Safety and efficacy of ghrelin agonist TZP-101 in relieving symptoms in patients with diabetic gastroparesis: a randomized, placebo-controlled study

Authors


  • Presented in part at the Joint International Neuro-gastroenterology and Motility Meeting, Chicago IL, August 28–29, 2009 and the American College of Gastroenterology Annual Meeting, San Diego, CA, October 23–28, 2009.

Address for CorrespondenceGordana Kosutic, MD, Tranzyme Pharma, 4819 Emperor Blvd., Ste 400, Durham, NC 27703, USA.
Tel: 919 313 4765; fax: 919 313-4700;
e-mail: gkosutic@tranzyme.com

Abstract

Background  Gastroparesis, a chronic disorder of abnormal gastric motility, is common in patients with diabetes mellitus. A synthetic, selective ghrelin receptor agonist, TZP-101, is in clinical development for treatment of gastroparesis. This double-blind, randomized, placebo-controlled study evaluated the safety and efficacy of multiple TZP-101 doses in patients with moderate to severe symptomatic diabetic gastroparesis.

Methods  Patients were admitted to the hospital and adaptively randomized to receive a single 30-min intravenous infusion of 20, 40, 80, 160, 320, or 600 μg kg−1 TZP-101, (n = 57) or placebo, (n = 19) for four consecutive days. Symptoms were evaluated daily with the patient-rated Gastroparesis Cardinal Symptom Index (GCSI) and Gastroparesis Symptom Assessment (GSA). Clinicians rated gastroparesis symptoms on treatment day 4.

Key Results  The 80 μg kg−1 dose was identified as the most effective dose. On day 4, there was statistically significant improvement compared with placebo in the severity of GCSI Loss of Appetite and Vomiting scores for that dose group (P = 0.034 and P = 0.006). In addition, at the 80 μg kg−1 dose, the proportion of patients with at least 50% improvement in vomiting score was significantly different (P = 0.019) compared with placebo. Meal-related GSA scores for Postprandial fullness were significantly improved in the 80 μg kg−1 TZP-101 group compared with placebo (P = 0.012). Clinicians rated the 80 μg kg−1 group better improved than placebo for overall symptom assessment (P = 0.047). Safety profiles were similar in the placebo and TZP-101 groups and all doses were well-tolerated.

Conclusions & Inferences  TZP-101 appears to be safe, well-tolerated, and effective at acutely addressing several gastroparesis symptoms.

Gastroparesis is a chronic disorder of abnormal gastric motility characterized by delayed gastric emptying and symptoms of nausea, vomiting, loss of appetite, early satiety, postprandial fullness, and epigastric pain and discomfort. Gastroparesis can occur idiopathically, but is common in patients with diabetes mellitus.1 Up to 50% of patients with type 1 diabetes may experience chronic gastrointestinal symptoms that can diminish quality of life,2 and are associated with impaired quality of life as well as increased morbidity and mortality relative to the general population.3,4 Patients with diabetes are particularly affected by the gastroparesis symptoms of loss of appetite, early satiety, and vomiting, which complicate nutrition and weight management, as well as glycemic control.5,6 Nausea, vomiting, and early satiety are present in approximately 90%, 70–80%, and 60–80%, respectively, of patients with gastroparesis.4,7

Treatment options for gastroparesis are limited and include dietary modifications, and prokinetic and antiemetic therapy. Current prokinetic options are limited by adverse events profiles and poor efficacy. Gastroparesis is termed refractory when symptoms cannot be managed pharmacologically or by diet and patients with refractory gastroparesis may require surgical or other invasive interventions (e.g., gastric electrical stimulation and/or jejunostomy).8–10 Therefore, new therapeutic options for gastroparesis management are needed and compounds targeting novel pathways are in development. Ghrelin, the natural ligand for the growth hormone secretagogue (hGHS-R1a) receptor, is produced primarily in the gastric mucosa and coordinates gastric emptying and motility.11,12 Administration of pharmacologic doses of exogenous ghrelin has been shown to exert significant gastrokinetic effects in patients with both idiopathic and diabetes-associated gastroparesis,13–15 while lower doses that stimulate physiologic GH plasma levels do not alter gastric emptying.16

After exogenous administration, ghrelin has a short half-life17 that limits its therapeutic potential. A synthetic, selective ghrelin receptor agonist, TZP-101, is in clinical development for treatment of dysmotility disorders including gastroparesis and postoperative ileus. TZP-101 is a macrocyclic peptidomimetic with potent binding affinity18 for the human ghrelin receptor (hGHS-R1a). TZP-101 is well-tolerated in healthy subjects and patients with diabetic gastroparesis administered in single or multiple doses.19,20,21 The pharmacokinetic parameters of TZP-101 in patients with gastroparesis and healthy subjects are comparable and there is little accumulation of TZP-101 after multiple dosing.20 TZP-101 accelerates gastric emptying in adults with diabetic gastroparesis21 and also improves gastrointestinal recovery in patients with postoperative ileus following abdominal surgery.22

The objectives of this double-blind, randomized, placebo-controlled study in patients with moderate to severe symptomatic diabetic gastroparesis were to: (i) identify the most effective dose of TZP-101 for evaluation in forthcoming phase 3 trials based on symptomatic and objective improvement criteria, and (ii) determine the safety and efficacy of multiple doses of TZP-101.

Methods

This multicenter, randomized, double-blind, placebo-controlled, dose-ranging study was conducted from September 2007 to January 2009 at eight study centers: four sites in the United States and one site each in Denmark, Norway, Sweden, and the United Kingdom (Clinical Trial Registry# NCT 00612014).23 The study was conducted in a hospital setting with patients admitted for 5 days to receive daily doses of intravenous study medication for 4 days. All participants gave written informed consent and the protocol (TZP-101-CL-G004) was reviewed and approved by the Institutional Review Boards/Ethics Committees for each study site prior to patient screening.

Study patients

Adult men and women (18–80 year) with type 1 or 2 diabetes mellitus and documented diagnosis of gastroparesis were eligible for the study. Confirmation of at least a 3-month history of symptoms consistent with gastroparesis, delayed gastric emptying (by scintigraphy or breath test assessed at screening or within 6 months of the study) and an endoscopy to rule out upper gastrointestinal tract obstructions were required. Patients who had undergone gastric resection, fundoplication, or major abdominal surgery or who had evidence of bowel obstruction in the 12 months prior to the study were excluded. Exclusion criteria also included patients with severe cardiovascular, pulmonary, hepatic, renal disease or eating disorders, or a history of hepatitis B or C, or HIV. Daily use of narcotics and use of medications with clinical evidence of interacting with cytochrome P450 isozyme 3A4 or prolonging the QT interval were prohibited. Women were required to use approved methods of birth control if not postmenopausal or surgically sterile.

Patients completed a pretreatment screening phase to ensure eligibility criteria and determine the severity of gastrointestinal symptoms during the previous 2 weeks. Symptoms were evaluated using the Gastroparesis Cardinal Symptom Index (GCSI), which uses a 6-point scale ranging from none (0) to very severe (5).24 Based on GCSI development data,24 a GCSI total score of ≥2.66 and a Postprandial Fullness/Early Satiety subscale score of ≥3.00 were required for enrollment. Patients completed the Michigan Neuropathy Screening Instrument (MNSI)25 and had cardiovascular autonomic dysfunction measurements that included beat-to-beat heart rate variation (abnormal defined as a difference ≤10 beats min−1 in a supine patient breathing 6 times min−1) and orthostatic hypotension evaluation (abnormal defined as a fall in systolic blood pressure of ≥20 mmHg or diastolic blood pressure ≥10 mmHg 3 min after standing up). Other screening assessments included hematology and clinical chemistry (including HbA1c measurements).

Prior to hospital admission, patients completed a 24-h recall GCSI questionnaire26 daily for 4 days to determine baseline gastrointestinal symptom severity. Internet-based or interactive voice response systems were used to randomize patients to receive daily 30-min intravenous infusions of matching placebo or TZP-101 (20, 40, 80, 160, 320, and 600 μg kg−1) coincident with the start of breakfast for 4 days in a row. For the inpatient portion of the study, patients received meals tailored to their individual caloric needs as determined for diet maintenance by a dietician. Throughout the study, insulin was administered to patients as required.

Placebo and active study drug were colorless solutions administered in identical volumes. A pharmacist at each site who was not blinded prepared the study treatments. All other clinical personnel were blinded to group assignment. Dosing compliance was monitored by the study nurse administering study drug.

Study endpoints

The primary endpoint was the 4-day mean change from baseline for the GCSI Postprandial Fullness/Early Satiety subscale score. The GCSI was completed each evening to evaluate the symptoms from the previous 24 h.24 Baseline was defined as the average of the four daily GCSI Postprandial Fullness/Early Satiety subscale scores prior to hospital admission.

Secondary endpoints with respect to the GCSI scores were the 4-day mean change from baseline in the total score, subscores, and individual symptom scores, and the changes from baseline for any scores on individual dosing days. Percent change from baseline and the proportion of patients with ≥50% improvement from baseline values were also calculated. Similar analyses were performed on GCSI data at the 30-day follow up visit.

During each dosing day, patients completed the modified 6-item meal-related Gastroparesis Symptom Assessment (GSA) just prior to the start of the infusion and breakfast, and then every 30 min until 4 h after the start of the study drug infusion (9 assessments per day).15 Cumulative GSA scores were calculated after administration of each dose, and after all dosings.

Clinicians reported a symptom severity assessment (CRSA, where 0 = none, 1 = mild, 2 = moderate, and 3 = severe) for each patient at screening and after 4 days of dosing (prior to discharge). Gastric emptying was determined (only in a small subset study) the day after the last dosing day prior to discharge and compared with screening gastric emptying rates. Gastric retention was determined via scintigraphy after ingestion of a radio-labeled meal and the data expressed as half-emptying time (upper limit of normal [ULN] = 135 min) and 4-h retention data (ULN = 10%). Half-emptying time was also determined using the acetate breath test (ULN = 100 min). Glucose was not controlled during gastric emptying measurements.

Other endpoints included monitoring bowel movement frequency, meal completion, the need for nasogastric tube insertion, rescue medication use, and hospitalizations due to gastroparesis for 1 month following the last dose of study drug.

Safety

Safety was evaluated by daily assessment of adverse events and injection site reactions. Vital signs, blood glucose, orthostatic hypotension, and 12-lead ECGs were monitored predosing and several times postdosing daily. Clinical and laboratory evaluations including physical examination and hematology were performed the day after final dosing prior to discharge from the hospital. Patient safety was monitored by phone call on day 7 and at a clinic visit on day 30 after administration of the last dose of study medication.

Statistical analyses

The sample size and numbers of patients assigned to each dose was determined by adaptive randomization27 rather than pair-wise comparisons. The primary endpoint, which was the change from baseline in postprandial fullness/early satiety GCSI score subscale, was used for dose–response model and patients were randomized to different doses using an adaptive allocation scheme. For the first 12 patients enrolled in the study, equal numbers of patients were randomly assigned across study dose groups. After that, the randomization probabilities were adjusted to reflect the accumulated data on the primary endpoint (for example, if the effect appeared to be larger in one study dose group, more patients were assigned to that group as enrollment continued). An Interactive Web Response System (IWRS) was used to dynamically randomize patients based on statistical output performed on a weekly basis.

The relationship between change in the subscale score and dose was modeled using a normal dynamic linear model (NDLM) as described by West and Harrison.28 The NDLM did not assume a monotonic relationship and flexibly modeled the dose–response curve. The objective was finding the best dose for the subscale score improvement while validating its superiority over placebo. The prospectively defined adaptive randomization rules were based on two goals: finding the ED90 (the lowest dose achieving 90% of the efficacy seen with the maximum effective dose) and determining the probability for the ED90 to be superior to placebo in a phase 3 study.

The trial was to stop when the sample size cap of 100 subjects was reached, or when sufficient information was available to show that a TZP-101 dose was sufficiently effective, had a high probability of being the ED90 (probability of at least 0.60 or 0.70) and was likely to be superior to placebo in a phase 3 study, or when continuation of the trial was deemed futile.

Bayesian analysis27 of the primary endpoint was the primary statistical analysis for the study. In addition, the primary and all secondary endpoints were analyzed using frequentist inferential analyses27 to test for differences between each TZP-101 dose and placebo. The linear mixed-effects model included subject baseline as a covariate and factors for dose, day and dose by day. These analyses were performed using sas version 9.1 or higher (SAS Institute Inc., Cary, NC, USA). A mixed-effects model was applied for the analysis of TZP-101 dose × day interaction terms for each of the nine individual GCSI symptom scores, for the three GCSI subscales and overall GCSI score. All efficacy analyses were based on two-sided tests with significance levels of < 0.05. The prospectively defined patient population included all randomized patients who received at least one dose of study drug. Missing data were imputed with the last observation carried forward (LOCF). A repeated-measures analysis was used to model subject results by visit. A one-way anova was used to analyze the LOCF for each measure as a function of the study dose and pair-wise and treatment effects analyzed without respect to time. Observed least-squares means and mean change from baseline were generated from the models and presented, along with 95% confidence intervals and P-values associated with the pair-wise comparisons of each dose to placebo.

Results

One hundred and six qualified patients were screened. Twenty-eight patients were not randomized due to lack of documented gastroparesis diagnosis (n = 9), other screening criteria failure (n = 10), or withdrawn consent/other (n = 9). Seventy-eight patients were randomized and 76 patients received study drug in the distribution to treatment groups shown in Table 1. Two randomized patients did not receive study drug due to voluntary withdrawal from the study. All 57 patients in the TZP-101 groups completed the study through the 30-day follow-up assessment. One patient in the placebo group was lost to follow-up.

Table 1.   Summary of patient demographics [number of patients (percent)] and clinical measures (mean ± SD)
ParameterTZP-101 Concentration (μg kg−1)
Placebo (n = 19)20 (n = 8)40 (n = 17)80 (n = 13)160 (n = 6)320 (n = 6)600 (n = 7)
  1. *Upper limit of normal was 135 min for the scintigraphy assay and 100 min for the breath test.

  2. †Two-week recall of symptoms. Six point GCSI symptom severity scale where: 0 = none, 1 = very mild, 2 = mild, 3 = moderate, 4 = severe and 5 = very severe.

Age, years 45.7 ± 12.6 41.4 ± 12.543.4 ± 15.1 44.0 ± 11.0 47.2 ± 11.2 40.8 ± 13.7 48.4 ± 13.8
Gender
 Male  9 (47)  2 (25)  5 (29)  3 (23)  2 (33)  2 (33)  2 (29)
 Female 10 (53)  6 (75) 12 (71) 10 (77)  4 (67)  4(67)  5 (71)
Race
 White 18 (95)  7 (88) 17 (100) 12 (92)  6 (100)  6 (100)  7 (100)
 African American00  0  1 (8)000
 Other0  1 (12)00000
Body mass index, kg m−227.5 ± 5.126.3 ± 5.226.3 ± 6.027.7 ± 5.826.9 ± 3.530.0 ± 3.825.6 ± 4.2
Duration of diabetes, years18.9 ± 12.418.0 ± 9.122.3 ± 9.117.6 ± 12.420.7 ± 10.721.7 ± 10.730.9 ± 18.0
Gastric half-emptying times*, min
Scintigraphy153.0 ± 28.6154.7 ± 18.2159.3 ± 61.1170.5 ± 75.0127.4 ± 10.4184.0 ± 66.6122.8 (n = 1)
Breath test168.6 ± 67.8186.0 ± 32.7172.4 ± 66.6161.6 ± 46.1141.0 ± 27.4162.0 ± 15.1139.3 ± 16.4
Duration of Gastroparesis Symptoms, years6.1 ± 7.12.9 ± 2.24.1 ± 33.52.6 ± 7.87.2 ± 8.44.5 ± 4.62.1 ± 2.0
HbA1c, %8.0 ± 1.69.1 ± 2.78.1 ± 1.98.6 ± 1.78.4 ± 1.29.2 ± 2.19.4 ± 1.5
GCSI scores†
 Total3.7 ± 0.73.5 ± 0.63.5 ± 0.53.5 ± 0.73.7 ± 0.73.7 ± 0.63.6 ± 0.6
 Postprandial fullness/early satiety3.9 ± 0.73.8 ± 0.64.0 ± 0.73.7 ± 1.03.9 ± 0.53.5 ± 0.73.4 ± 1.2

Patient demographics and baseline characteristics are shown in Table 1 and were similar across groups, except for higher proportions of women in the TZP-101 groups compared with the placebo group. Mean patient age ranged from 41 to 48 years across groups, and the majority of patients in each group were white. The mean duration of diabetes ranged from 18 to 30 years and duration of gastroparesis symptoms ranged from 2 to 7 years. Fifty-seven patients (75%) had type 1 diabetes mellitus. Sixty-six patients (87%) had signs of or a clearly developed autonomic dysfunction. Mean percent HbA1c was not statistically different for any of the TZP-101 groups compared with placebo and the majority of patients in all groups had less than optimal glucose control. Gastric half-emptying times at screening were consistently above the ULN for both the scintigraphic and breath test methods. The mean GCSI total scores and the Postprandial Fullness/Early Satiety subgroup scores for the 2 weeks prior to the study were comparable between the groups as shown in Table 1, and ranged from 3.4 to 4.0 points, indicating moderate to severe gastroparesis symptoms. Average total GCSI and Postprandial Fullness/Early Satiety subgroup scores from daily assessments for the 4 days prior to admission, used as the baseline for efficacy analyses, also indicated moderate to severe gastroparesis symptoms.

Efficacy

Primary analyses of ED90 The Bayesian analyses of the primary endpoint (Postprandial Fullness/Early Satiety subscale of the GCSI) determining posterior predictive probabilities identified the 80 μg kg−1 TZP-101 dose as the ED90 with a moderate probability of 0.341. However, there was a high probability (0.650) that the ED90 lies between the 40 and 80 μg kg−1 doses. The estimated probability for the 80 μg kg−1 dose to generate a successful phase 3 trial was 0.705. Predictive probabilities for all dose groups are shown in Table 2.

Table 2.   Predictive probabilities for the ED90 and success in the phase 3 (based on the averaged 4-day GCSI Postprandial Fullness/Early Satiety subscale score)
 TZP-101 concentration, μg kg−1
Placebo (n = 19)20 (n = 8)40 (n = 17)80 (n = 13)160 (n = 6)320 (n = 6)600 (n = 7)
  1. Pr ED90: predictive probability that a TZP-101 dose is the ED90; the ED90 is the lowest dose achieving 90% of the efficacy seen with the maximum effective dose.

  2. Pr phase 3: predictive probability that the TZP-101 dose will be successful in a phase 3 trial.

  3. *ED90 and the target dose for phase 3 studies determined using Bayesian analysis.

Pr ED900.1820.3070.341*0.0930.0220.054
Pr phase 3 success0.5200.6930.7050.4380.1900.254

Efficacy analyses  The average change from baseline across the four dosing days in the GCSI Postprandial Fullness/Early Satiety subscale score (primary endpoint) was numerically improved in all groups, but the differences between placebo and TZP-101 were not statistically significant. Mean changes from baseline (SD) were −1.3 (1.1), −1.2 (0.9), −1.8 (1.0), −1.8 (1.2), −1.3 (0.9), −0.9 (1.1), and −1.2 (1.5) for the placebo, 20, 40, 80, 160, 320, and 600 μg kg−1 TZP-101 groups, respectively (a greater negative change indicates more improved scores). The lack of statistical significance was driven primarily by the improvement from baseline in both the placebo and TZP-101 groups on dosing day 1, likely because of the elective hospitalization that triggered placebo responses across all groups immediately upon patient admission.

To better understand potential differences between placebo and TZP-101 effects, dose × day interaction analyses was performed. Scores in the placebo group tended to remain stable on days 2, 3, and 4, while the separation in level of improvement between TZP-101 doses and placebo increased over time as shown for the 80 μg kg−1 dose group in Fig. 1. The dose × day interaction analyses for Loss of Appetite, Early Satiety, Vomiting, Postprandial Fullness/Early Satiety subscale, and overall GCSI score show that patients in the 80 μg kg−1 TZP-101 group had a greater rate of decline in their scores and improvement of symptoms over time than the placebo group. The rate of improvement in scores for the 80 μg kg−1 dose group vs placebo approached significance for vomiting (P = 0.079) and postprandial fullness/early satiety (P = 0.095) (Fig. 1).

Figure 1.

 The dose*day interaction analyses for Loss of Appetite, Early Satiety, Vomiting, Postprandial Fullness/Early Satiety subscale, and overall GCSI score for the placebo and 80 μg kg−1 TZP-101 groups. Lower scores indicate less severe symptoms. P values are associated with differences in the rate of improvement (slopes) between the TZP-101 and placebo groups for each symptom. Patients in the 80 μg kg−1 TZP-101 group had greater rates of decline in their scores indicating greater improvement of symptoms over time compared with the placebo group.

Changes from baseline on day 4 in Loss of Appetite, Early Satiety, Vomiting, Postprandial Fullness/Early Satiety subscale score, and overall GCSI score for the 80 μg kg−1 dose and placebo groups are shown in Fig. 2. Loss of appetite and vomiting scores were significantly different from placebo. In addition, there was a clear trend for improvement in the Postprandial Fullness/Early Satiety subscale with 80 μg kg−1 dose. No other statistical trends or significance were seen for other GCSI individual symptom scores or subscales.

Figure 2.

 Day 4 GCSI mean change from baseline in scores for the Fullness/Satiety Subscale, individual symptoms of Loss of Appetite, Inability to finish a meal, Vomiting, and the Total score. Numbers that are more negative indicate greater improvement in symptoms. The graph to the right shows the mean change from baseline in the scores for Vomiting at the 30-day follow-up assessment. P values are shown comparing TZP-101 to placebo.

The percent improvements from baseline and proportion of patients with at least a 50% improvement on day 4 in Loss of Appetite, Early Satiety, Vomiting, Postprandial Fullness/Early Satiety subscale, and overall GCSI score for the 80 μg kg−1 dose and placebo groups are summarized in Table 3. The percent improvement from baseline, and proportion of patients with at least 50% improvement in vomiting score, were significantly different in the 80 μg kg−1 group compared with placebo and there was a clear trend for improvement in the Postprandial Fullness/Early Satiety subscale score with 80 μg kg−1 dose.

Table 3.   Percent improvement from baseline and the percent of patients with at least a 50% improvement in GCSI individual symptom scores for Loss of Appetite, Early Satiety, Vomiting, the Postprandial Fullness/Early Satiety Subscale, and overall score on day 4
Gastroparesis-related symptom endpointPlacebo, 0 μg kg−1 (n = 19)TZP-101, 80 μg kg−1 (n = 13)P value*
  1. *Determined using a Frequentist MIXED model (incorporating dose, time, dose–time interaction and baseline) to compare each dose to placebo.

  2. †Trend toward statistical significance.

Loss of appetite
 % improvement from baseline51.277.60.126
 % patients with ≥50% improvement41.272.70.137
Early satiety
 % improvement from baseline57.581.30.120
 % patients with ≥50% improvement47.172.70.253
Vomiting
 % improvement from baseline59.497.80.021
 % patients with ≥50% improvement50.01000.019
Postprandial Fullness/Early Satiety Subscale
 % improvement from baseline46.866.60.089†
 % patients with ≥50% improvement41.281.80.054†
Overall GCSI score
 % improvement from baseline51.062.70.434
 % patients with ≥50% improvement52.963.60.705

At the 30-day follow-up assessment, the change from baseline for vomiting was statistically significant for the 80 μg kg−1 TZP-101 group relative to placebo (−0.8 vs 0.4, P = 0.024, Fig. 2). In addition, there was a significant difference in the proportion of patients with ≥50% improvement in vomiting score at the 30-day follow-up assessment: 60.0%vs 15.4% for the 80 μg kg−1 and placebo groups, respectively (P = 0.039). No other statistically significant changes were observed at the follow-up assessment.

The daily meal-related GSA symptom questionnaire completed by patients showed that all meal-related individual symptom severity scores, as well as the cumulative GSA score were rated as less severe in the 80 μg kg−1 TZP-101 dose group compared with placebo on day 4. The difference between the 80 μg kg−1 TZP-101 dose group and placebo was statistically significant for postprandial fullness (P = 0.012). The analyses of dose × day interactions for the GSA scores showed a greater rate of decline (i.e., improvement) over 4 days of dosing in scores for the TZP-101 80 μg kg−1 TZP-101 dose group compared with placebo for all symptoms, similar to the results shown for the GCSI assessment. Statistically significant differences between the rates of decline were observed for postprandial fullness (P = 0.006), post-meal belching (P = 0.049), and bloating (P = 0.012)

Clinician-rated assessments of gastroparesis symptoms completed at the end of treatment showed that the 80 μg kg−1 dose group was better than placebo for overall symptom assessment: mean changes from baseline (where more negative numbers indicate greater improvement) in total scores were −4.5 ± 4.2 and −6.8 ± 2.6 for the placebo and 80 μg kg−1 TZP-101 groups, respectively (P = 0.047). All changes from baseline in individual symptoms were numerically improved in the 80 μg kg−1 TZP-101 group compared with placebo, and differences approached significance for bloating, abdominal distension, and early satiety. Mean changes from baseline for bloating, abdominal distension, early satiety, nausea, and vomiting were −0.7 vs−0.13 (P = 0.082), −0.6 vs−0.12 (P = 0.053), −1.1 vs−1.6 (P = 0.09), −1.3 vs−1.6 (P = 0.49), and −0.8 vs−1.2 (P = 0.15) for the placebo and 80 μg kg−1 TZP-101 groups, respectively.

Gastric emptying was assessed prior to discharge for the final 20 subjects enrolled in the study and compared with the half-emptying time that was recorded at screening. The average percent change (improvement) was 25% and 8% for combined TZP-101 dose groups and the placebo group, respectively. The difference between groups in this small number of patients was not statistically significant. Additional exploratory analysis was performed for the scintigraphy data only. Eleven patients (n = 4 for placebo and n = 7 for all TZP-101 doses) had gastric retention recorded 4 h after eating a solid meal at screening and at discharge. At discharge, all subjects in the all TZP-101 group had normalized gastric emptying rate (<10% retention at 4-h time point). The percent retention in the TZP-101 group was 23.9% at screening and 5.0% at discharge, and the observed change in percent retention from screening to discharge was 80%, P = 0.024 (Fig. 3). Change in percent retention in the placebo group was not statistically significant.

Figure 3.

 Percent retention at the 4 h time point after eating a radiolabeled solid meal for patients in the placebo and TZP-101 (all doses) groups at screening and after 4 days of treatment (postdose). The horizontal dashed line indicates the value for upper limit of normal (10% retention).

A statistically significant increase in number of bowel movements occurred in the 80 μg kg−1 group across the dosing days compared with placebo (P = 0.0032) (not shown). There were no changes in meal consumption between groups. Two of 57 patients treated with TZP-101 (3%) and two of 19 (10.5%) treated with placebo were hospitalized for gastroparesis symptoms within the 30-day follow-up period. No patients required nasogastric tubes and nine patients required rescue medications during follow-up.

Safety

There were no clinically relevant differences between TZP-101 and placebo-treated patients in clinical or laboratory evaluations. Safety profiles were similar in the placebo and TZP-101 groups and all doses appeared to be well-tolerated. Similar percentages of patients receiving placebo or TZP-101 experienced at least one adverse event: 46% (n = 26/57) of patients receiving TZP-101 and 37% (n = 7/19) of patients in the placebo group. None of the adverse events appeared to be dose related. Four of 13 patients (31%) in the 80 μg kg−1 had at least one adverse event. The most common (>2% incidence) adverse events for all TZP-101 doses combined were hyperglycemia (n = 6, 11%), nausea (n = 4, 7%), diarrhea (n = 3, 5%), and abdominal pain, urinary tract infection, nasopharyngitis, back pain, and headache (n = 2, 4% for each event). In the placebo group, each of these adverse events was reported by one patient (6%). Five of the six cases of hyperglycemia in TZP-101-treated patients occurred at one clinical site where glucose values >240 mg dL−1 were classified as adverse events. Hyperglycemia in the remaining patient was associated with a change in diabetes management from insulin pump to subcutaneous insulin injection.

The majority of events in both the placebo (n = 23) and TZP-101 groups (n = 92) were mild or moderate: 11/23 mild (48%) and 2/23 moderate (9%) events in the placebo group and 62/92 mild (67%), and 26/92 (28%) moderate events in all TZP-101 groups. However, severe adverse events were more frequent in the placebo group 43% (10/23) vs 5% (4/92) in TZP-101 dose groups. Of the four severe events in the TZP-101 groups, two were not related and two were possibly related to study drug.

Fourteen serious adverse events occurred in five patients: two (10%) patients in the placebo group reported 11 serious events and three (5%) patients receiving TZP-101 reported three serious events. Two patients in the TZP-101 40 μg kg−1 did not have all four doses due to sinus tachycardia in one patient, and dizziness and poor venous access in the second patient.

The time-averaged ECG analysis of change of baseline for heart rate, PR, QRS, QT, and QTc (Bazett correction) did not identify dose-dependent changes or notable differences from placebo. Further, no evident effect of TZP-101 on these ECG parameters was identified in a time-matched analysis comparing baseline ECG parameters to each subsequent time point where recordings were obtained (45 min, 90 min, and 4-h post infusion, post-final dose, and clinical follow-up visit).

Discussion

This is the first study to show clinically and statistically significant improvements in several gastroparesis-related symptoms in patients with diabetes treated for 4 days with the first-in-class ghrelin agonist TZP-101. A combination of patient and clinician evaluations were used to measure TZP-101 effects on gastroparesis symptoms after meals, daily, at the end of treatment, and 30 days after treatment. The rate of symptom improvement in TZP-101-treated patients increased throughout the 4 days of treatment. Statistically significant symptom improvement in TZP-101-treated compared with placebo-treated patients on day 4 of treatment was observed with one or more of the evaluation tools for overall symptom scores and individual symptoms of postprandial fullness and/or early satiety, nausea and/or vomiting, loss of appetite, abdominal distension, bloating, and belching. After 4 days of dosing, TZP-101 normalized gastric emptying in a small subset of patients with available 4-h scintigraphic data.

A sustained decrease in vomiting score was noted at the 30-day follow-up assessment. The mechanism behind this sustained improvement in vomiting is not known. Similarly, patients with drug-refractory gastroparesis undergoing temporary, short-term gastric electrical stimulation have experienced sustained improvement in vomiting frequency.29

Since patients in this study underwent elective hospitalization, the treatment period was limited to 4 days and was selected to minimize hospital stay rather than optimize efficacy. This limitation will be addressed in future studies, where it is likely that several dosing schedules will be assessed in order to better understand onset of efficacy and determine any potential long-term benefits of TZP-101 treatment.

The patients in the study reflected a population with advanced diabetes as evidenced by poor glucose control, evidence of neuropathy, and gastroparesis with delayed gastric emptying and moderately severe symptoms overall. Postprandial fullness has been described as the primary upper gastrointestinal symptom associated with delayed gastric emptying in patients with both idiopathic and diabetes-related gastroparesis.30,31 The present results and those of a previous study in patients with diabetic gastroparesis21 suggest that this clinical profile may characterize patients with potential to receive benefit from TZP-101 treatment. Improvements in the gastroparesis symptoms of loss of appetite, early satiety, and vomiting can be particularly relevant for diabetes patients, as these symptoms can contribute to malnutrition and poor glycemic control,1 and consequently result in decreased quality of life and poor health-related outcomes.2 Recent published data indicate a statistically significant correlation between nausea/vomiting (measured on GCSI N/V Subscale) and delayed gastric emptying (a key pathophysiological parameter in gastroparesis), underlying importance of those symptoms in assessing the effectiveness of gastroprokinetic drugs.32 In the future, subjects with vomiting-predominance who are not responsive to standard medical therapy may be identified and selected for future studies using a proposed predominant symptom classification system33 for gastroparesis, and this may prove useful for assessing potential benefits of TZP-101.

The Bayesian statistical analysis identified TZP-101 80 μg kg−1 as the most effective dose and dose with the highest probability of success in subsequent phase 3 trials. The TZP-101 dose–response appears to be a non-monotonic curve. The non-monotonic dose–response curves have also been reported for ghrelin peptide and a non-peptide ghrelin agonist with respect to gastric emptying in an animal model,34 and following stimulation of the ghrelin receptor with the synthetic growth hormone secretagogue hexarelin.35

The safety and tolerability of TZP-101 were similar to those described in previous studies.19,21 All TZP-101 doses were well-tolerated and there were no dose-related trends in incidence or severity of adverse events or clinically relevant changes. Extensive cardiovascular safety assessments of TZP-101 have not revealed any issues of concern. The most frequent adverse event in the TZP-101 group was hyperglycemia observed in six patients. These events were described in five patients at one site where glucose values >240 mg dL−1 were conservatively classified as hyperglycemia. Adjustment of insulin pump to subcutaneous injection occurred in the other patient. In previous studies, TZP-101 administration did not alter blood glucose levels in healthy volunteers while patients with diabetes had secure blood glucose control using a hyperinsulinemic-euglycemic clamp technique.19–21

Certain limitations of the study are apparent: elective hospitalization of patients for this study was artificial and resulted in significant placebo effects immediately upon admission to the hospital that particularly impacted day 1 symptom assessments. The observed initial response in the placebo group tended to remain stable for the four treatment days while the separation between TZP-101 doses and placebo in level of improvement for some gastroparesis symptoms increased over time. Placebo effects are most often observed when patients provide subjective evaluation of symptoms, and this phenomenon is well known and documented in gastrointestinal disorders.36 In a clinical trial of patients with diabetic gastroparesis receiving a motilin receptor agonist, 63% of patients on placebo reported a good or excellent global response, which was not different from the responses in the active treatment arms.37 Placebo responses are influenced by both conscious expectation and unconscious conditioning regarding treatment.38 The setting, number, and quality of caregiver/patient interactions also influence placebo responses in patients with gastrointestinal disorders.36,39 Thus, the in-patient setting alone created an environment that may have influenced the subjective assessment of symptom severity provided by patients during the initial day of the present study. Another limitation of the study was that enrollment was closed at 76 patients due to slow accrual. Thus, the statistical power for efficacy analysis may have been diminished in some dose groups. However, the adaptive design used to promote efficiency and limit the number of patient needed across doses remained optimal for the primary analysis of determining the ED90 for phase 3 dose selection.

In summary, the results suggest that TZP-101 is safe, well-tolerated and effective at acutely addressing the spectrum of gastroparesis symptoms. The 80 μg kg−1 dose was identified as the dose with the best probability for success in future phase 3 studies.

Acknowledgments

Funding for the study was provided by Tranzyme Pharma. Medical writing support funded by Tranzyme Pharma was provided by Patrice Ferriola. Jason Connor of Berry Consultants provided support for adaptive design and Bayesian analysis and interpretation. Stat-Tech services provided statistical analysis. Technical support was provided by Thorbjørn Jensen of Cyncron A/S. The authors thank the coordinators, clinical personnel, and research staff at each study site for their contributions to the conduct of the study.

Disclosures

L Shaughnessy and G Kosutic are employees of Tranzyme Pharma and provided study concept, design and supervision, and drafting and revision of the manuscript. JC Pezzullo, N Ejskjaer, and R McCallum are consultants to Tranzyme Pharma. JC Pezzullo provided statistical analyses and participated in critical revision of the manuscript. N Ejskjaer, G Dimcevski, J Wo, PM Hellstrom, LC Gormsen, I Sarosiek, E Søfteland, T Nowak, and R McCallum were clinical investigators for the study who were provided research support by Tranzyme Pharma. They participated in study concept and design, data acquisition, and manuscript drafting and critical revision. All study participants were aware of potential conflicts of interest.

Ancillary