidirectional relationship between control of glycemia and diabetic gastroparesis

There is a bidirectional relationship between glycemic control and GE in DM, that is, hyperglycemia can delay GE wheras GE disturbances affect glycemic control (and perhaps variability) (90). The effects of hyperglycemia on GE may be influenced by the duration of hyperglycemia. Hence, it is useful to characterize glycemic exposure as follows: immediate (i.e., fasting glucose before the study), recent (i.e., HbA_1c before the GE study), and remote (e.g., HbA_1c 20 years ago). In the context of DM, the term “immediate” is more appropriate than “acute” because DM patients have longstanding hyperglycemia.

As reviewed elsewhere (90), acute hyperglycemia delayed GE in healthy people (111, 112). Two studies, with a total of 18 patients, observed that compared with euglycemia (blood glucose of 4 to 8 mmol/L), marked hyperglycemia (blood glucose of 16 to 20 mmol/L) prolonged the GE half-time of solids in type 1 DM by an average of 17 minutes (increased from 124 to 141 minutes) and 31 minutes (increased from 74 to 105 minutes) (113, 114). In one study, hyperglycemia also prolonged the GE of nutrient liquids (113). In another study, mild hyperglycemia (8 mmol/L vs 4 mmol/L) was associated with greater gastric retention of solids and liquids in eight healthy people and nine patients with type 1 DM (115). However, the difference in gastric retention at 100 minutes was relatively small (i.e., 36% at 4 mmol/L vs 44% at 8 mmol/L). Moreover, in this study and others that evaluated the effects of severe acute hyperglycemia on GE (113, 114), the effects of normoglycemia and hyperglycemia were assessed on 2 separate days. The intraindividual day-to-day variability in GE during normoglycemia is unclear. Also, the fasting blood glucose immediately before the GE study was not associated with the GE rate in patients with type 1 DM from the NIDDK consortium (116), the Epidemiology of Diabetes and Intensive Complications (EDIC) study (58), and at Mayo Clinic (60). Indeed, among patients with poorly controlled type 2 DM, a higher fasting blood glucose was correlated with faster GE (59), which is consistent with findings from a mouse model (61). Taken together, these findings suggest that the effects of acute hyperglycemia on GE are relatively modest.

Hyperglycemia may delay GE by decreasing gastric motility. Compared with euglycemia (blood glucose of 5 to 8 mmol/L), hyperglycemia (blood glucose ≥10 mmol/L) reduced antral motility in healthy people (117). More severe hyperglycemia (blood glucose of 16 to 19 mmol/L) reduced antral motility and GE in eight patients with type 1 DM (114). Antral contractility triturates food into smaller particles. Hence, it is conceivable that hyperglycemia-induced inhibition of antral motility can delay GE. However, in this study, the correlation between antral motility and GE of solids was not significant (114). Hyperglycemia also increased gastric compliance during distention in healthy people (118) and in people with type 1 DM and reduced duodenal pressure activity, which may facilitate GE (84). Finally, hyperglycemia [mean blood glucose of 15.9 mmol/L (range, 10.7 to 17.5)] is also accompanied by a shorter length (mean of 88 minutes vs 158 minutes) of the migrating motor complex cycle in healthy people (119).

Two studies of 18 patients with type 1 DM showed that hyperglycemia (15 mmol/L) increased proximal gastric compliance during gastric distention and the perception of fullness, nausea, and bloating during distention relative to euglycemia (6 mmol/L) (84, 120). Relative to euglycemia (blood glucose of 6 mmol/L), the perception of nausea and fullness and gastric relaxation induced by duodenal infusion of triglycerides was greater during hyperglycemia (blood glucose of approximately 15 mmol/L) (118). Although these studies in patients with type 1 DM suggest that hyperglycemia increased gastric accommodation, patients with DM have impaired postprandial gastric accommodation, possibly due to a vagal neuropathy, which is postulated to explain early satiety (67, 121). Although increased perception may explain more severe symptoms (e.g., nausea) during hyperglycemia, greater postprandial relaxation would not. Besides hyperglycemia, normoglycemic hyperinsulinemia is also associated with reduced postprandial accommodation and increased perception of gastric distention (122).

HbA_1c levels reflect the magnitude of glycemic exposure during the past 3 months. There are conflicting data for the relationship between HbA_1c and GI manifestations of DM. Cross-sectional studies suggest that higher glycosylated hemoglobin concentrations are associated with a higher prevalence of GI symptoms and slower GE among people with DM in the community and in outpatient clinics (10, 123). However, among 78 randomly selected participants with type 1 DM in the EDIC cohort, HbA_1c (mean, 7.7%) just before the GE study was not associated with GE. However, several parameters of remote glycemic exposure were independently associated with delayed GE, that is, the HbA_1c at entry into the Diabetes Control and Complications Trial (DCCT) ∼27 years previously (OR, 1.6; 95% CI, 1.1, 2.3), the duration of DM (OR, 1.2; 95% CI, 1.01, 1.3) before DCCT entry, and mean HbA_1c over 27 years during DCCT-EDIC (OR, 2.2; 95% CI, 1.04, 4.5) (58).

Strict glycemic control improves neural, renal, and retinal disturbances in type 1 and, to a lesser extent, in type 2 DM (124, 125). However, improved glycemic control did not improve GE 1 week later in 10 patients with type 2 DM (126). Likewise, overnight and sustained (6-month) improvements in glycemic control did not affect GE in 30 patients with poorly controlled DM (59). This is in contrast to the findings of the DCCT, in which intensive insulin therapy for 6.5 years reduced the risk of diabetic nephropathy, retinopathy, and peripheral and cardiac autonomic neuropathy by 40% to 60% compared with conventional insulin therapy (124). These differences between the former intensive and conventional treatment groups persist for as long as 14 years despite the loss of glycemic separation between groups (127–129). Perhaps these differences are explained by epigenetic changes secondary to hyperglycemia (130, 131).

In addition to hyperglycemia, electrolyte imbalances due to diabetic ketoacidosis (e.g., hypokalemia) and uremia may also aggravate impaired motor function in DM patients. Amylin analogs such as pramlintide (132) or glucagon-like peptide-1 (GLP-1) receptor agonists (such as liraglutide and exenatide) may cause iatrogenic gastroparesis (133–135). Similarly, medications used to treat complications of DM such as tricyclic antidepressants, which exhibit anticholinergic properties, and opioids may delay GE (136–141).

Normally the stomach empties nutrients into the small intestine at a tightly regulated rate averaging 2 to 3 kcal/min (38, 142). The GE rate is correlated with the postprandial increase in blood glucose concentration in healthy people and in those with DM (90). Indeed, pharmacologically mediated delayed GE partly explains the efficacy of GLP-1 analogs in type 2 DM whereas acceleration of GE increased postprandial glycemia in type 1 DM (143) and type 2 DM (144, 145). Hence, it seems plausible that delayed GE may contribute to hypoglycemia in DM (146). However, the evidence for an association between delayed GE and hypoglycemia in DM is limited. Two studies have evaluated the relationship between innate (i.e., not pharmacologically mediated) delayed GE and glycemia in type 1 DM. In a small cross-sectional study, postprandial insulin requirements were lower in patients with type 1 DM with than without gastroparesis (147). Perhaps this suggests that gastroparesis predisposes to hypoglycemia, particularly early in the postprandial period (so-called gastric hypoglycemia) (146). These observations predated the use of insulin analogs and widespread use of insulin pumps for DM. An alternative explanation is that shared risk factors, such as autonomic neuropathy, predispose to hypoglycemia in patients with gastroparesis. In a clinical trial of 30 patients with moderately controlled type 1 DM on an insulin pump and minimal GI symptoms, delayed GE was associated with lower glycemia, but not hypoglycemia, measured with a continuous glucose monitoring (CGM) device (148). In contrast, delayed GE was associated with higher glucose values during the entire day. This glucose increment associated with delayed GE is, at a minimum, equivalent to that requiring 15% more insulin. Further studies are necessary to ascertain the impact of GE disturbances on guiding therapy in DM.

“Similar to CVS, cannabinoid hyperemesis syndrome is also characterized by stereotypical episodic vomiting.”

Clinical Features

The cardinal symptoms of gastroparesis include early satiety, postprandial fullness, nausea, vomiting, bloating, upper abdominal pain, and weight loss (2). In the NIDDK Gastroparesis Registry, symptoms begin at a later age in patients with type 2 DM (49 ± 11 years) than in patients with type 1 DM (34 ± 10 years) or idiopathic gastroparesis (36 ± 15 years) (18). In ∼50% of patients with gastroparesis (both idiopathic and diabetic), the symptoms began acutely (18). Approximately 19% of patients with idiopathic and 14% each with type 1 and type 2 DM had features of an infectious prodrome (18, 149). In almost half of the patients, the symptoms waxed and waned over time; ∼33% of patients with idiopathic and diabetic gastroparesis had periodic exacerbations and 10% had a cyclical pattern.

The average duration of symptoms prior to diagnosis of gastroparesis was 6 years in type 1 DM and 4 years in type 2 DM (18). The evaluation was generally prompted by vomiting in diabetic gastroparesis and abdominal pain in idiopathic gastroparesis. In this and another study (150), retching and vomiting were more severe in diabetic than in idiopathic gastroparesis. In contrast, patients with idiopathic gastroparesis had more severe early satiety and postprandial fullness.

Among patients with gastroparesis, caloric intake is typically reduced, causing significant weight loss. In the NIDDK consortium registry, patients with diabetic gastroparesis consumed, on average, ∼1280 kcal daily (151). Approximately 60% of patients with diabetic gastroparesis consumed a calorie-deficient diet (i.e., an intake of <60% of the estimated total energy requirements) (151) and half reported weight loss (18). Indeed, in one study of 129 patients, weight loss >4.5 kg predicted delayed GE (4). Patients with more severe GI symptoms (pain, fullness, bloating, and constipation) are more likely to consume fewer calories. Only a minority of patients supplemented their dietary intake of vitamins and minerals. Despite this, 51% of patients with type 1 DM and 91% with type 2 DM are overweight (18). Overestimation of energy requirements in individuals with obesity (151) or lower energy expenditure in patients with gastroparesis (152) may explain this paradox. Perhaps ingestion of sweet, salty, and starchy foods, which are better tolerated by patients with gastroparesis, may also contribute to obesity in these patients (153).

As observed by Kassander (154) in 1958 and subsequently documented by others, many patients, indeed 60% in his report, with DM and delayed GE are asymptomatic; hence, delayed GE is probably “more often overlooked than diagnosed” (155, 156). The lack of symptoms may partly explain why GE disturbances are underrecognized among patients with delayed GE and the limited correlation between delayed GE and symptoms. A meta-analysis of studies that evaluated the association between symptoms and delayed GE identified two studies in which GE was evaluated with optimal techniques (100). Early satiety and fullness were associated (OR, 2.0; 95% CI, 1.0, 4.1) with delayed GE in DM. However, no other symptoms (i.e., nausea, vomiting, abdominal pain, and bloating) were associated with delayed GE. In comparison, among all patients (i.e., DM and non-DM), the association between symptoms and delayed GE were stronger; all of these symptoms were associated with delayed GE. Among asymptomatic patients, erratic glycemic control, especially early postprandial hypoglycemia, may be the only sign of gastroparesis (157–159). However, establishing a diagnosis of gastroparesis does not necessarily rectify these excursions; erratic glycemic control occurs in 50% of patients with known gastroparesis (18).

Diagnosis

Among patients with DM and upper GI symptoms, a diagnosis of gastroparesis is established by documenting delayed GE and excluding gastric outlet obstruction, preferably with endoscopy. It is necessary to assess GE because symptoms alone cannot predict whether patients have normal, rapid, or delayed GE (83). Medications that may delay (e.g., opioid analgesics, anticholinergic agents, GLP-1 analogs, and pramlintide) or accelerate (e.g., metoclopramide, domperidone, and erythromycin) GE should be discontinued before the test; for most medications, 48 to 72 hours is sufficient. Although guidelines recommend that the GE should not be assessed during severe hyperglycemia (blood glucose ≥275 mg/dL) (160), there is limited evidence that the fasting blood glucose prior to the GE study is associated with delayed GE (58–60). Indeed, hyperglycemia was associated with faster GE in patients with poorly controlled type 2 DM (59).

GE can be evaluated with scintigraphy (Fig. 2), which is the criterion standard test, GEBT, or with wireless capsule motility (161) (Table 2). The assessment of GE with ultrasound and MRI is limited to research studies. Typically, the GE of solids is evaluated with a meal containing eggs labeled with ^99mTc sulfur colloid; images are acquired at baseline and 1, 2, and 4 hours. Postprandial scans at 1 and 2 hours can identify accelerated GE, whereas scans at 2 and 4 hours identify delayed GE (162). There are two commonly used meals, with normal or low fat content, which in turn influences the normal values of GE. The regular fat meal (296 kcal, 31% protein, 32% carbohydrate, and 37% fat) consists of two scrambled eggs, one slice of whole wheat bread, and one glass of skim milk (163). GE is faster in men than in women; sex-appropriate normal values are available for this meal. At 4 hours, >23% retention (or <76% emptying) in men and >24% retention in women reflects delayed GE. For the GE half-time, the intraindividual variability in 60 patients with upper GI symptoms was 20% (164).

Table 2.

Noninvasive Measurements of GE

	Gastric Emptying by Scintigraphy	Stable Isotope Breath Test	Wireless Pressure and pH Capsule
Measurements	Gastric emptying	Gastric emptying	Gastric emptying and pressure amplitude
Apparatus	External gamma camera and isotope-labeled meal	Breath collection vials and stable isotope-labeled meal	Intraluminal capsule with miniaturized strain gauge and pH measurement
Study duration	Generally 4 h; longer when small bowel and colon transit are also measured	3–4 h	6 h; longer when small bowel and colon transit are also measured
Performance and interpretation	Excellent. Meals, data acquisition, and interpretation are standardized	Excellent. Meals, data acquisition, and interpretation are standardized for the commercially available 13C spirulina GEBT	Meals, data acquisition, and interpretation are standardized. The sensitivity and specificity for diagnosing gastroparesis are 65% and 87%, respectively. Compared with controls, gastric pressures are lower in diabetic but not idiopathic gastroparesis

The Society of Nuclear Medicine and the American Neurogastroenterology and Motility Society recommend a 4-hour test using radiolabeled egg white meal with jam, toast, and water (160). This meal, which contains 255 kcal, with 72% carbohydrate, 24% protein, 2% fat, and 2% fiber, is used by studies performed by the NIDDK gastroparesis research consortium. For this meal, delayed GE is defined as >60% retention at 2 hours and/or 10% at 4 hours. Moderate to severe delayed GE has been defined as >20% retention at 4 hours (110). Unfortunately, many centers perform scans for 2 hours only, which is suboptimal.

The radiation dosimetry for these tests is dependent on the meal used, the patient’s sex, and GI transit time (165). In individuals with normal intestinal transit, the ^99mTc–low-fat egg white sandwich meal, as endorsed by many societies (160), is estimated to expose the stomach wall to 1 mSv of radiation in men and 1.23 mSv in women. The radiation exposure is greater in individuals with delayed transit and to the colon than the stomach. In comparison, in 2009 the average effective dose from natural and medical sources of radiation was 3.1 and 3.0 mSv, respectively (166).

The GEBT employs a meal containing a substrate, either octanoic acid or the blue-green algae Spirulina platensis, labeled with ¹³C, which is a stable isotope. The meal is emptied from the stomach, absorbed in the small intestine, and catabolized in the liver, from where it enters the body’s bicarbonate pool, and is then excreted as ¹³CO₂ in the breath. The exhaled ¹³C is detected by mass spectrometry. The rate-limiting step in this process is the stomach emptying rate. The ¹³C spirulina GEBT (238-kcal meal) has been validated against scintigraphy (167, 168), approved by the FDA, and used in several clinical trials in diabetic gastroparesis (73, 169, 170). This test had a sensitivity of 89% and specificity of 80% for identifying delayed GE. The nondigestible, wireless transmitting capsule (Medtronic) samples and transmits pH, pressure, and temperature data at regular intervals to a portable receiver worn by the patient (171). Passage of the capsule from the antrum through the pylorus into the duodenum is identifiable by an abrupt change from an acid gastric pH to an alkaline duodenal pH associated with a burst of contractions. In the pivotal study, the GE time measured by this capsule and scintigraphy were correlated (171). However, in 6 of 87 healthy patients, the capsule did not empty from the stomach in 5 hours, which is the upper limit of normal. The sensitivity and specificity for diagnosing gastroparesis are 65% and 87%, respectively. Similar to scintigraphy, the wireless capsule has the advantage of being able to evaluate small bowel and colonic transit times in a single examination.

At endoscopy, some patients with delayed GE have retained food in the stomach after an overnight fast. Anecdotally, some such patients have normal GE by scintigraphy. Possible explanations for this discrepancy include ingestion of food before an endoscopy, day-to-day variations in GE, the use of medications (e.g., opioids) that can delay GE, or differences between the gastric motor mechanisms responsible for antral motility and emptying of smaller particles during scintigraphy (i.e., type 2 antral motor activity) and indigestible larger particles (i.e., ≥3 mm size) ingested with meals that are emptied by the antral component of the migrating motor complex during fasting or sleeping. Some patients have a gastric bezoar, which is most commonly encountered with DM and may be present at diagnosis (172). In one series of 50 patients with upper GI symptoms and gastroparesis diagnosed by upper GI barium study, a bezoar was identified in 6% of patients (173). However, in that series, only 13 of the 20 (65%) patients who had a solid-meal GE scintigraphy demonstrated delayed emptying. Overall, the sensitivity of a bezoar for the diagnosis of gastroparesis is unclear.

Assessments of pattern or timing of gastric slow waves, which are generated and propagated by the ICCs and then conducted into the adjacent smooth muscle cells, are also used to diagnose gastric motility disorders (174). Gastric slow waves originate in the proximal body of the stomach and migrate to the distal antrum with a regular rhythm of ∼3 cycles per minute (cpm). With electrogastrography, the gastric electrical rhythm is recorded percutaneously. Spectral analysis is used to identify the dominant frequency and electrical activity that is within, below (i.e., bradygastria), or above (i.e., tachygastria) the normal range.

The utility of traditional electrogastrography for identifying gastric motility disorders is markedly limited by several factors. Compared with cardiac myoelectricity, cutaneously recorded gastric slow waves have a very weak amplitude, necessitating an amplifier. Electrical activity from other organs (e.g., heart, respiration, small intestine, and colon) and even body movements need to be discriminated from gastric electrical activity, necessitating the use of filters (175). Even total gastrectomy patients display a 3-cpm rhythm, which is thought to originate from the colon (176). The methods for electrogastrography (i.e., electrode positions, recording periods, test meals, analytic software, filters, and normal reference values) are not standardized (177). The amplitude of cutaneous recordings is an imprecise index of the strength of contraction. Indeed, distention of the atonic canine stomach with a balloon increased the amplitude of electrogastrography signals even when gastric contraction was completely abolished (178). Parameters of interest include the dominant frequency and power, percentage normal rhythm, percentage bradygastria, percentage tachygastria, instability coefficient, and power ratio. The range of normal frequencies is ∼2 to 4 cpm, but it varies considerably among studies (174). Furthermore, although bradygastrias and tachygastrias are considered abnormal, they also occur in healthy people (179). Hence, electrogastrography is predominantly used as a research rather than a clinical tool (174, 180).

In contrast to transabdominal electrogastrography, high-resolution electrical mapping is performed by placing an array of numerous (e.g., 256) closely spaced electrodes on the outer surface of the stomach with a laparotomy under general anesthesia, generally in patients undergoing surgery for another indication (e.g., placement of gastric electrical stimulator) (175, 181). In one study, patients with gastroparesis had markedly abnormal initiation and/or propagation of gastric electrical activity (181). At present, the use of this invasive technique is limited to research studies.

Differential diagnosis

By definition, the symptoms of gastroparesis are similar to those of benign or malignant gastric outlet obstruction (182), which can be readily excluded with an upper endoscopy. Gastroparesis can be readily differentiated from most other organic causes of upper GI symptoms (e.g., acute cholecystitis) (183).

There is a bidirectional relationship between DM and pancreatic cancer (i.e., adenocarcinoma) (184). Longstanding type 2 DM is a risk factor for pancreatic cancer. Conversely, robust epidemiologic studies suggest that DM may be a manifestation of overt or occult pancreatic cancer. Indeed, ∼1% of people with new-onset DM >50 years of age are diagnosed with pancreatic cancer within 3 years (185); patients are hyperglycemic for a mean duration of 30 to 36 months before their cancer diagnosis (186). Similar findings were reported in US veterans (187). Hence, clinicians should meticulously inquire about (especially meal-related) GI symptoms in patients with newly diagnosed DM and should have a low threshold for investigating these symptoms.

DM is also associated with chronic intestinal pseudo-obstruction, a condition resulting in abnormal small intestinal motility and dilatation (188). When the stomach is involved in this process, delayed GE may occur (189). Abdominal imaging and/or small bowel manometry to identify chronic intestinal pseudo-obstruction should be performed, where appropriate.

Differentiating gastroparesis from other nonorganic GI disorders associated with nausea and vomiting can be more challenging (190). With the exception of weight loss, which is more common in patients with delayed GE, upper GI symptoms do not differentiate between patients with diabetes with normal delayed or rapid GE (25, 83, 191). Likewise, in the NIDDK gastroparesis cohort of patients with diabetes and nondiabetic patients with chronic nausea and vomiting, the patients with normal and delayed GE were virtually indistinguishable (192). A careful history is essential to differentiate regurgitation, which occurs in GERD and in the rumination syndrome, from vomiting, which occurs in gastroparesis (193). This distinction is critical because regurgitation and rumination are simply and effectively managed with a behavioral intervention (i.e., diaphragmatic breathing). In contrast to vomiting, which entails forceful retching, regurgitation and rumination are characterized by the repetitive, effortless regurgitation of recently ingested food into the mouth followed by rechewing and reswallowing or expectoration of food. Not infrequently, rumination is a habit, often initiated by a belch, a swallow, or by stimulation of the palate with the tongue. Contraction of the abdominal muscle with relaxation of the lower esophageal sphincter causes regurgitation, which typically, occurs within 15 minutes of starting a meal. Vomiting from gastroparesis occurs later in the postprandial period. In some patients, the distinction of vomiting from regurgitation can be difficult for two reasons. First, delayed GE in gastroparesis may predispose to regurgitation; hence, vomiting and regurgitation may coexist. Second, the initial effortless regurgitation of food is followed by vomiting thereafter.

Anorexia and bulimia, which are psychiatric disorders, can also present with similar features (193). Anorexia nervosa primarily occurs in adolescent and young adult women. Patients with anorexia have a distorted body image and fear of obesity. They engage in compulsive dieting and self-imposed starvation to maintain a profoundly low body weight, which in turn can delay GE. GI symptoms (e.g., anorexia, early satiety, nausea, and vomiting) are common. Refeeding and restoration of normal body weight can normalize both GI symptoms and GE. Patients with bulimia nervosa have recurrent episodes of binge eating often followed by self-induced vomiting, strict dieting or fasting, the use of laxatives or diuretics, or vigorous exercise to prevent weight gain. In bulimia, GE may be normal, rapid, or delayed. Additionally, it is being increasingly recognized that young individuals with childhood onset type 1 DM have an increased risk of eating disorders with clinical presentation precipitated by change in modality of insulin therapy such as initiation of insulin pump therapy (194).

In addition to regurgitation, eating disorders, and self-induced vomiting, there are two other causes of unexplained nausea and vomiting, that is, cyclical vomiting syndrome (CVS) and cannabinoid hyperemesis syndrome (190). Typically affecting young adults, CVS is characterized by discrete recurrent episodes of intense nausea and vomiting that generally last 3 to 5 days and occur approximately every 3 to 4 months. Vomiting often starts abruptly, although a prodrome of nausea and abdominal pain can occur. Symptoms usually fade gradually and in between episodes, patients may have milder symptoms but not vomiting. It may be linked to the menses (catamenial CVS), precipitated by pregnancy, or associated with DM. Common precipitants of CVS episodes include stress, sleep deprivation, infections, foods, motion sickness, and medications. When the episodes of vomiting become closer together, differentiation of coalescent CVS from the more typical daily symptoms of gastroparesis in an adult can be challenging. In contrast to gastroparesis, in CVS the episodes of vomiting are stereotyped, weight loss is uncommon, and up to 60% of individuals have rapid GE (190, 195).

Similar to CVS, cannabinoid hyperemesis syndrome is also characterized by stereotypical episodic vomiting. Additionally, patients present after prolonged excessive cannabis use, which is often associated with a predilection for hot showers, and vomiting episodes are relieved by sustained cessation of cannabis use. All patients with DM and vomiting should be asked about cannabis use. Typically, GE in CVS and cannabinoid hyperemesis syndrome is normal or rapid; however, 14% of a large series of patients had delayed GE (196).

Other potential explanations for GI symptoms (e.g., alcohol or medications such as nonsteroidal anti-inflammatory drugs or opioids) should also be sought (183). Rarely, endocrine disorders associated with metabolic derangements may also mimic gastroparesis. Adrenal insufficiency can result in symptoms similar to that of gastroparesis and is also associated with delayed GE (197, 198). Physical signs such as hyperpigmentation of mucous membranes and cutaneous scars and abnormally low cortisol levels, or an impaired cortisol response to adrenocorticotropic hormone, are suggestive of this condition. Renal tubular acidosis (in children) (199) and hypercalcemia (200) have similarly been mistaken for gastroparesis; basic laboratory testing including pH and chloride and calcium levels should exclude these possibilities.

Other GI manifestations of DM

Diabetic diarrhea and fecal incontinence

It is useful to consider the pathophysiology of diabetic diarrhea into conditions that are or are not associated with malabsorption. Adrenergic denervation (201), artificial sweeteners (e.g., sorbitol), and bile acid malabsorption (202, 203) predispose to diarrhea without malabsorption.

Diarrhea accompanied by features of malabsorption (e.g., anemia, macrocytosis, or steatorrhea) should prompt investigations for bacterial overgrowth, small bowel mucosal disease, or pancreatic insufficiency. Small intestinal dysmotility predisposes to bacterial overgrowth, which can cause bile salt deconjugation, fat malabsorption, and diarrhea. Type 1 DM is associated with celiac disease (Table 3) (204). Pancreatic exocrine insufficiency (e.g., due to chronic pancreatitis or cystic fibrosis) is associated with DM (205). It may cause or, rarely, be consequent to DM (205).

Table 3.

GI Dysfunctions

GI Manifestations of Diabetes	Associated Diseases	Clinical Presentation
Reduced gall bladder motility		Gallstones, cholecystitis
Antral hypomotility and pylorospasm	Gastroparesis	Gastric stasis, bezoars
↓ Gastric accommodation		Dyspepsia
↓ α2-Adrenergic tone in enterocytes	Exocrine pancreatic insufficiency	Impaired absorption of fluids and electrolytes in the small intestine
Small bowel dysmotility	Celiac disease, small bowel bacterial overgrowth	Rapid or slow small bowel transit
Colonic dysmotility	Bile acid malabsorption	Constipation or diarrhea
Anorectal dysfunctions		Disordered defecation or fecal incontinence

In patients with diabetic diarrhea, loose stools and anorectal dysfunctions may cause fecal incontinence. Compared with continent people with DM and healthy controls, patients with DM and fecal incontinence have reduced sensation (206, 207). A pudendal neuropathy may result in reduced anal squeeze pressure. A sympathetic neuropathy may impair internal anal sphincter functional and anal resting pressures. Besides managing diarrhea, pelvic floor biofeedback therapy, which can improve anal sphincter function and rectal sensory disturbances, should be considered (207).

Constipation

The mechanisms have not been carefully studied and are poorly understood. Similar to chronic idiopathic constipation (208), colonic dysmotility and anorectal dysfunctions (i.e., impaired relaxation of the anal sphincters and puborectalis during defecation) may contribute to constipation in DM (209). Colonic dysmotility is characterized by an impaired colonic contractile response to a meal and delayed colonic transit (210). Patients with reduced rectal sensation may not perceive the desire to defecate. Acute hyperglycemia inhibited the colonic contractile response to gastric distention and proximal colonic contraction elicited by colonic distention in healthy people (211). However, acute hyperglycemia did not significantly affect fasting or postprandial colonic motility, tone, compliance, and sensation, or rectal compliance and sensation in healthy people (212).

The management of constipation is guided by the pathophysiology (208). Patients with a defecatory disorder should be referred for pelvic floor biofeedback therapy. For patients who do not have a defecatory disorder, therapy should begin with a dietary fiber supplement and/or osmotic or stimulant laxatives. If these agents do not work, other agents (e.g., a secretagogue or cholinesterase inhibitor) may be considered (213, 214).

Therapy

Principles of management

The goals of therapy are to reduce symptom burden, to ensure adequate nutritional intake and, to improve QOL. We recommend a stepwise approach that initially incorporates safe and less invasive interventions, reserving more invasive options for patients in whom the initial therapies are ineffective (Fig. 3). Although there is evidence to support many of the individual steps, the efficacy of this stepwise approach has not been evaluated in clinical trials.

Figure 3.

Stepwise approach to the management of gastroparesis. [From Adil E. Bharucha, MBBS, MD; Yogish C. Kudva, MBBS; and David O. Prichard, MB, BCh, PhD. Used with permission of the Mayo Foundation for Medical Education and Research. All rights reserved.]

Several medications, especially GLP-1 analogs that are used for the treatment of DM and obesity, anticholinergic agents, and opioids delay GE and induce upper GI symptoms (136–139). Opioids also delay small intestinal and colonic transit and are associated with the narcotic bowel syndrome (136, 137). Delayed colonic transit can delay GE via reflex mechanisms. In the United States, the use of opioids in the overall population and in patients with gastroparesis has increased to alarming levels (215). Indeed, 44% of patients in the NIDDK gastroparesis consortium cohort were taking opioids (110). Because opioids delay GI transit, are associated with dependence and addiction, and may cause hyperalgesia, we strongly discourage the use of opioids in patients with gastroparesis (216).

Antiemetics are used as first-line therapy, often before the diagnosis is established. Dietary modification utilizing a low-fat, low-fiber small-particle diet is beneficial but infrequently followed (110, 217). When necessary, improving glycemic control is desirable, but there is little evidence to suggest that it improves GE. In the United States, metoclopramide and erythromycin are the only FDA-approved medications that can accelerate GE. Although these drugs improve symptoms and facilitate increased nutritional intake, their use is associated with adverse effects (218, 219). Administration of these medications in suspension, where available, rather than tablet form may reduce the risk of erratic pharmacokinetic profiles secondary to delayed GE by virtue of the differing GE profiles of liquids, solids, and large particles (39, 40). Ghrelin and 5-hydroxytryptophan receptor agonists, which are in advanced clinical trials, appear to be promising options.

If these measures are insufficient, supplemental nutrition, preferably enteral, administered through a feeding jejunostomy often accompanied by a venting gastrostomy, should be considered (193). The evidence for more invasive options (i.e., GES, pyloromyotomy, and gastrectomy) is of low quality (193). These should be considered on a case-by-case basis, and only at specialist centers. In primarily uncontrolled studies, GES improved symptoms but not GE. Hence, the enthusiasm for GES varies considerably across centers. Endoscopic pyloromyotomy is the new kid on the block. Subtotal or total gastrectomy is rarely performed.

Nutritional support

In the NIDDK gastroparesis consortium study, ∼60% of patients with diabetic gastroparesis consumed a calorie-deficient diet (i.e., an intake of <60% of the estimated total energy requirements) (151). Less than 40% of patients took a multivitamin supplement, and <10% and 20% supplemented dietary iron and calcium intake, respectively. Although 60% of patients with type 1 DM and 39% of patients with type 2 DM had consulted a dietitian, no patients followed a suggested gastroparesis diet (151). These findings underscore the gaps and the scope for improving adherence to nutritional guidelines.

Primarily based on physiologic principles and supported by relatively small clinical trials, a low-fat, low-fiber, small-particle or liquid diet is recommended for gastroparesis (220, 221). Patients should begin by restricting the ingestion of foods in the “difficult to digest” category, and expand, as necessary, to exclude foods in the moderate and easy to digest groups. Practical and specific dietary guidelines are provided elsewhere (217). The rationale for these recommendations is that solids need to be ground into particles ≤2 mm before they can be emptied from the stomach, and lipids in the small intestine delay GE via enterogastric feedback mechanisms (38).

Among 12 patients with gastroparesis, of whom 3 had diabetic gastroparesis, the symptoms were most severe after high-fat solid meals, followed thereafter by low-fat solids, high-fat liquids, and low-fat liquids (152). The severity of nausea more than doubled after the high-fat solid meal. In a survey of 45 patients with gastroparesis of whom 39 had idiopathic gastroparesis, fatty-, acidic-, spicy-, or roughage-based foods were reported to provoke symptoms, whereas bland, sweet, salty, and starchy foods were associated with tolerable symptoms (153).

Among 56 patients with insulin-treated DM and gastroparesis, dietary advice, which was underpinned primarily by foods of a small particle size and administered by dietitian on seven occasions during 2 weeks, improved virtually all symptoms (i.e., nausea and vomiting, abdominal fullness and bloating, lower but not upper abdominal pain, heartburn, and regurgitation) compared with a standard diabetic diet with similar nutrient composition after 20 weeks (217). The GE improved to a greater extent in the intervention group than in the control group. Although anxiety declined in the intervention group, QOL and HbA_1c did not significantly improve in either group.

Improving glycemic control

The symptoms of gastroparesis (e.g., nausea, vomiting, and erratic caloric intake) increase the challenge of balancing carbohydrate intake while avoiding hypoglycemia (222). However, as detailed in the section on the bidirectional relationship between glycemia and GE, there is limited evidence that delayed GE is associated with hypoglycemia in DM (90). To avoid the unpredictable hypoglycemic episodes associated with delayed GE (222), patients may be more comfortable with suboptimal glycemic control. Medications that accelerate GE may worsen glycemic control. In one study, domperidone was associated with clinically significant hypoglycemia in 5 of 12 patients with DM (223). Dose-related worsening of glycemic control was observed in 15% of patients who received relamorelin, necessitating adjustments to medications in some patients (169).

There have been several advances in CGM techniques, insulin preparations, and automated insulin delivery, especially for patients with type 1 DM. By informing patients when the blood glucose is trending lower, CGM can reduce the duration of or avert hypoglycemia. Postprandial glucose measurements also provide patients with a refined understanding of the magnitude, onset, and duration of hyperglycemia after different foods (224). Through a fine cannula placed in subcutaneous tissues, insulin pumps infuse rapid-acting insulin at a basal rate throughout the day and provide insulin boluses with meals and in between meals. [The basal rate can be preset to change at any time (e.g., at night) or altered on demand.] Bolus calculators use several factors based on meal intake or suboptimal intermeal glucose status (e.g., using either carbohydrate intake or premeal/target blood glucose levels and insulin sensitivity). Integrating pumps with CGM closes the loop between the glucose signal and insulin delivery; that is, insulin delivery is automated (225). Such systems, termed closed loop or artificial pancreas (AP), have been developed in the last decade. Early versions, which were approved in Europe, and thereafter in the United States, suspended insulin delivery when CGM glucose decreased below a predetermined set point (e.g., 70 mg/dL). More recently, hybrid closed-loop systems, which were approved by the FDA in September 2016, modulate basal insulin delivery (i.e., increase, decrease, or suspend) in real time but continue to rely on patient decision-making for meal insulin bolus delivery. In a multicenter cohort of 124 patients with reasonably well-controlled type 1 DM (mean baseline HbA_1c of 7.7% in adolescents and 7.3% in adults) who were switched to a hybrid closed-loop system, the proportion of in-target glucose values increased from 60% to 67%; no severe hypoglycemic or diabetic ketoacidotic episodes were observed (226). The Medtronic 670G Hybrid AP was FDA approved in September 2016 and has been marketed in the United States since March 2017 (226, 227). The same AP model and other competing systems are currently being tested in randomized clinical trials. More details are available in a recent review (228).

Conceivably, recent advances in monitoring and therapy of DM may be beneficial for patients with DM and gastroparesis who have increased glycemia or glycemic variability (222, 224, 229). It is our impression that the adoption of CGM and insulin pumps in patients with type 1 DM and gastroparesis has been slow. One study evaluated the combination of CGM and continuous subcutaneous insulin infusion in 45 patients with DM and gastroparesis, of whom 32 had type 1 DM (222). Patients were trained consecutively for up to 8-week run-in periods on the Medtronic iPRo2 CGM and 522/23 insulin pump, and subsequently followed for 24 weeks. During this period, the HbA_1c improved significantly by 1.1%; the total Gastroparesis Cardinal Symptom Index score and QOL also improved significantly. The mean glucose, duration for which glucose was <70 mg/dL, and the duration for which glucose was >180 mg/dL were all significantly lower. Since this trial, the CGM systems and insulin pumps have been substantially improved. A more rapid acting insulin is now available, will be tested in future AP systems, and may provide better control of glycemia (230).

In type 2 DM, several medications, including newer agents (e.g., SGLT-2 inhibitors) to improve glucose control, are available and can be used alone or in combination (231). GLP-1 receptor agonists should be avoided because they delay GE. Conceivably, oral agents may be combined with CGM and insulin pumps even for type 2 DM.

Medications

Metoclopramide

Metoclopramide reduces emesis via central effects (i.e., at chemoceptor trigger zone) (218), which are attributable to antagonism of dopamine receptors, and peripheral effects, which are attributable to cholinergic agonism and increased foregut motility. The latter comprises increased lower esophageal sphincter and gastric fundus tone, gastric and duodenal motility with synchronized antroduodenal motility that accelerates GE (232–234). It can be administered via the oral (tablet and liquid) (218), parenteral (218), rectal (235), or nasal routes (236).

Acute IV administration of metoclopramide increases the rate of GE (235, 237, 238). The efficacy of oral metoclopramide for gastroparesis has been evaluated in seven blind clinical trials (236, 239–244) (Table 4). In four of five double-blind, placebo-controlled, short-term studies in diabetic gastroparesis, symptoms improved to a greater extent with metoclopramide than placebo. In two studies, metoclopramide accelerated the rate of GE (239, 240); a third trial may have included patients who were studied previously (240, 243). The limited correlation between the improvements in symptoms and GE suggest that the centrally mediated effects of metoclopramide are at least as important as the peripherally mediated effects. These trials did not evaluate the long-term efficacy of metoclopramide.

Table 4.

Effects of Metoclopramide on Symptoms and GE in Patients with Diabetic Gastroparesis

Study	Design	Etiology (N)	Dose	Duration	Symptom Response	Effects on GE	Comments
Perkel et al., 1980 (242)	DB	PSG (20)	10 mg QID PO	3 wk	MET: 46% improvement (P < 0.01). Placebo: 22% reduction (P < 0.02). Response to MET > placebo (P < 0.01)		10 of 26 had AEs (6 restlessness, 3 drowsiness or confusion, and 1 dystonic reaction); 4 patients withdrew from the study
	PC	IG (27)
	PG	DG (5)
	RCT
Snape et al., 1982 (244)	DB	DG (10)	10 mg QID PO	3 wk	MET: Less vomiting but no change in abdominal pain or bloating. Placebo: No changea	GE of liquid meal at 60 min increased from 33% to 57% after treatment with MET in DM (P < 0.01). Placebo had no effect. For reference, GE was 80% at 60 min in healthy volunteers	No relationship between change in symptoms and emptying
	PC
	XO
	RCT
McCallum et al., 1983 (240)b	DB	DG (40)	10 mg QID PO	3 wk	Reduction in fullness and nausea (P < 0.05) relative to placebo.	Solid meal (595 kcal) retention at 90 min: reduced from 91% to 68% (n = 7, P < 0.05)	11 of 18 had AEs (restlessness, headache, cramps, reduced strength, diarrhea, and constipation)
	PC
	PG
	RCT
Ricci et al., 1985 (243)b	DB	DG (13)	Acute: 10 mg IV	3 wk	50% improvement in total symptom score (P < 0.01 relative to placebo). Only satiety did not improve significantly	Solid meal (595 kcal) retention at 90 min: reduced from 91% to 79% (P < 0.05, norm 52%) after acute IV. Oral MET also improved GE; the data were also reported in a different study (240)	No relationship between effects on symptoms and GE; 4 of 13 had AEs (sedation, amenorrhea, tremor, headache, agitation)
	PC		Chronic: 10 mg QID PO
	XO
	RCT
Erbas et al., 1993 (239)	SB	DG (13)	MET (10 mg TID PO) or ERY (250 mg TID PO)	3 wk	MET and ERY reduced symptoms by >50% (both P < 0.05). Symptoms improved to a greater extent with ERY than with MET (P < 0.05)	Solid meal retention at 90 min: 55% at baseline, 42% after MET, and 41% after ERY (both P < 0.05 relative to baseline)	4 of 13 AEs with MET (weakness, drowsiness, cramps, tachycardia); ERY more effective than MET without AE
	XO
Patterson et al., 1999 (241)	DB	DG (93)	MET	4 wk	∼40% reduction for both MET (P = NS) and DOM (P < 0.01)		CNS-related AEs (e.g., somnolence, restlessness, asthenia, anxiety) more common with MET than DOM; 9 patients (3 DOM and 6 MET) discontinued early due to AE
	PG		10 mg QID PO
	RCT		or
			DOM
			20 mg QID PO
Parkman et al., 2015 (236)	DB	DG (285)	Nasal spray (10 or 14 mg QID)	4 wk	Symptom reduction above placebo in women (P < 0.03 for both doses of MET) but not men (P = NS). Placebo response rate was 30%–48%; significance unspecified		Dysgeusia, fatigue, and headache were more common in MET then placebo
	PG
	PC
	RCT

Abbreviations: AE, adverse event; CNS, central nervous system; DB, double-blind; DG, diabetic gastroparesis; DOM, domperidone; ERY, erythromycin; IG, idiopathic gastroparesis; MET, metoclopramide; NS, not significant; PC, placebo controlled; PG, parallel group; PO, per os (by mouth); PSG, postsurgical gastroparesis; QID, four times a day; RCT, randomized controlled trial; SB, single-blind; TID, three times a day; XO, crossover.

^aNo statistical analysis of symptom response provided in this article.

^bProbable overlapping cohorts.

Unfortunately, the use of metoclopramide is limited by adverse effects; indeed, more than one-third of patients reported acute adverse effects in blinded trials (Table 4). Because metoclopramide crosses the blood–brain barrier, its use is associated with hyperprolactinemia and a risk of extrapyramidal adverse effects (245). Hyperprolactinemia is associated with a risk of galactorrhea, amenorrhea, gynecomastia, and impotence. Acute reversible dystonic reactions are the most commonly reported extrapyramidal adverse effects and occur most frequently in younger women. Drug-induced parkinsonism, which is usually evident within 3 months of drug initiation, usually resolves with a few months of drug withdrawal. Tardive dyskinesia, a potentially irreversible movement disorder, is more likely to occur in older women and in patients exposed to higher cumulative doses of metoclopramide. Retrospective estimates of the incidence of tardive dyskinesia range from 0.1% to 1% in 2000 to 2800 patient-years of exposure. Conversely, metoclopramide is the most common cause of drug-induced tardive dyskinesia (245). For this reason, the FDA approval to relieve “symptoms in adults with acute and recurrent diabetic gastroparesis” is accompanied by a black box warning regarding tardive dyskinesia (246). To reduce this risk, metoclopramide should be administered as a liquid formulation at the lowest effective dose, starting with 5 mg 15 minutes before meals and at bedtime. Metoclopramide is not only metabolized by the cytochrome oxidase CYP2D6 but it is also analogous to neuroleptic agents (e.g., haloperidol), a competitive inhibitor of this enzyme. Pharmacogenomic testing can potentially identify individuals who have reduced CYP2D6 activity and, consequently, an increased risk of tardive dyskinesia (247).

Domperidone

Similar to metoclopramide, domperidone is a dopamine receptor antagonist (219). It is only available as an oral formulation and improves symptoms to a similar extent as metoclopramide (241). However, in contrast to metoclopramide, domperidone does not cross the blood–brain barrier. Hence, the risk of extrapyramidal adverse effects is extremely low (219).

In open-label and blinded controlled trials, domperidone improved symptoms and GE in gastroparesis (223, 241, 248–253) (Table 5). Short-term open-label studies demonstrated a substantial improvement in symptoms (223, 241, 251, 253). Longer, open-label studies suggest that this improvement is sustained for up to 2 years (249, 252). In one study, the improvement in symptoms was persistent but the effects on solid GE declined after administration for 5 to 7 weeks (223). Similar to metoclopramide, this suggests that centrally mediated effects of domperidone are important.

Table 5.

Selected Studies Regarding the Efficacy of Domperidone on Symptoms and GE in Patients with Gastroparesis

Study	Design	Etiology (N)	Dose	Duration	Symptom Response	Effect on GE	Comments
Nagler and Miskovitz, 1981 (250)	DB	Unclear (11)	10 mg QID PO	4 wk	No difference between DOM and placebo (P = NS)		It is unclear whether these patients truly had gastroparesis. Two patients each had 1 AE (i.e., abdominal pain and skin rash)
	PC
	XO
	RCT
Horowitz et al., 1981 (223)	Acute phase:	DG (12)	40 mg PO	Single dose		Solid meal (270 kcal) retention at 100 min: placebo 79% vs DOM 46% (P < 0.005)	More frequent hypoglycemic episodes while taking DOM but no change in HbA1c during study period
	DB
	PC
	XO
	Chronic  phase: Open		20 mg TID PO	5–7 wk	66% reduction (P < 0.001)	Solid meal (270 kcal) retention at 100 min: 67% (P = NS vs baseline value of 79%)
Koch et al., 1989 (249)	Open	DG (6)	20 mg QID PO	6 mo	80% reduction (P < 0.01)	Solid meal retention at 120 min: 78% at baseline vs 57% on therapy (P = NS)	Normalization of gastric dysrhythmias with patients regressing toward normal 3-cpm frequency. AE not reported
Soykan et al., 1997 (252)	Open	IG (12)	20 mg QID PO	23 (6–48) mo	68% reduction (P < 0.05)	Solid meal (274 kcal) retention at 120 min: 87% at baseline vs 57% on therapy (P < 0.05)	15 of 17 patients regarded QOL as greater than good. AE: gynecomastia in 3 patients, increased prolactin in all
		DG (3)
		PSG (2)
Silvers et al., 1998 (251)	Induction  phase: SB cohort	DG (287)	20 mg QID PO	4 wk	63% reduction in symptoms (P < 0.001)		QOL substantially improved in those who responded
	Withdrawal phase:	DG (208)	20 mg QID PO	4 wk	Deterioration in symptoms in both placebo and DOM group. No difference based on physician global assessment (P = NS). Significant difference by patient diaries (P = 0.01) and questionnaires (P = 0.03)		AE: 63% in placebo vs 60% in DOM. Three DOM patients withdrew due to AE. Improvement in QOL maintained in patients receiving DOM but deteriorated in those randomized to placebo (P < 0.05)
	DB
	PG
	PC
	RCT
Patterson et al., 1999 (241)	DB	DG (93)	MET (10 mg QID) or DOM (20 mg QID)	4 wk	∼40% reduction in symptoms for both MET (P = NS) and DOM (P < 0.01)		3 DOM patients withdrew due to AE
	PG
	RCT
Schey et al., 2016 (253)	Open	IG (90)	10 mg TID/QID PO	Average of 2 mo	Significant reduction in almost all symptoms (P < 0.001 for most); 45% reported > moderately improved symptoms (CPGAS ≥4)		AE: headache, palpitations, and tachycardia. Treatment discontinued in 12% of patients
		DG (16)
		PSG (9)

Abbreviations: AE, adverse event; CPGAS, Clinical Patient grading Assessment Scale; DB, double-blind; DG, diabetic gastroparesis; DOM, domperidone; ERY, erythromycin; IG, idiopathic gastroparesis; MET, metoclopramide; NS, not significant; PC, placebo controlled; PG, parallel group; PO, per os (by mouth); PSG, postsurgical gastroparesis; QID, four times a day; RCT, randomized controlled trial; SB, single-blind; TID, three times a day; XO, crossover.

Domperidone increases serum prolactin levels and can cause gynecomastia (219). In one study, “five (of 12) patients observed more frequent hypoglycemic episodes while taking domperidone and reduced their insulin dose” (223). However, HbA_1c levels did not significantly change during the 5- to 7-week study period (223). Although domperidone appears to have negligible effects on QT interval in healthy people at doses up to 20 mg four times a day (254), concerns regarding the cardiac effects of this medication have prompted the European Medications Agency to recommend that use of this medication be limited to short-term treatment of nausea and vomiting at a dose <30 mg/d. Furthermore, caution should be exercised in patients who have moderately or severely impaired liver function or at increased risk of cardiac events (e.g., prolonged corrected QT interval, serious ventricular arrhythmia, torsades de pointes, or sudden cardiac death) (255). Coadministration of domperidone with QT-prolonging drugs is contraindicated except for apomorphine. Domperidone is not FDA approved for use in the United States but it is available under an FDA investigational new drug protocol that stipulates that electrolyte and electrocardiogram monitoring must be performed before starting and during treatment.

Motilin agonists (erythromycin and azithromycin)

Motilin is a 22–amino acid peptide hormone expressed throughout the gut (256). Motilin receptors, however, are predominantly located proximal to the ligament of Trietz (257). Stimulation initiates phases 2 and 3 interdigestive antral contractions that promote GE (258–261).

Erythromycin, a macrolide antibiotic, which is available as oral (tablet and suspension) and IV formulations, acts as a motilin receptor agonist and stimulates gastric migrating motor complexes at doses substantially lower than those used for antibacterial effects (262, 263). At a low dose, IV erythromycin (40 mg) induced a premature migrating motor complex in some patients with diabetic gastroparesis (263). A higher dose (200 mg) predictably induced giant gastric antral contractions that did not migrate to the small intestine and accelerated GE (263). IV administration evokes a greater effect than does oral administration on GE (264), and there appears to be no incremental effect for PO doses >250 mg (265). The erythromycin-induced acceleration of GE is attenuated by higher blood glucose concentrations (266–270) and after repeated administration (3 to 4 weeks) (271).

Short-term treatment of 2 to 4 weeks improved symptoms in some but not all open-label studies (239, 264, 271–275) (Table 6). One single-blinded, crossover study suggested that symptoms improved to a greater extent with erythromycin than with metoclopramide (239). Two retrospective studies suggested that the response to treatment may be sustained in some patients (276, 277). Among 11 patients with gastroparesis, of whom 8 had DM, the improvement in symptoms with erythromycin was sustained for 6 months in 71% of patients (276). In another study of 25 patients including 2 patients with DM, oral erythromycin (50- to 100-mg suspension four times daily) provided substantial symptoms relief in both the short term (6 to 8 weeks, n = 18) and long term (11 ± 7 months, n=18) (277). In both reports, symptoms improved to a greater extent in the acute phase than in the chronic phase; early treatment failure predicted long-term failure.

Table 6.

Selected Studies Regarding the Efficacy of Erythromycin on Symptoms and GE in Patients with Gastroparesis

Study	Design	Etiology (N)	Dose	Duration	Results Symptoms	Effect on GE	Comments
Janssens et al., 1990 (272)	Acute phase:	DG (10)	200 mg IV	One dose	Improved in 75% of symptomatic patients. Two patients on tube feeding resumed oral feeds. Three patients reported fewer hypoglycemic attacksa	Retention of a solid meal (231 kCal) at 120 min: reduced from 64% (placebo) to 4% (P < 0.001) with acute IV and 28% (P < 0.05) with chronic PO ERY	Loss of liquid/solid GE discrimination (i.e., both phases emptied simultaneously) after administration of IV ERY. AE: none
	DB
	PC
	XO
	Chronic phase:		250 mg PO TID	4 wk
	Open
Urbain et al., 1990 (271)	Acute phase:	DG (12)	200 mg IV		NA	Retention of a solid meal at 120 min reduced from 58% (baseline) to 6% with acute IV and 27% (P < 0.05) with chronic PO ERY	Loss of liquid/solid emptying discrimination after administration of IV ERY. AE: none
	Open
	Chronic phase:		500 mg PO TID	3 wk
	Open
Richards et al., 1993 (273)	Acute phase:	IG (10)	6 mg/kg IV		No overall reduction in total symptom score; 35% reduction (P < 0.05) in global assessment score	Retention of solid meal (275 kCal) at 120 min reduced from 85% (baseline) to 20% (P < 0.01) after acute IV and 48% (P < 0.01) with chronic PO ERY	4 AEs resulting in withdrawal; rash (2), cramps (1). and vomiting (1); 50% remained on medication at 8 mo (range, 7–11); 50% discontinued treatment due to lack of effect
	Open	DG (4)
	Chronic phase:		500 mg before meals and at bedtime	4 wk
	Open
Erbas et al., 1993 (239)	SB	DG (13)	MET (10 mg TID) vs ERY (250 mg TID)	3 wk	>50% reduction in symptom score for MET and ERY (both P < 0.05). Symptoms improved to a greater extent with ERY than MET (P < 0.05)	Retention of solid meal at 90 min reduced from 55% pretreatment to 42% with MET and 41% with ERY (both P < 0.05 relative to baseline)	No AE with ERY
	XO
Samson et al., 1997 (274)	DB	DM (12) of which 5 had DG	250 mg PO TID	2 wk	Overall, effects NS. Symptom reduction (P < 0.05) in patients with antral hypomotility	NA	None
	PC
	XO
	RCT
Ramirez et al., 1994 (264)	Acute phase:	PSG (9)	200 mg PO		Overall, no symptom improvement (P = NS)	GE of solids (GE t1/2) declined from 154 to 56 min with IV ERY (P < 0.01) and from 146 to 87 min with PO ERY (P < 0.05)	Acute IV administration had a greater effect on GE than PO
	Open
	XO
	Chronic phase:		150 mg PO TID	2 wk
	Open
Fiorucci et al., 1994 (275)	Acute phase:	SS (12)	2 mg/kg		Significant reduction in upper GI symptoms (P < 0.01)	60% reduction (P < 0.01) in GE t1/2 (solid meal, 1000 kcal) with both acute IV and chronic PO administration	1 AE (vomiting). Emptying measured with ultrasound
	Open
	Chronic phase:		250 mg PO TID	4 wk
	Open

Abbreviations: AE, adverse event; DB, double-blind; DG, diabetic gastroparesis; ERY, erythromycin; GE t_1/2, gastric emptying half-time; IG, idiopathic gastroparesis; MET, metoclopramide; NA, not available; NS, not significant; PC, placebo controlled; PO, per os (by mouth); PSG, postsurgical gastroparesis; RCT, randomized controlled trial; SB, single-blind; SS, systemic sclerosis; TID, three times a day; XO, crossover.

^aNo statistical analysis of symptom response provided in this article.

Limited evidence suggests that a second macrolide antibiotic, azithromycin, may also improve the rate of GE (278–280). Relative to erythromycin, azithromycin has a longer half-life, fewer drug interactions, and fewer adverse effects. However, there are no reports regarding the effect of azithromycin on gastroparesis symptoms. Furthermore, similar to erythromycin, azithromycin use is associated with a risk of sudden cardiac death, particularly in those with underlying cardiac disease (281). A third compound, derived from macrolide antibiotics but with no antibiotic properties, increased GE but failed to reduce symptoms in patients with DM and upper GI symptoms (282, 283).

Ghrelin antagonists

Ghrelin is a natural ligand for the growth hormone secretagogue receptor expressed throughout the GI tract (284). At pharmacologic doses, ghrelin promotes gastric motility in healthy people and in idiopathic gastroparesis (285). Two ghrelin agonists (i.e., TZP-102 and relamorelin) have been assessed in diabetic gastroparesis. A phase 2b trial of TZP-102 in patients with diabetic gastroparesis was terminated early because the drug was ineffective (286). There are two large phase 2 trials of relamorelin (RM-131), which is a subcutaneously administered ghrelin receptor agonist, in diabetic gastroparesis. The first trial, with 204 patients, observed that relamorelin reduced the frequency of vomiting by 60% (i.e., significantly vs placebo) and also accelerated the GE half-time by 23% (170). However, relamorelin only improved bloating, abdominal pain, early satiety, and nausea in patients who experienced vomiting symptoms at baseline. In the second study, relamorelin significantly reduced the GE half-time by 12% and also substantially improved the symptoms of bloating, nausea, postprandial fullness, abdominal pain, and the composite daily diary score in 393 patients with moderate to severe diabetic gastroparesis, of whom 90% had type 2 DM (169). Compared with baseline, the composite GI symptom score declined by an average of 318 U for placebo, and between 500 and 554 U for the three doses of relamorelin. However, in contrast to the phase 2a study, relamorelin did not significantly affect the primary end point (i.e., frequency of vomiting) in this study. Although adverse effects were not more frequent for relamorelin than placebo, the glycemic control worsened in 15% of patients with diabetes treated with relamorelin. Five patients were hospitalized with hyperglycemic complications, of whom three had diabetic ketoacidosis. Phase 3 trials are in progress.

Serotonin receptor antagonists

Cisapride, which is a serotonin receptor agonist that improved the symptoms of gastroparesis, was withdrawn from the US market due to concerns regarding unexpected cardiac events. More recently, 5-HT(4) receptor agonists are being evaluated for treating gastroparesis. Reported in abstract form, prucalopride, which is currently used to treat constipation, improved the GE rate and symptoms of gastroparesis in a blinded crossover study (287). However, 4 of 28 patients discontinued the medications due to adverse effects. Velusetrag accelerated GE in healthy volunteers (288) and in patients with idiopathic and diabetic gastroparesis (289). The reports of phase 2 studies in gastroparesis are awaited (clinicaltrials.gov).

Management of nausea and abdominal pain

Several antiemetic agents such as phenothiazines (e.g., prochlorperazine), antihistamines (e.g., promethazine or meclizine), transdermal scopolamine, and serotonin receptor antagonists (e.g., ondansetron) are used to treat nausea associated with gastroparesis. However, the efficacy of these agents has not been evaluated in controlled clinical trials. A placebo-controlled study of the neurokinin-1 receptor antagonist aprepitant, which is approved by the FDA for managing chemotherapy-associated nausea and vomiting, reported mixed results in patients with nausea and vomiting with or without gastroparesis (290). Of 126 patients, 29% had DM and 57% had delayed GE; the remainder had normal (40%) or rapid (3%) GE. Compared with placebo, aprepitant did not reduce the severity of nausea evaluated with a visual analog scale, which was the primary end point. However, aprepitant significantly reduced the severity of nausea, retching, and vomiting evaluated with other indices (e.g., patient assessment of upper gastrointestinal symptom severity index questionnaire).

Two studies suggest that acupuncture also improves symptoms in diabetic gastroparesis (291, 292). Active acupuncture also improved the symptoms of gastroparesis and shortened the GE half-time from 144 ± 56 minutes to 99 ± 29 minutes in nine patients with diabetic gastroparesis; in contrast, sham acupuncture, which was administered in 10 patients, did not improve symptoms or GE (291).

Although abdominal pain is a common symptom in diabetic gastroparesis, no studies have addressed optimal management of this symptom. In a phase 3 study of relamorelin, patients with vomiting symptoms at baseline experienced a 40% reduction in abdominal pain (170). However, relamorelin is not currently available for clinical use. Tricyclic antidepressants such as amitriptyline that are beneficial for treating pain of neuropathic origin and improving symptoms associated with functional GI disorders may be considered (293). However, amitriptyline has anticholinergic effects that can delay GE. Nortriptyline was not better than placebo in patients with idiopathic gastroparesis (294). The effect of tricyclic antidepressants on diabetic gastroparesis has not been studied. Other neuromodulators such as pregabalin or gabapentin may be considered, but no evidence supports this practice. Many patients eventually require or use opioids to manage the abdominal pain associated with gastroparesis (295).

Multidisciplinary approaches

In many patients with functional GI disorders, a holistic perspective, as embodied in the biopsychosocial model, is preferable to an exclusive focus on peripheral (i.e., GI) or central factors for managing symptoms (104). However, at present, the management of diabetic gastroparesis predominantly relies on a biologic (i.e., medications, control of glycemia, and GES) rather than the biopsychosocial model. Comprehensive, intensive, outpatient-based, multidisciplinary pain management programs are very effective for managing symptoms in patients with idiopathic functional GI disorders (e.g., chronic abdominal pain) (296). These programs integrate pharmacologic, psychologic, and behavioral interventions that are tailored to the symptoms, functional impairment, and psychologic distress experienced by patients. The utility of such approaches for managing diabetic gastroparesis has not been evaluated. However, prompted by the high prevalence of depression and impaired QOL in diabetic gastroparesis, the mismatch between improvement in GI symptoms and QOL (110), and our clinical experience, we recommend consideration of these approaches. Indeed, the use of antidepressants independently predicted the improvement in GI symptoms in the NIDDK consortium (110) and the treatment of depression improved QOL in DM (297). Also, because patients with gastroparesis feel frustrated, misunderstood, and burdened by their illness, training to increase resilience may be worthwhile (298, 299).

Gastric electrical stimulation

GES is performed by electrically stimulating the stomach through electrodes implanted along the greater curvature of the stomach. It is primarily used for the treatment of nausea and vomiting, often secondary to gastroparesis. There are two patterns, referred to as pacing and electrical stimulation. Gastric pacing employs depolarizations at frequencies slightly above the intrinsic slow wave frequency (i.e., three cycles per minute) with long duration pulses (300). Similar to a cardiac pacemaker, pacing entrains the stomach and improves GE in dogs and humans. In the only human study of gastric pacing, which was conducted in nine patients with gastroparesis, of whom five had DM, gastric retention at 2 hours decreased from 77% ± 3% (mean ± SEM) to 57% ± 9%, and the mean symptom score declined by almost 50% (301). Eight patients were able to discontinue jejunal feedings thereafter. However, entraining slow waves requires long duration pulses, which consume much energy. These devices are too large to be portable or implantable.

In contrast to pacing, GES is conducted with a pulse generator (Enterra, Medtronic) that provides ongoing high-frequency, low-energy electrical stimulation that does not entrain gastric muscle, does not pace the stomach, and does not consistently improve GE compared with sham stimulation (302). The mechanisms of action of high-frequency GES are unclear. In clinical and animal studies, high-frequency GES modulates vagal and spinal afferent pathways, and it may affect central nausea and vomiting centers, pain thresholds, vagal efferent function, and gastric sensitivity (302, 303). GES also induces gastric relaxation in dogs. In cohort studies, GES reduces nausea and vomiting in humans (304). Two double-blinded randomized trials suggest perhaps that GES may be beneficial in some patients with diabetic gastroparesis. In the first study, 33 patients, of whom 16 had diabetic gastroparesis, were randomized in a double-blind crossover design to stimulation ON or OFF for 1-month periods (305). Thereafter, all patients were programmed to stimulation ON and evaluated at 6 and 12 months. At 1 month, DM patients in the ON group reported a lower total symptom score and median weekly vomiting frequency [6.0 (ON) vs 12.8 (OFF)]. The reduction in symptoms was statistically significant in the entire cohort, but not separately in patients with DM, likely because the study was underpowered. In DM patients, the mean gastric retention at 4 hours declined significantly from 46% at baseline to 16% at 12 months. These effects were greater in patients with diabetic than idiopathic gastroparesis. In another study, 55 patients with treatment-refractory diabetic gastroparesis underwent GES (306). The stimulator was turned ON for 6 weeks after surgery. Thereafter, patients were randomly assigned to two groups with the device ON or OFF for 3 months. Following that, patients crossed over to the other group for the next 3 months. Finally, the device was turned ON in all patients and they were followed up, unblinded, for 4.5 months. At 6 weeks, the median weekly vomiting frequency declined by 57% compared with baseline; however, during the crossover period, differences between device ON or OFF were not significant. At 1 year, vomiting had improved vs baseline (306). In summary, these results suggest that GES may be beneficial in some patients, but there are questions regarding the extent to which these benefits are explained by placebo effects (307). The FDA approved the Enterra device for treating gastroparesis under a humanitarian device exemption.

Two studies have evaluated the effects of a pyloroplasty and GES in gastroparesis (308, 309). When both procedures were performed simultaneously in 27 patients (11 with DM), symptoms and GE improved substantially (308). However, the effects on symptoms were statistically significant in the overall group but not in the patients with DM. Although the improvement in GE seemed impressive, it is unclear whether these results were statistically significant. In another study of 49 patients undergoing GES placement, 8 of 17 with diabetic gastroparesis also had a pyloroplasty performed (309). Among patients with DM, the total symptom severity score improved significantly, by 60%, after GES alone and to a lesser extent (45%) after GES plus pyloroplasty. GE also improved in both groups but the results were not significant. Hence, further studies are necessary to evaluate this approach before it can be recommended.

Approximately 10% of patients experience a severe device-related event after GES (305, 306, 310, 311); a greater percentage experience problems during long-term follow-up (33, 312). The most recent meta-analysis and systematic review suggests that baseline symptom severity and regression toward the mean account for a large proportion of the documented beneficial effect of GES (304). Based on these data, the American College of Gastroenterology guidelines recommended GES only as a compassionate treatment in patients with refractory gastroparesis (193).

Endoscopic therapy for gastroparesis

Intrapyloric injection of botulinum toxin

The efficacy of botulinum toxin for achalasia (313) and the recognition that a subset of patients with gastroparesis, predominantly related to DM, have pyloric dysfunction (68), prompted physiologic and clinical studies of the effect of botulinum toxin on pyloric muscle. The toxin reduced pyloric muscle contractility in vitro by inhibiting the release of acetylcholine from neurons and through direct effects on smooth muscle cells (314). In vivo, botulinum toxin attenuated pyloric pressure waves (315). Early, open-label clinical studies in DM (316, 317) and idiopathic (318) gastroparesis suggested that botulinum toxin improved symptoms and GE. Retrospective cohort studies suggested that higher doses of botulinum toxin may afford greater benefit (319, 320). However, more rigorous randomized, controlled trials in patients with moderately severe diabetic and idiopathic gastroparesis demonstrated that high and low doses of botulinum toxin administered into the pylorus were not beneficial compared with placebo (321, 322). Based on these findings, intrapyloric botulinum toxin should not be used to manage gastroparesis. Indeed, it is conceivable that administration of botulinum toxin into the antrum, which is adjacent to the pylorus, inhibits antral contractility and GE.

Venting gastrostomy

One study evaluated the effects of venting the stomach through an endoscopically placed venting gastrostomy tube on five to six occasions per week, usually postprandially, in patients with idiopathic gastroparesis (323). During a mean follow-up of 29 months, symptoms improved, on average by 54%, and patients regained 4.5 kg of weight. In another study among patients with chronic intestinal pseudo-obstruction, the rate of hospital admission fell by 83% after insertion of a venting gastrostomy (324). Although no studies have examined the effect of this intervention in patients with diabetic gastroparesis, the American College of Gastroenterology guideline conditionally recommends that venting gastrostomy may be performed for symptom relief (193).

Jejunostomy

In patients who cannot tolerate oral feeding, enteral feeds can be given through a jejunostomy, placed via surgery, or via direct percutaneous endoscopic jejunostomy, which is often combined with a venting gastrostomy. The procedure is typically preceded by a trial of nasojejunal feeding for 48 to 72 hours to ensure that patients tolerate the same. In a large series of patients who underwent an upper GI endoscopy to place a jejunostomy tube for enteral feeding, the procedure was successful in ∼70% of cases (325). Severe adverse events (e.g., bowel perforation, jejunal volvulus, major hemorrhage, aspiration) occur in 5% of cases and moderately severe adverse events (e.g., fistulae, pain requiring removal, infection requiring admission) occur in 7% of cases (325). Where successful, a jejunostomy offers more reliable enteral access than does a gastrostomy tube with a jejunal extension, because the extension often slips back into the stomach (326). However, a retrospective review of 26 patients with diabetic gastroparesis utilizing jejunostomy during a 4-year period (32) documented 23 major complications that required surgery or hospitalization in 14 patients, including 1 death related to jejunostomy placement. Yet, in this series, approximately half of the patients reported improvement, albeit not statistically significant, in nausea/vomiting, hospitalizations, and nutritional status. Furthermore, overall health improved significantly. Therefore, despite the potential risks, when nutrition requirements cannot be maintained with oral feeding, if tolerated, enteral nutrition is preferable to parenteral nutrition.

Transpyloric stenting

The only retrospective case series reported 30 gastroparesis patients of whom 8 had DM (327). Symptoms were not evaluated with rigorous assessments. Stent placement was technically successful in 98% of patients and symptoms improved substantially in 75%. Patients with predominant nausea and vomiting appeared to achieve the greatest symptomatic benefit. On average, patients gained 5 kg in weight after the procedure. Among the 16 patients with a repeat GE study, improvement or normalization occurred in 11. Despite the stent being fixed in place (predominantly with sutures or clips), stent migration occurred in 59% of cases. No adverse results were reported secondary to stent migration.

“Patients with diabetic gastroparesis may respond less favorably to gastric per-oral endoscopic myotomy than patients with idiopathic… or postsurgical gastroparesis.”

Gastric per-oral endoscopic myotomy

This relatively new endoscopic procedure was introduced in 2013 by adapting techniques that were used for esophageal per-oral endoscopic myotomy (328). Via endoscopy, a 5-cm submucosal tunnel is created between the muscularis propria and the gastric mucosa, facilitating access to the pyloric muscle. The circular and/or longitudinal layer of the pylorus and 2 to 3 cm of the antrum are then incised using electrocautery. The entry point to the submucosal tunnel is sealed when the endoscope is withdrawn. Between 2013 and 2017, fewer than 20 cases were reported. In 2017 and 2018, >130 additional patients in five discrete case series were described (329–333) (Table 7). Symptom improvement, evaluated with validated instruments, was reported in three (329, 330, 333); one demonstrated no significant difference (332); and in one series there was no objective symptom assessment (331). A significant improvement in GE at 4 hours was documented in two (329, 333); two others demonstrated no significant difference (330, 332); and in one the significance of the reduction was not reported (331). However, the reported adverse effect profile was minor when patients adhered to the prescribed postoperative protocol. Patients with diabetic gastroparesis may respond less favorably to gastric per-oral endoscopic myotomy than do patients with idiopathic gastroparesis or postsurgical gastroparesis (329, 330). These early results are promising; however, as reminded by the history of GES, a controlled study is desperately needed.

Table 7.

Selected Studies Regarding the Efficacy of Gastric Per-oral Endoscopic Myotomy (G-POEM) on Symptoms and GE in Patients with Gastroparesis

Study	Etiology (N)	Follow-Up (mo)	Mean Gastroparesis Cardinal Symptoms Index Score			Gastric Retention at 4 h (%)			Hospital Stay (d)	Complications (n)
			Before	After	P Value	Before	After	P Value
Gonzalez et al., 2017 (330)	DG (7)	6	3.3 ± 0.9	1.1 ± 0.9	<0.001	40 ± 34	28 ± 45	0.07	5	Capnoperitoneum (5), bleed and abscess (1, protocol violation), bleed (1)
	IG (15)
	PSG (5)
	SS (2)
Khashab et al., 2017 (331)	DG (11)	6	NA	NA	NA	37 ± 23	17 ± 16	NA	3 (range, 1–12)	Capnoperitoneum (1), ulcer (1, conservatively managed)
	IG (7)
	PSG (12)
Dacha et al., 2017 (329)	DG (9)	8 ± 3	3.4 ± 0.5	1.4 ± 0.9	<0.01	63	25	0.007	2.5 ± 0.7	None
	IG (6)
	PSG (1)
Rodriguez et al., 2017 (333)	DG (12)	1	4.6 ± 0.9	3.3 ± 1.4	<0.001	37 ± 25	20 ± 26	0.03	1 (range, 0–4)	Death (1, autopsy-proven unrelated)
	IG (27)
	PSG (8)
Malik et al., 2018 (332)	DG (1)	3	2.1 ± 0.8	1.9 ± 1.0	NS	49	33	0.1	2.5 ± 1.4	Pulmonary embolism (1, previous history of deep vein thrombosis and pulmonary embolism)
	IG (4)
	PSG (8)

Abbreviations: DG, diabetic gastroparesis; IG, idiopathic gastroparesis; NA, not applicable; NS, not significant; PSG, postsurgical gastroparesis; SS, systemic sclerosis.

Surgical therapy for gastroparesis

Pyloroplasty

Initially, surgical pyloroplasty, which is designed to increase the GE rate, was performed in patients who had both GERD and gastroparesis. In retrospective studies, a Heineke–Mikulicz pyloroplasty improved GE times and also, in some patients, improved symptoms in patients with gastroparesis (334–338) (Table 8). However, most studies did not evaluate symptoms with validated assessments (334, 336, 338). Hibbard et al. (334) demonstrated a reduction in average GE half-time (from 320 minutes to 112 minutes) and the need for prokinetic medications together with a significant reduction in the symptom severity score (335), which has not been validated in gastroparesis. The most comprehensive retrospective series followed patients for >1 year after pyloroplasty at a tertiary referral center (337). GE, which was documented with a variety of techniques, improved in almost all patients (90%) and normalized in 60% (337). The severity of symptoms, evaluated with the Gastroparesis Cardinal Symptom Index decreased significantly from 33.8 ± 6.6 to 12.4 ± 8.4 (mean ± SD). However, loss of appetite, stomach fullness, and feeling excessively full after meals were unaffected by the procedure.

Table 8.

Selected Studies Regarding the Efficacy of Surgical Pyloroplasty on Symptoms and GE in Patients with Gastroparesis

Study	Etiology (N)	Follow-Up	Mean Gastroparesis Cardinal Symptoms Index Score			Gastric Emptying Half-Time (GE t1/2)			Complications (n)
			Before	After	P Value	Before	After	P Value
Hibbard et al., 2011 (334)a	DG (7)	3 mo	Significant reduction in all aspects of the symptom severity score (335) at 3 mo: nausea (P < 0.001), vomiting (P < 0.001), bloating (P < 0.001), abdominal pain (P < 0.001), and GERD symptoms (P = 0.013)			320	112	<0.001	No major surgical complications
	NDG (21)
Toro et al., 2012 (336)	DG (5)	3 mo	Postoperative symptom improvement was reported by 82% of the patients (P < 0.001)			180	60	<0.001	No intraoperative complications. During follow-up period, two patients were identified with superior mesenteric artery syndrome
	NDG (45)
Mancini et al., 2015 (337)	DG (15)	>12 mo	33.8 ± 6.6	12.4 ± 8.7	<0.005	Reduction in median GE t1/2 in 13 matched pairs = 76 min (P < 0.001); 4 h retention in 9 matched pairs reduced from 30% to 10% (P = 0.04)			Suture line leak (1)
	NDG (31)
Shada et al., 2016 (338)a	Not stated (177)	6 mo	Significant reduction in heartburn, reflux, abdominal pain, dysphagia, gas bloat, nausea, and emesis (all P < 0.001). No significant reduction in satiety (P = 0.14)			167	74	<0.001	Return to OR for suspected leak (4, 2 confirmed), pulmonary embolus (1), suture line bleeding (1) and, wound infection (4)

Abbreviations: DG, diabetic gastroparesis, NDG, nondiabetic gastroparesis; OR, operating room; t_1/2, half-time.

^aProbable overlapping cohorts.

Gastrectomy

Gastrectomy is a seldom used, last resort for diabetic gastroparesis. In one case series, the symptoms improved, weight stabilized, and the incidence of hospitalizations declined after subtotal (70%) gastrectomy in four patients with diabetic gastroparesis (339). In another cohort, symptoms, assessed subjectively, improved in six of seven patients with refractory diabetic gastroparesis (340). In a mixed cohort of patients with obesity with gastroparesis (two diabetic, five idiopathic), severity and frequency of symptoms improved by 50%. However, of the two patients with diabetes, “one had partial remission and one had improvement with less medication use” (341). In a retrospective comparison of GES and gastrectomy, 87% of patients, of whom 12 had diabetic gastroparesis, reported improved symptoms after a gastrectomy (310). However, objective assessments of symptom severity were incomplete. Complications occurred in 23% and death in 3% of patients. In summary, a gastrectomy should rarely, if ever, be considered for diabetic gastroparesis.

Natural History and Outcomes in the Community and in Clinical Practice

The natural history of gastroparesis has been studied in different settings. In Olmsted County, one-third of all incident cases of gastroparesis died within 5 years and another one-third required hospitalization, medications, or tube feeding related to gastroparesis (4). Compared with the Minnesota white population, overall survival was significantly lower in patients with gastroparesis.

Among 86 patients evaluated at a tertiary referral center in Australia, 20 patients (16 had type 1 DM) were reevaluated 12 years after the first study and 13 patients (12 had type 1 DM) were reevaluated ∼25 years after the first study (342, 343). Twenty-five years after the baseline assessment, the GE was not significantly different. However, the correlation between initial and subsequent assessments was limited, as evidenced by a correlation coefficient of 0.56 (342).

In the NIDDK gastroparesis consortium, patients managed at seven tertiary referral centers were better at 48 weeks after the initial assessment. The improvement was comparable in patients with idiopathic and diabetic gastroparesis (110). After excluding approximately one-third of patients who did not complete questionnaires at 48 weeks, the remaining 262 patients had idiopathic (68%) or diabetic gastroparesis (32%). Among patients with DM, GI symptoms, use of total parenteral nutrition, and hospitalizations declined and the QOL, both mental and physical, improved. On average, these patients had gained 1.52 kg. However, anxiety and depression did not improve. In another study of 34 patients with DM and GI symptoms, these symptoms and glycemic control improved over time (i.e., a mean of 3.2 years later) (25).

A subsequent report compared outcomes in patients with type 1 and type 2 diabetic gastroparesis in the NIDDK cohort described above (116). At baseline, patients with type 1 and type 2 DM had moderate glycemic control (i.e., average HbA_1c of 8.1% in type 1 DM and 7.2% in type 2 DM), severe symptoms of gastroparesis, as evidenced by an average Gastroparesis Cardinal Symptom Index score of ∼3, and several hospitalizations [5.1 ± 6.4 (mean ± SD) in type 1 and 3.2 ± 6.6 in type 2 DM] in the past year for gastroparesis. Gastric retention was moderate or severe in 82.1% patients with type 1 DM and 55.9% of patients with type 2 DM. However, these patients were taking several medications (e.g., antidepressants, pain modulators) that can delay GE; indeed, ∼47% of patients were on opioids. Hence, it is unclear to what extent the delayed GE is explained by diabetic gastroparesis per se, rather than by concomitant medications. Fifty-six percent of patients with type 1 DM and 79% of patients with type 2 DM completed the 48-week visit. Despite increased use of prokinetic drugs, proton pump inhibitors, anxiolytic agents, and implantation of gastric electrical stimulators in 21% of patients, the severity of gastroparesis symptoms did not change significantly in type 1 DM. In comparison, the severity of several symptoms and the mean Gastroparesis Cardinal Symptom Index score, which was lower by an average of 0.2 U, declined in patients with type 2 DM. The number of emergency room visits, hospitalizations, use of total parenteral nutrition, anxiety and depression scores, and QOL did not significantly change in patients with gastroparesis related to type 1 or type 2 DM. The HbA_1c was not correlated with GI symptoms or GE. At 48 weeks, the HbA_1c was not significantly different vs baseline in type 1 or type 2 DM. In summary, among patients with severe gastroparesis who were managed by experienced gastroenterologists at tertiary referral centers, symptoms improved in type 2 but not in type 1 DM patients. However, follow-up was not available in some patients.

Dudekula et al. (28, 344) studied hospitalizations and overall outcomes in gastroparesis. Strikingly, the 326 patients in this cohort “were hospitalized, on average, slightly more than once annually, for about 8 days and underwent 320 endoscopies, 366 computed tomography scans, 390 abdominal X-rays, 90 upper GI contrast studies and 163 GE studies; 37 patients exceeded an annual radiation exposure of 20 mSv at least once. The majority of tests were confirmatory, with results not altering treatment. Vomiting and pain were the most common cause for emergency encounters and diagnostic testing.” They concluded that repeated testing contributes to the cost and risks but had a limited impact on outcome. A second study, derived from an outpatient cohort of 709 patients with gastroparesis, of which 21% had DM, suggests that between 2003 and 2012, treatment was moving away from prokinetics and focusing on symptom-oriented therapy and/or confounding mood disorders.

Conclusions and Future Perspectives

Diabetic gastroparesis is defined by symptoms of variable severity and objective evidence of delayed GE in the absence of gastric outlet obstruction. In addition to scintigraphy, delayed GE can also be diagnosed with the GEBT and a wireless pressure and pH capsule. High-resolution electrogastrogram is a research tool that reveals abnormal initiation and/or propagation of electrical activity in diabetic gastroparesis. Whereas earlier studies suggested that gastroparesis almost exclusively affected patients with complicated type 1 DM, it is now clear that the disease affects patients with type 1 and type 2 DM. Among patients with diabetes, the diabetic phenotype is not different among patients with normal, delayed, and rapid GE. The symptoms of diabetic gastroparesis can be severe; the impact on QOL is comparable to inflammatory bowel disease. Whereas older studies focused on the roles of autonomic neuropathy and hyperglycemia, more recent studies in some animal models and humans have uncovered enteric nervous dysfunctions, including loss of enteric nerves, ICCs, and an immune infiltrate. There is evidence that interactions between the innate immune system, specifically macrophages, and ICCs and the neuromuscular apparatus may mediate gastroparesis in nonobese diabetic mice and in humans. The management relies upon a stepwise approach that incorporates dietary modifications, antiemetic, and prokinetic agents; GES may be considered. Unless patients present with new symptoms, it is not necessary to repeat diagnostic tests in patients with established gastroparesis who present with increased symptoms.

This foundation provides the impetus for future research into the epidemiology, pathophysiology, and management of diabetic gastroenteropathy. Most studies of diabetic gastroenteropathy are relatively small and have been conducted at tertiary centers in patients with moderate or severe symptoms. In contrast, GE disturbances are often asymptomatic. Hence, it is necessary to assess GI sensorimotor functions and their relationship with symptoms preferably in community-based studies. These studies will likely provide a more refined characterization of the magnitude and phenotypes of diabetic gastroenteropathy, the natural history of the condition, which is poorly understood, and its relationship with the diabetic phenotype. From a pathophysiological perspective, it is necessary to continue applying concepts emanating from animal models of diabetic gastroparesis and from other complications of DM in humans (e.g., epigenetic disturbances) to diabetic gastroenteropathy in humans, and to better understand the relationship between hyperglycemia and diabetic gastroenteropathy. In addition to studies that are designed to improve control of glycemia (e.g., newer insulin regimens) and facilitate GE (e.g., endoscopic pyloromyotomy), such insights will provide the impetus for developing novel approaches that target the underlying pathophysiology of the disorder. Stronger collaborations between gastroenterologists, endocrinologists, and basic scientists than currently exist are necessary to tackle these questions.