Efficacy of mitemcinal, a motilin agonist, on gastrointestinal symptoms in patients with symptoms suggesting diabetic gastropathy: a randomized, multi-center, placebo-controlled trial


Dr R. W. McCallum, Center for GI Nerve and Muscle Function, University of Kansas Medical Center, Kansas City, KS 56160, USA.
E-mail: shinshiosm@chugai-pharm.co.jp


Background Mitemcinal, an oral motilin agonist, accelerates gastric emptying.

Aim To investigate if mitemcinal was superior to placebo in relief of symptoms attributed to gastroparesis.

Methods In a randomized, double-blind design, 392 insulin-requiring diabetics with symptoms attributable to gastroparesis were treated for 3 months with placebo, mitemcinal 5 or 10 mg bid. On a weekly basis, patients assessed whether there was adequate relief of their gastroparesis symptoms. Patients were classified as Complete Responders (CR) if there were three consecutive positive monthly responses, which required at least 50% of their weekly responses in a month being positive. An Overall Responder (OR) had at least 75% positive weekly responses for the whole treatment period.

Results Mitemcinal 10 mg produced a significantly better response rate than placebo with a 10.6% increase in the OR (< 0.05 vs. placebo). Mitemcinal 10 mg also produced statistically significant increases in the CR and OR in the subgroup identified by baseline body mass index (<35 kg/m2) and haemoglobin A1c (<10%) (< 0.01 vs. placebo). Adverse events did not differ from placebo frequency levels.

Conclusions Mitemcinal can induce a statistically significant response to treatment in a subset of diabetic gastroparesis where future prokinetic clinical trials should be focused.


Gastroparesis is a serious and well-recognized disorder that can occur acutely in response to surgery or medications, chronically as a well-known complication of diabetes mellitus (DM), in an idiopathic form, or following surgery that damaged the vagal nerve. Patients with chronic gastroparesis may be asymptomatic or they may display a variety of symptoms including abdominal distension/bloating, persistent fullness, early satiety, anorexia and nausea with or without vomiting.

Treatment for gastroparesis has involved prokinetic agents that enhance and coordinate gastrointestinal (GI) motility and transit of material in the GI tract (e.g. cisapride, domperidone, metoclopramide and erythromycin).1 Infusion of motilin, an endogenous peptide hormone present in the duodenum, has been shown to accelerate gastric emptying (GE) in patients with diabetic gastroparesis.2 As motilin is a peptide hormone that cannot be administered orally, research has focused on finding motilin receptor agonists that can be administered orally. Erythromycin, one of the macrolide antibiotics, is known as a motilin receptor agonist,3 and studies have shown that erythromycin, administered either intravenously or orally, is capable of accelerating GE in patients with diabetic gastroparesis.4, 5 However, erythromycin has a number of drawbacks—it is not acid stable, its antibacterial properties can disrupt the intestinal bacterial flora and promote resistance in bacterial strains, and it can induce cardiac arrhythmias.6

Other macrolide analogues that possess motilin agonist properties were expected to show improvement in gastroparesis symptoms, but have not consistently done so. The motilin agonist ABT-229 showed lack of efficacy in relieving the symptoms of gastroparesis and functional dyspepsia in patients with or without delayed-GE.7, 8 It has been postulated that a long half life, leading to tachyphylaxis, could have contributed to the lack of therapeutic benefit.9

A macrolide analogue currently in clinical development is mitemcinal,10 which accelerates GE in animals and humans and hence has the potential to clarify whether a motilin agonist can ameliorate gastroparesis symptoms.10, 11 We investigated in a double-blind study if mitemcinal can be efficacious in DM patients with symptoms of gastroparesis, as well as identifying whether various factors [body mass index (BMI), haemoglobin A1c (HbA1c), sex, diabetes type and age] affected the response (symptoms of gastroparesis) to mitemcinal or to placebo. Because obesity12 and poor glycaemic control13 are thought to be independent risk factors for upper GI symptoms, we also examined the efficacy of mitemcinal in the subgroup of patients with BMI < 35 kg/m2 and HbA1c < 10%.


Research ethics

For all participating sites, Institutional Review Board (IRB) approval for the protocol and patient informed consent was obtained before the site began screening patients. Written informed consent was obtained for all patients before any study-related procedures were performed. The trials were carried out according to the guidelines provided by the Declaration of Helsinki and the International Ethical Guidelines for Biomedical Research Involving Human Subjects.

Study design

This randomized, double-blind, placebo-controlled, parallel-group study was conducted in 122 sites in the USA. The study duration was 16 weeks, consisting of a 12-week double-blind phase (treatment with mitemcinal or placebo), followed by a 4-week single-blind phase (treatment with placebo only).

During the 2-week screening period, patients were asked to call in daily to an interactive voice response system (IVRS, an electronic touch-tone telephone system connected to a central database) and the severity of three gastroparesis symptoms—early satiety (defined as feeling, soon after starting to eat, that the stomach is over full); persistent fullness (defined as an unpleasant sensation of slow digestion or persistence of food in the stomach) and bloating or abdominal distension (defined as a tightness located in the upper abdomen)—were recorded. Each symptom was scored using a 5-point scale (0 = none and 4 = very severe). A gastric emptying test (GET) was performed during the screening period unless a documented GET was performed within 6 months prior to screening. No GET was performed during the double-blind period of the study.

Randomization was performed electronically using the IVRS to assign patients in a blinded fashion in a 1:1:1 ratio to treatment (mitemcinal 5 or 10 mg bid or placebo bid). The randomization was stratified according to the result of the GET (delayed-GE or normal). Patients returned to the investigator site monthly for follow-up and to receive study medication for the next month. Patients were instructed to take the study medication with meals.


Eligible patients were males or females, >18 years of age, with insulin-requiring diabetes type 1 or type 2, and with a 3-month history of symptoms attributable to gastroparesis. During the screening period, at least one of the three cardinal gastroparesis symptoms (early satiety, persistent fullness and abdominal distension/bloating) must have met or exceeded a symptom severity score of 1.8. Patients were excluded from the study if they had prior GI surgery, unstable medical or surgical condition, gastroparesis associated with other conditions such as scleroderma or current use of prokinetics or any medications known to affect GE.


Efficacy was assessed by having patients report daily information regarding symptoms by means of the IVRS. On a weekly basis, patients assessed whether they had adequate relief of their gastroparesis symptoms and responded either with 1 = yes or 2 = no. Patient response to the weekly IVRS global symptom assessment (GSA) question was the primary efficacy parameter for the study.

Patients were classified as Monthly Responders (MRs) if at least 50% of their weekly responses to the GSA question were positive for that month (i.e. patients reported that they had experienced adequate relief of their gastroparesis symptoms for that month).2

Primary efficacy endpoints consisted of the rate of Complete Responders (CRs) (defined as patients who were MRs for 3 consecutive months) and the rate of Overall Responders (ORs) (defined as patients who had positive responses for at least 75% of their weekly responses for all 3 months).

Drug plasma concentration for pharmacokinetic (PK) analysis was determined from samples collected during the course of the study. A 2 h (±15 min) post-dose blood sample was drawn at week 4 to assess the highest drug plasma concentration at steady state. A trough blood sample was drawn 8–16 h post-dose at week 12 to determine the lowest drug plasma concentration at steady state.

Safety and tolerability were monitored throughout the trial. At the monthly follow-up visits, patients were asked about adverse events (AEs) and received a physical exam, measurement of vital signs and weight, collection of routine clinical labs and a 12-lead electrocardiography (ECG).

Statistical methods

The analysis of efficacy used the ‘intent-to-treat’ (ITT) population, which included all randomized patients. Efficacy was based on MR, CR and OR rates analysed using the Cochran–Mantel–Haenszel test, with a 5% significance level for overall and pair-wise comparison of mitemcinal vs. placebo. No adjustment was made for inflation of type 1 error due to the multiplicity of testing. Multivariate analysis with a logistic model was conducted to calculate the odds ratio in each subgroup stratified by the indicated threshold to determine the effect on the placebo response. Each subgroup was analysed using the following covariates: diabetes type, gender, GET result, BMI, age, HbA1c, years from diabetes diagnosis, and symptom severity. Because the study is an initial investigation for this indication, additional analyses were incorporated to identify subgroups more likely to benefit from treatment. Specifically, a subgroup analysis was conducted on the patients with BMI <35 kg/m2 and HbA1c <10% to determine mean weekly response, CR and OR rates in this subpopulation. Another subgroup analysis was conducted on patients with delayed or non-delayed GE.


The study was conducted between May 22, 2002 and July 27, 2004 at 122 sites in the USA. A total of 654 patients were screened for this study, 262 patients were screening failures, resulting in 392 patients who were randomized (Figure 1). The subgroup (BMI <35 kg/m2 and HbA1c <10%) analysis involved 247 patients. Most of patients completed the protocol-specified study visits.

Figure 1.

 Trial profile.

Demographics and baseline characteristics

The patient population was adequately balanced across the three treatment groups with regard to demographic characteristics (Table 1). The patients in the subgroup analysis had similar demographic and baseline characteristics as for the overall population (Table 1). The distribution of DM was 61.2% type 1 and 38.8% type 2, and the percentages of type 1 and type 2 were similar across all three treatment groups. The overall mean number of years from diabetes diagnosis to randomization for the study was 21.1 years. The gastroparesis symptoms that were most troublesome for the patients overall were abdominal distension/bloating (92.6%), early satiety (80.6%), nausea (59.7%) and abdominal pain (55.9%). These symptoms were reasonably well balanced across the treatment groups. Overall, 47.7% patients at baseline had not previously taken a prokinetic agent. The most common prokinetic agent previously used was metoclopramide (46.2%), followed by cisapride (16.3%), erythromycin (7.9%), domperidone (4.3%) and tegaserod (2.8%). For the patient population, 86.7% reported a lack of response to any prior prokinetic agent, and this was reasonably balanced across the treatment groups. Finally, only half of the patients included in the study had measurably delayed-GE.

Table 1.   Demographic and Baseline characteristics for the overall patient population (ITT population) and for the subpopulation with BMI <35 kg/m2 and HbA1c < 10%
VariableOverall population (ITT)Subpopulation (BMI < 35 kg/m2 and HbA1c < 10%)
Placebo (n = 131)MitemcinalTotal (n = 392)Placebo (n = 82)MitemcinalTotal (n = 247)
5 mg bid (n = 131)10 mg bid (n = 130)5 mg bid (n = 95)10 mg bid (n = 70)
  1. Abd, abdominal; GE, gastric emptying; HbA1c, haemoglobin A1c; BMI, body mass index; ITT, intent-to-treat.

Age, years
 Mean (s.d.)48.3 (11.6)47.4 (11.7)50.5 (11.7)48.7 (11.7)48.3 (12.6)45.3 (11.9)50.5 (11.2)47.8 (12.1)
Sex, n (%)
 Male47 (35.9)44 (33.6)48 (36.9)139 (35.5)33 (40.2)34 (35.8)27 (38.6)94 (38.1)
 Female84 (64.1)87 (66.4)82 (63.1)253 (64.5)49 (59.8)61 (64.2)43 (61.4)153 (61.9)
BMI, kg/m2
 Mean (s.d.) 31.3 (7.7)29.3 (7.0)31.7 (8.6)30.7 (7.8)27.5 (3.7)26.5 (4.0)27.0 (4.2)27.0 (4.0)
HbA1c, %
 Mean (s.d.)7.9 (1.9)7.6 (1.5)7.6 (1.6)7.7 (1.7)7.4 (1.0)7.5 (1.0)7.2 (1.2)7.4 (1.1)
Diabetes type, n (%)
 Type 182 (62.6)83 (63.4)75 (57.7)240 (61.2)64 (78.0)70 (73.7)48 (68.6)182 (73.7)
 Type 249 (37.4)48 (36.6)55 (42.3)152 (38.8)18 (22.0)25 (26.3)22 (31.4)65 (26.3)
GE status, n (%)
 Delayed64 (48.9)64 (48.9)64 (49.2)192 (49.0)41 (50.0)49 (51.6)38 (54.3)128 (51.8)
 Non-delayed67 (51.1)67 (51.1)66 (50.8)200 (51.0)41 (50.0)46 (48.4)32 (45.7)119 (48.2)
Years since DM diagnosis
 Mean (s.d.)21.3 (11.0)20.8 (11.6)21.3 (13.1)21.1 (11.9)24.9 (10.8)22.1 (11.6)25.0 (13.7)23.9 (12.0)
Symptom score, mean (s.d.)
 Early satiety2.28 (0.72)2.18 (0.82)2.04 (0.86)2.17 (0.81)2.24 (0.78)2.18 (0.86)2.07 (0.92)2.17 (0.85)
 Persistent fullness2.50 (0.71)2.47 (0.76)2.30 (0.68)2.43 (0.72)2.46 (0.76)2.49 (0.82)2.35 (0.65)2.44 (0.75)
 Abd distension/bloating2.58 (0.78)2.56 (0.77)2.45 (0.75)2.53 (0.77)2.49 (0.84)2.53 (0.83)2.48 (0.78)2.50 (0.81)
 Composite2.45 (0.67)2.41 (0.70)2.26 (0.69)2.38 (0.69)2.40 (0.70)2.40 (0.74)2.30 (0.70)2.37 (0.72)

Multivariate analysis of the placebo response

As patients randomized to placebo showed a comparatively high rate of response, ranging from 20% to 50% (Figure 2a), a multivariate analysis of the CR rate in the placebo group was performed. This revealed that BMI 35 kg/m2, followed by male gender, showed the highest odds ratio for placebo response and were seen as significant independent covariates for a greater placebo response (Table 2). DM type, GE status, age and HbA1c all affected placebo response rates without achieving statistical significance, and DM history and symptom severity were not predictive.

Figure 2.

 Complete response (CR), monthly response (MR) and overall response (OR) rate in patients for the intent-to-treat population (a) and for patients with BMI < 35 kg/m2 and haemoglobin A1c < 10% (b). M1 = month 1; M2 = month 2; M3 = month 3; M4 = month 4 (all the patients received placebo in month 4). Tiny numerals just below horizontal axis indicate the numbers of patients who replied to at least 1 weekly global symptom assessment question during the corresponding period. Patients were classified as CR or OR when three consecutive MRs (at least 50% of their weekly responses in a month) or at least 75% of their weekly responses for the whole treatment period, respectively, were positive. The difference in response rates between placebo and mitemcinal groups and the P-value if < 0.01 are indicated in the graph. *< 0.05 vs. Placebo.

Table 2.   Multivariate analysis of complete response rate in placebo group (intent-to-treat population)
 CovariateOdds ratio (95% CI)
  1. Characteristics found to significantly influence placebo response are indicated with italic font.

  2. BMI, body mass index.

BMI≥35 kg/m2vs. <35 kg/m23.082 (1.004–9.464)
GenderMale vs. female2.730 (1.023–7.287)
Type of DiabetesType 2 vs. Type 11.489 (0.503–4.410)
GE Test ResultDelayed vs. non-delayed1.862 (0.722–4.805)
Age≥50 year vs. <50 year1.749 (0.668–4.578)
Haemoglobin A1c≥8% vs. <8%1.219 (0.472–3.146)

Regarding the HbA1c classified by 8% threshold, the odds ratio of 1.219 for placebo response was not significantly different from 1.000 (Table 2). However, if a higher threshold of HbA1c were applied, the outcome would be changed although the sample size would become small. For example, if a threshold of 10% were used, then the odds ratio (in those patients whose BMI <35 kg/m2) increased to 2.05, suggesting that the elevation in HbA1c also correlated with the increase of placebo response rate. Thus, decreased placebo response rates would be observed for female patients with HbA1c <10% and BMI <35 kg/m2.


Treatment with mitemcinal produced a dose-dependent improvement in CR rate in the ITT dataset (Figure 2a). There was a dose-related pattern of increase in the rate of CR from placebo (21.3%) to mitemcinal 5 mg (24.8%) and mitemcinal 10 mg (28.2%); however, the difference in response rate was not great enough to reach statistical significance for the overall three group comparison or for either pair-wise comparison of mitemcinal vs. placebo. Only on the rate of OR did mitemcinal 10 mg produce a significantly better response rate than placebo, with a 10.6% increase (P < 0.05 vs. placebo).

Treatment with mitemcinal produced a dose-dependent improvement in MR (Figure 2a). After 3 months of treatment with mitemcinal 10 mg bid, the MR rate approached 60%.

Comparing the endpoints between CR and OR, the lowest placebo response was seen with OR rates. In the fourth month, when all patients received placebo, the difference between the rate of CR for the patients who previously received placebo and those who previously received mitemcinal 10 mg was −3.1% (Figure 2a).

For the subgroup analysis (BMI <35 kg/m2 and HbA1c <10%), mitemcinal 10 mg bid produced a significantly greater rate of CR (35.3%) compared with placebo (14.6%, P = 0.003; Figure 2b). The MR rates for both the first and third months were also significantly greater for mitemcinal 10 mg bid vs. placebo (month 1: 54.4% vs. 29.3%; month 3: 60.7% vs. 38.8%, P < 0.05 for both comparisons). In the fourth month, when all patients received placebo, the MR rates were indistinguishable for patients who previously received placebo or mitemcinal (mitemcinal 10 mg bid, 50.0%; mitemcinal 5 mg bid, 53.9%; and placebo, 49.2%). The OR rate was also significantly greater for mitemcinal 10 mg bid (33.8%) compared with placebo (9.8%, P = 0.0002).

A subgroup analysis for patients with non-delayed GE (Figure 3a) compared with those with delayed-GE (Figure 3b) showed interesting results. First, the patients with delayed-GE showed a higher placebo response rates (18–58%) compared with patients with non-delayed GE (2–40%). Secondly, a statistically significant effect of mitemcinal on response rate was only observed for patients with non-delayed emptying, although the size of the subgroup was limited. Statistically significant improvement for both dosages of mitemcinal were observed for CR and month 1 response rates. Statistically significant improvement relative to placebo were only observed for the higher dosage (10 mg bid) for the month 3 response rate and the OR rate.

Figure 3.

 Complete response (CR), monthly response (MR) and overall response (OR) rate in patients for the intent-to-treat population with non-delayed gastric emptying (a) and for patients with delayed-gastric emptying (b). M1 = month 1; M2 = month 2; M3 = month 3; M4 = month 4 (all the patients received placebo in month 4). Tiny numerals just below horizontal axis indicate the numbers of patients who replied to at least 1 weekly global symptom assessment question during the corresponding period. Patients were classified as CR or OR when three consecutive MRs (at least 50% of their weekly responses in a month) or at least 75% of their weekly responses for the whole treatment period, respectively, were positive. The difference in response rates between placebo and mitemcinal groups and the P-value if < 0.01 are indicated in the graph. *< 0.05 vs. Placebo.

For each of the three protocol-defined gastroparesis symptoms for each treatment group during the double-blind treatment period, there was a consistent decrease in symptom severity; however, the differences from placebo were not statistically significant for the overall ITT population. The results for the subgroup analysis show a dose-dependent decrease in symptom score, also not achieving statistical significance (Figure 4).

Figure 4.

 Serial changes of symptom composite score in patients with BMI < 35 kg/m2 and haemoglobin A1c < 10%. Symptom composite score was evaluated as an average of three gastroparesis symptoms (early satiety, persistent fullness and abdominal distension/bloating) in which each symptom was scored using a 5-point scale (0 = none and 4 = very severe). BL = baseline; M1 = month 1; M2 = month 2; M3 = month 3; and M4 = month 4. Symbols and bars present mean ± S.E. +< 0.10 vs. placebo.

In the analysis of the response of patients during the fourth month when all patients received placebo, the majority of patients who were CR continued to be responders during month 4 (placebo 88.9%, 5 mg 82.8% and 10 mg 75.0%), and this was reasonably balanced across treatment groups without an apparent dose-related pattern. Return of symptoms during the single-blind placebo period was examined by comparing month 3 with month 4. Both the mean and median changes from month 3 to month 4 were small, and the lack of an increase in symptom severity during month 4 suggests that rebound symptoms did not occur upon termination of treatment with mitemcinal.

Pharmacokinetic results

The peak mean (±s.d.) plasma concentrations of mitemcinal were 0.4 ± 0.5 and 1.3 ± 1.4 ng/mL for 5 mg and 10 mg mitemcinal, respectively. At the same time, the trough mean plasma concentrations of mitemcinal were 0.3 ± 0.4 and 0.6 ± 0.6 ng/mL for 5 mg and 10 mg mitemcinal, respectively. Thus, although trough concentrations were roughly linear with dose, the peak concentrations tripled with the doubling of the dose.

Safety and tolerability

Statistical analysis of all of the safety and tolerability data indicated no significant differences for placebo compared with the different dosages of mitemcinal (P > 0.05). At least one serious AE was experienced by 12.5% of patients in the placebo group, 11.1% of patients in the 5 mg treatment group and 10.8% of patients in the 10 mg group. The percentage of patients with one or more treatment-emergent AEs was balanced across treatment groups (Table 3). A numerical increase in the percentage of patients experiencing frequent bowel movements was observed in patients receiving mitemcinal (5 mg bid, 6.3% and 10 mg bid, 8.3%) compared with those receiving placebo (5.5%). The overall incidence of severe AEs by treatment group, experienced by 18.8% of placebo patients, 15.9% of mitemcinal 5 mg patients and 20.0% of mitemcinal 10 mg patients did not show a dose-dependent pattern.

Table 3.   Summary of most frequent treatment-emergent adverse events (occurring in >6% of safety population in any treatment group)
Adverse event, n (%)Placebo (n = 128)   MitemcinalTotal (n = 374)
5 mg bid (n = 126)10 mg bid (n = 120)
  1. NOS, not otherwise specified.

Hypoglycaemia NOS73 (57.0)81 (64.3)61 (50.8)215 (57.5)
Nausea16 (12.5)14 (11.1)12 (10.0)42 (11.2)
Blood glucose decreased12 (9.4)12 (9.5)13 (10.8)37 (9.9)
Tremor8 (6.3)13 (10.3)11 (9.2)32 (8.6)
Vomiting NOS11 (8.6)8 (6.3)7 (5.8)26 (7.0)
Diarrhoea NOS7 (5.5)8 (6.3)10 (8.3)25 (6.7)
Upper respiratory tract infection NOS9 (7.0)8 (6.3)8 (6.7)25 (6.7)

The most common laboratory abnormality in this study was hyperglycaemia, as expected for a patient population with diabetes. The changes in serum glucose were similar across the treatment groups, suggesting that mitemcinal did not have an effect on serum glucose. There was no evidence of any meaningful effect of mitemcinal on laboratory results, vital signs, physical examination findings or body weight. Finally, mitemcinal did not show any evidence of ECG abnormality.


Results in the patient population suggest that mitemcinal has efficacy in relieving the symptoms of gastroparesis at the doses studied but the findings were not consistently able to achieve statistical significance. The primary endpoints were two measures of efficacy that required different levels of symptom relief, OR required 75% of weekly responses to be positive compared with 50% for CR. Mitemcinal 10 mg produced a significantly better response rate than placebo with a 10.6% increase in the OR (< 0.05 vs. placebo). The placebo response rate for OR (14.2%) was lower than that of CR (21.3%) in the ITT dataset, and the difference from placebo achieved by mitemcinal was larger for OR than for CR and was statistically significant for both the overall population (Figure 2a) and for the subgroup analysis (Figure 2b). The patients in this study did not meet the criteria for irritable bowel syndrome (IBS). Specifically there was no relationship of cramping abdominal pain or eating stress with bowel movement habit after the meal. Furthermore, bowel movements did not relieve the abdominal pain, as required by the Rome II & III criteria of IBS.14, 15

The safety of mitemcinal was similar to that of placebo. The observation of more patients receiving mitemcinal than placebo experiencing an increase in stool frequency would tend to support the idea that mitemcinal was exhibiting a prokinetic effect on the patients. Many patients with diabetic gastropathy have concomitant gut neuropathy, and constipation is common; 16therefore, this prokinetic effect on stool frequency could potentially be beneficial.

The subanalysis results indicated that mitemcinal 10 mg bid was statistically superior to placebo in diabetic gastroparesis patients with a BMI <35 kg/m2 and HbA1c <10%. The strongest evidence of a mitemcinal treatment effect is the dose-related pattern of increase in the rate of CR and its dissipation after the termination of mitemcinal treatment (Figure 2b). For both efficacy and PK parameters, tachyphylaxis was not observed in this clinical study over the 3-month double-blind period. These results suggest that a motilin agonist with PK characteristics different from such predecessors as ABT-229 can improve symptomatic gastroparesis in the specified subpopulation of diabetics (patients with BMI <35 kg/m2 and HbA1c <10%). Because mitemcinal appeared safe and well tolerated at the doses administered in the study, these results suggest that mitemcinal is a promising agent deserving further evaluation as a new treatment for the symptoms of diabetic gastroparesis as well as other gastropathies.

Results of another subanalysis indicated that the symptomatic improvement associated with mitemcinal treatment was more impressive in patients with non-delayed GE than in patients with delayed-GE. This may provide one more piece of evidence suggesting that there is no consistent relationship between delayed-GE and symptoms in patients with gastroparesis.17 This is encouraging because many diabetic patients have symptoms suggestive of gastroparesis but have a non-delayed GE test. Consequently, it is possible that mitemcinal might be appropriate for a broad range of patients rather than simply being confined to the subpopulation of diabetic patients with gastroparesis.

A high placebo response rate is commonly seen in clinical trials assessing symptom relief in functional GI disorders, and it is an ongoing problem.18 Our study demonstrated that placebo response was significantly affected by BMI and gender. Previously, Talley had identified glycaemic control as an important variable in gastroparesis trials.19 Assuming that obese diabetic patients with inadequate glycaemic control could exhibit a large placebo response for relief of symptoms that might obscure treatment responses to drugs, subgroup analysis was performed to assess efficacy of treatment in non-obese diabetic patients with adequate glycaemic control. With the resulting reduction in placebo response, a statistically significant improvement in the efficacy of mitemcinal compared with placebo was observed. However, a caveat about this conclusion is that it is the result of a post hoc analysis and the strategy of selecting a low placebo response rate subgroup may have introduced a bias. Additional trials in this subgroup should address these bias concerns.

In summary, our study found that BMI and gender were independent covariates for a greater placebo response in insulin-requiring diabetics with symptomatic gastroparesis. Mitemcinal 10 mg produced a significantly better response rate than placebo with a 10.6% increase in the OR (< 0.05 vs. placebo). The subpopulation with BMI <35 kg/m2 and HbA1c <10% had a modest placebo effect, but mitemcinal 10 mg bid in this subpopulation gave a statistically significant response and is a promising new therapy. The identification of specific subgroups of patients that might represent an ideal target population for prokinetics is an important observation that should be incorporated into clinical trials assessing putative prokinetic agents.


  • 2

    MRs were defined in the original protocol and statistical analysis plan to have at least two (rather than 50%) positive responses per month. Since a month does not always consist of 4 weeks, the definition was corrected, and all the analyses were revised to reflect this change.


Declaration of personal interests: R. W. McCallum has received research grant support from Chugai Pharmaceutical Co., Ltd., Novartis Pharmaceuticals Corp., Salix Pharmaceuticals, Medtronic, Inc., Dynogen Pharmaceuticals, Inc., and The SmartPill Corporation. Dr McCallum has also served on the advisory board for The SmartPill Corporation, as a consultant for Takeda, Chugai, Novartis and Medtronic, and received honoraria as a speaker from Takeda Pharmaceuticals, Novartis and TAP Pharmaceuticals.

Declaration of funding interests: Funding for this research was provided by Chugai Pharmaceutical Co., Ltd. Editorial support was provided by Carol A. Lewis, PhD, of Wolters Kluwer Health. O. Cynshi is an employee of Chugai Pharmaceutical Co., Ltd.


The following investigators participated in the study:

T. Abell, University of Mississippi Medical Center, Jackson, MS; R. Albery, Lovelace Scientific Resources, Phoenix AZ; K. Amin, Conyers, GA; C. Arauz-Pacheco, Radiant Research, Dallas North, Dallas, TX; D. Arkin, North Atlanta Endocrinology & Diabetes, PC, Lawrenceville, GA; G. August, Columbus, GA; C. J. Baumgartner, Radiant Research, Edina, MN; B. Bleau, Tacoma Digestive Disease Research Center, Tacoma, WA; B. Bode, Atlanta Diabetes Associates, Atlanta, GA; L. Bonner, Retreat Hospital, Richmond, VA; S. Brady, Anchor Research Center, Naples, FL; D. Buth, Professional Research Network of Kansas, Wichita, KS; R. Chasen, Maryland Digestive Disease Research, Laurel, MD; R. Cherlin, Los Gatos, CA; M. Conway, Lovelace Scientific Resources, Inc, Albuquerque, NM; P. Dandona, Diabetes Center of WNY, Buffalo, NY; J. DiBaise, University of Nebraska Medical Center, Omaha, NE; B. Dolin, Gastroenterology, Ltd., Peoria, IL; A. Firek, Loma Linda University, Loma Linda, CA; R. Gaona, Pro-Research Group, San Antonio, TX; S. Garg, Barbara Davis Center for Childhood Diabetes, Denver, CO; S. Gebhart, Emory University School of Medicine, Atlanta, GA; D. Geenen, Wisconsin Center for Advanced Research, Milwaukee, WI; G. Gerety, The Endocrine Group, LLC, Albany, NY; B. J. Goldstein, Jefferson Medical College of Thomas Jefferson University, Philadelphia, PA; R. Graf, Tacoma, WA; R. Hardi, Chevy Chase Clinical Research, Chevy Chase, MD; L. Harstine, Trilogy, Wichita, KS; B. Hoffman, Medical University of South Carolina, Charleston, SC; R. Hogan, GI Associates Research, Jackson, MS; J. Hone, Center for Diabetes and Endocrinology, Arvada, CO; A. K. Jain, Medical Research Institute, Slidell, LA; M. Mazen Jamal, Long Beach VA Medical Center, Long Beach, CA; J. Johanson, Rockford Gastroenterology Associates, Ltd., Rockford, IL; S. Karne, Medical Affiliated Research Center, Inc. (MARC), Huntsville, AL; M. Kipnes, Diabetes & Glandular Disease Research Assoc, San Antonio, TX; E. LaCava, Evergreen Diabetes & Endocrinology Medical Group, Kirkland, WA; G. Ledger, St. John’s Medical Research, Springfield, MO; H. Lilienfeld, Diabetes Care Institute, Tampa, FL; J. Liljenquist, Rocky Mountain Inst. Of Clinical Research, Idaho Falls, ID; L. Loman, Suncoast Clinical Research, New Port Richey, FL; G. Lopez, Associated Medical Group, Battle Creek, MI; A. Malik, Advanced Clinical Research, Ltd., No. Providence, RI; I. Marcadis, Palm Beach Research Center, West Palm Beach, FL; S. Mayeda, The Endocrine Medical Group, Inc., Orange, CA; M. Meredith, University of Wisconsin—Madison, Madison, WI; J. Mersey, MODEL Clinical Research, Baltimore, MD; C. Monder, Arizona Center for Clinical Research, Glendale, AZ; J. Morelli, Stoneboro, PA; E. Morris, Western Montana Clinic, Missoula, MT; N. Murali, Gastroenterology Associates, Orangeburg, SC; B. Neustater, BioQuan Research Group, North Miami, FL; M. Olyaee, Kansas University Medical Center, Kansas City, KS; C. Owyang, University of Michigan Health System, Ann Arbor, MI; C. Page, Endocrinology Consultants of East Tennessee, Knoxville, TN; H. Parkman, Temple University Hospital—GI Section, Philadelphia, PA; R. Peterson, Hyperion Clinical Research, Charleston, WV; S. Plevin, Suncoast Clinical Research, Inc, New Port Richey, FL; J. Popp, Jr., Columbia Gastroenterology Associates, PA, Columbia, SC; E. Portnoy, Westlake Medical Research, Westlake Village, CA; C. Raine, Diabetes Control Center, Orangeburg, SC; R. A. Ramanujan, Diabetic Care Associates, Binghamton, NY; J. Reed, Southeastern Endocrine & Diabetes, Atlanta, GA; D. Riff, AGMG Clinical Research, Anaheim, CA; M. Rodriguez, Gastroenterology Associates of Manatee, Bradenton, FL; R. Rood, Grand Rapids Associated Internists, Grand Rapids, MI; M. Rosenberg, Research Foundation of America, Los Angeles, CA; M. Safdi, Consultants for Clinical Research, Cincinnati, OH; I. Sarosiek, Kansas University Medical Center, Kansas City, KS; C. Schmitt, Southeastern Clinical Research, Chattanooga, TN; M. Shepherd, Endocrinology Consultants, Tupelo, MS; T. W. Sherraden, Tallahassee Endocrine Associates, Tallahassee, FL; H. J. Simon, Physicians Research Center, Toms River, NJ; B. Sindler, Baltimore, MD; D. Stanton, Community Clinical Trials, Orange, CA; E. Stephens, Oregon Health & Science University, Portland, OR; M. Stern, Atlanta Academic Research Group, Decatur, GA; L. Stonesifer, Federal Way, WA; D. Sutton, NE Florida Endocrine & Diabetes Research Center, Jacksonville, FL; M. Swaim, Regional Research Institute, Jackson, TN; S. Swami, Prime Care Clinical Research, Clinton, OK; O. Tatpati, Wichita, KS; A. Thannoun, Pharma Tex Research, Amarillo, TX; T. Tobey, P. Tung, Endocrinology & Diabetes Consultants, PC, Dover, NH; G. B. Waldon, Mercy Health Center, Bentonville, AR; R. Weinstein, Diablo Clinical Research, Inc., Walnut Creek, CA; P. Weissman, Baptist Diabetes Associates, Miami, FL; P. Winkle, Orange County Clinical Research, Cypress, CA; J. Witte, Idaho Gastroenterology Associates, Boise, ID; J. Wo, Digestive Health Center, University of Louisville, Louisville, KY; J. Wolosin, Sharp Rees-Stealy Medical Group, Inc., San Diego, CA; F. Zieve, McGuire VA Medical Center, Richmond, VA.