Clinical trial: alvimopan for the management of post-operative ileus after abdominal surgery: results of an international randomized, double-blind, multicentre, placebo-controlled clinical study


Prof. M. W. Büchler, Chirurgische Universitätsklinik Heidelberg, Im Neuenheimer Feld 110, 69120 Heidelberg, Germany.


Aliment Pharmacol Ther28, 312–325


Background  Post-operative ileus (POI) affects most patients undergoing abdominal surgery.

Aim  To evaluate the effect of alvimopan, a peripherally acting mu-opioid receptor antagonist, on POI by negating the impact of opioids on gastrointestinal (GI) motility without affecting analgesia in patients outside North America.

Methods  Adult subjects undergoing open abdominal surgery (= 911) randomly received oral alvimopan 6 or 12 mg, or placebo, 2 h before, and twice daily following surgery. Opioids were administered as intravenous patient-controlled analgesia (PCA) or bolus injection. Time to recovery of GI function was assessed principally using composite endpoints in subjects undergoing bowel resection (= 738).

Results  A nonsignificant reduction in mean time to tolerate solid food and either first flatus or bowel movement (primary endpoint) was observed for both alvimopan 6 and 12 mg; 8.5 h (95% CI: 0.9, 16.0) and 4.8 h (95% CI: −3.2, 12.8), respectively. However, an exploratory post hoc analysis showed that alvimopan was more effective in the PCA (n = 317) group than in the non-PCA (n = 318) group. Alvimopan was well tolerated and did not reverse analgesia.

Conclusion  Although the significant clinical effect of alvimopan on reducing POI observed in previous trials was not reproduced, this trial suggests potential benefit in bowel resection patients who received PCA.


Post-operative ileus (POI) is defined as a transient impairment of gastrointestinal (GI) motility that results in the failure of ordered peristalsis of intestinal contents. It affects almost all patients undergoing abdominal surgery and often manifests following other surgeries in the vicinity of the abdomen.1–5 The signs and symptoms of POI include abdominal distension and bloating, nausea, vomiting, abdominal cramping and pain, absence of bowel sounds, an inability to tolerate solid food, and delayed passage of, or inability to pass, flatus or stools.2–4, 6, 7 Thus, POI causes increased morbidity7 and reduces patients’ quality of life.8

The pathophysiology of POI is multifactorial involving a combination of pathways and mechanisms. These include disruption of normal hormonal, neural and local factor signalling, which lead to abnormal electrophysiology and motility of the GI tract, and are believed to result primarily from surgical trauma/ bowel manipulation and the post-operative administration of opioid analgesia.4, 6, 9–11 Although opioids are used widely to control post-operative pain, they contribute to the duration and severity of POI by affecting the migrating motor complex, causing nonpropulsive segmental contractions of the small and large intestines, inhibiting intestinal ion and fluid secretion, and increasing transit time.7, 12–14

Historically, POI has been perceived by many healthcare professionals as a normal and expected consequence of abdominal surgery, for which specific medical treatment is unnecessary.4, 6 However, there is growing recognition that POI imposes a substantial burden on patients and healthcare resources, and that effective management is required to reduce the impact and associated costs of this condition.

Alvimopan is a novel, investigational orally administered, peripherally acting mu-opioid receptor (PAM-OR) antagonist that has the potential to reduce the duration of POI without affecting analgesia.15, 16 In phase III clinical studies in subjects undergoing abdominal surgery, alvimopan accelerated the time to GI recovery and decreased the length of hospital stay, without affecting centrally mediated opioid analgesia.17–20 These phase III studies of alvimopan in POI were conducted in centres in the US and Canada. This study was designed to assess the efficacy and safety of alvimopan 6 and 12 mg twice daily (b.d.) vs. placebo in reducing the time to post-operative recovery of GI function in subjects undergoing abdominal surgery outside North America.


Study design and participants

This was a randomized, double-blind, placebo- controlled, parallel-group, phase III clinical study (SB-767905/001) of alvimopan vs. placebo in hospitalized subjects undergoing major abdominal surgery [either elective partial small/large bowel resection (BR) with primary anastomosis or radical total abdominal hysterectomy (rTAH)]. Following a protocol amendment, the subject population was revised to include only subjects undergoing elective partial small/large BR with primary anastomosis (with no further recruitment of rTAH subjects) as greater benefit had been demonstrated in BR subjects after release of data from the US studies.17, 18, 20 The study was conducted at 70 hospitals in 11 countries, predominantly within the European Union (Austria, Belgium, France, Germany, Greece, Poland, Portugal, Spain, Sweden, UK and New Zealand). The study was conducted according to the principles of Good Clinical Practice and the Declaration of Helsinki (1996 version). The study protocol and amendments were approved at each hospital by independent ethics committees, institutional review boards and regulatory authorities. All subjects gave written informed consent before participating in the study.

Subjects were eligible to be included in the study if they were: aged ≥18 years, scheduled for elective partial small or large BR with primary anastomosis to be performed by open laparotomy, scheduled to receive opioids for intraoperative analgesia and post-operative pain management, and scheduled to have the nasogastric tube removed no later than the morning following surgery. Subjects were excluded from the study if they were scheduled to have laparoscopic surgery, a total colectomy, colostomy or ileostomy or a reversal of a stoma, if they had a complete bowel obstruction, or if they were scheduled to receive other treatments or techniques aimed at reducing ileus (such as epidural anaesthetic techniques, epidural opioids, non-opioid epidural anaesthesia, laxatives to stimulate the gut after surgery, or prophylactic use of antiemetics or low-dose naloxone). Subjects were also excluded if they had an American Society of Anesthesiologists (ASA) Physical Status Score of IV or V and were taking or had recently taken opioid analgesics, defined as receiving a course of opioid analgesics for ≥7 days before surgery. Subjects with a clinically significant cardiac arrhythmia or a history of alcohol and/or substance abuse were not permitted to participate in the study. Females who were pregnant, lactating or of child-bearing potential and not using adequate contraception were also excluded.


Following the screening visit prior to the day of surgery, eligible subjects were randomized 1:1:1 (using appropriate centralized computer software) stratified by centres to receive either alvimopan (6 or 12 mg) or matched placebo orally 2 h prior to surgery and then twice daily from the beginning of post-operative day (POD) 1 until hospital discharge, or for a maximum of 7 days of post-operative treatment. The total clinical study duration from signing the informed consent until the last clinical assessment on the day of discharge or day 14 for those still hospitalized was a maximum of 21 days. A follow-up review was conducted approximately 6 weeks postsurgery. The study design is summarized in Figure 1.

Figure 1.

 Study design.

Subjects were encouraged to ambulate ‘as much as can be tolerated’ from day 1 onwards and feeding was advanced as tolerated by the subject. All concomitant medications taken during the study were recorded. Use of nonsteroidal anti-inflammatory drugs (NSAIDs) and other non-opioid analgesics was allowed, unless administered via epidural. Post-operative pain management with opioids was via either intravenous (i.v.) patient-controlled analgesia (PCA) or ward staff-administered i.v. or intramuscular bolus injection. Importantly, this differed from the studies conducted in North America, where virtually all subjects received i.v. PCA for post-operative opioid pain management and the use of some NSAIDs (such as ketorolac or COX-2 inhibitors) was restricted, otherwise the participating hospitals could implement their usual pain management protocols.


Demographic and baseline assessments.  Demographic and baseline assessments included: age, gender, race, height, weight, physical examination, 12-lead ECG, past and current medical conditions and an assessment against the ASA Physical Status Score.

Primary and secondary endpoints.  The primary study endpoint was the time to recovery of GI function (GI3). A subject was considered to have met the primary endpoint when (s)he had first tolerated food (recovery of upper GI function) and either passed flatus or experienced a bowel movement for the first time (recovery of lower GI function). Tolerance of food was defined as the first time the subject was able to eat solid food (any food that required chewing) without vomiting or experiencing significant nausea within 4 h following the meal, or without having to revert to enteral fluids only. The passage of flatus was determined by questioning the subject; the passage of a bowel movement was determined by reference to nursing records or by the clinical judgement of the investigator or designee following questioning the subject.

Secondary endpoints included time to first tolerance of solid food and first bowel movement, whichever of these occurred last (GI2): time until ready for hospital discharge, based solely upon investigator-judged recovery of GI function; time until ready for hospital discharge, based on investigator-judged medical fitness of the subject as a whole; time to authorization of hospital discharge by the investigator; average daily post-operative opioid consumption; and total pre- and intraoperative opioid dosage. It is important to note that time to first flatus is likely to be a less objective and unreliable endpoint than time to first BM because a subject must be conscious and willing to report it.21 Therefore, the GI2 endpoint may provide a more objective assessment of upper and lower GI recovery than GI3 which had been used as a primary endpoint in previous US studies but has been replaced with GI2 for a recently completed study.22

The primary and secondary endpoints were assessed twice daily (morning and evening) from POD 1 until hospital discharge or for a maximum of 14 PODs. The usual routine (nonprotocol specified) post-operative assessments were not recorded on the Case Report Forms.

Safety and tolerability assessments.  Safety was assessed by monitoring the occurrence and severity of unexpected adverse events, clinical laboratory evaluations (haematology, serum chemistry with liver function tests and dipstick urinalysis), ECG and vital signs. Adverse events and vital signs were monitored at screening, pre- and postsurgery, and on all PODs until hospital discharge. Adverse events were also monitored at clinical follow-up (5–7 days after last dose of study medication). Laboratory assessments were performed at the screening visit and at hospital discharge. Data were collected for specified surgery-related adverse events (i.e. nausea and vomiting, post-operative pain, flatulence and constipation) only if these became more severe or occurred with a higher frequency than expected for the subject’s condition.

Subjects who did not complete the course of study medication were considered to have prematurely withdrawn from investigational treatment, but remained under follow-up according to the study protocol unless the subject withdrew their consent. For all subjects who withdrew prematurely, efforts were made to perform a physical examination, 12-lead ECG, record concomitant medication use and opioid use, assess adverse events, obtain blood and urine samples for laboratory evaluations and contact the subject for clinical follow-up. Study medication was discontinued if the nasogastric tube had to remain in place beyond the morning of POD 1 or had to be reinserted. Subjects were withdrawn from the study altogether, if they received the preoperative dose of study medication less than 1 h or more than 4 h before the start of surgery.

Other assessments.  Other assessments included post-operative pain scores, which subjects rated twice daily using a visual analogue scale (VAS) from 0 mm (no pain) to 100 mm (worst pain imaginable). Daily opioid consumption was recorded, including the total number of tablets or capsules taken if the subject progressed to oral opioids in hospital, and total opioid consumption was calculated. Post-operative total daily opioid consumption was calculated in 24-h periods from the end of surgery, defined as the time of placement of the last suture or staple.

Statistical analysis

The original sample size calculation for this study was based on a mixed population (rTAH and BR) which resulted in an estimate of 342 evaluable subjects required to detect a 20% improvement in recovery of GI function by 84 h postsurgery with 90% power. Combined results from three North American studies17, 18, 20 indicated a treatment benefit of alvimopan against placebo represented by a hazard ratio (HR) of 1.33 in BR subjects. On this basis, in this study, it was estimated that a sample size of 240 BR subjects in each treatment group would be required to detect a treatment difference reflecting a HR of 1.33 with 80% power and an overall two-sided significance level of 5%. Assuming 10% of randomized subjects would not be eligible for the modified intent-to-treat (MITT) analysis (see below), a total of 801 BR subjects (267 in each treatment group) were required to be randomized and the protocol was amended.

The primary analyses for all efficacy endpoints were conducted on the BR subject group [BR(MITT)], which consisted of all randomized subjects who received at least one dose of study medication, who underwent protocol-defined BR and had at least one postsurgery GI efficacy assessment. Safety analyses were performed on the All(safety) (subjects who underwent BR or rTAH) population and the BR(safety) population; subjects in the safety populations received at least one dose of study medication and had at least one post-baseline safety assessment.

The Cox-proportional hazards model unadjusted for covariates was used to analyse the effects of treatment on the time to event endpoints. Comparisons between the alvimopan and placebo groups are expressed as HRs with 95% confidence intervals (CIs) and P-values (calculated using the Wald chi-squared test). Significance conclusions were based on Hochberg’s step-up method for controlling the overall Type I error (α) to be ≤5%.23 That is, for multiple comparisons, a P-value of ≤0.025 was required if the other treatment group’s P-value was >0.05. Both comparisons are considered statistically significant, if both P-values are ≤0.05. In addition, the mean time to recovery of GI function was estimated using a Kaplan–Meier (K–M) survival curve, with mean differences and 95% CIs between the alvimopan and placebo groups calculated from the area under the K–M survival curve for each group. Subjects were assigned a censored time for an event if they discontinued prior to recovery of GI function or completed the study without recovery of GI function. Imputation was used for partially missing dates or times to events. Analysis of variance was used to analyse the treatment effect on pain VAS score and daily post-operative opioid consumption.

All statistical analyses were performed using Statistical Analysis Software (sas) version 8.2 on the Uniplexed Information and Computing System (UNIX) platform.

Post hoc analysis

Despite the similarities in the study design between this study and the North American studies in terms of alvimopan dosing and study endpoints enabling data comparisons between studies, there was a key difference in post-operative pain management (route of opioid administration and use of non-opioid analgesics). All subjects in the North American studies were scheduled to receive opioids via PCA, but in the current international study, subjects could receive post-operative opioids via either PCA or ward staff-administered parenteral bolus. Recognition that this could impact the interpretation of the study results prompted a post hoc analysis to be conducted on the BR(MITT) population, which stratified data by whether opioids were administered via PCA or not. As with the analysis in the overall study population, survival curves and mean time to events were estimated from the K–M product limit method. The results of these analyses can only be considered hypotheses generating as they were not preplanned.



A total of 911 subjects were enrolled in the study and randomized to treatment: 302 to alvimopan 6 mg, 304 to alvimopan 12 mg and 305 to placebo. Following the amendment the BR subjects (741) formed the population of interest and the main analysis. Three subjects were excluded from any analyses: one subject allocated to alvimopan 6 mg and one to alvimopan 12 mg did not receive any study medication and a further subject allocated to alvimopan 12 mg had no post-baseline assessments. Of the majority of the remaining 738 BR subjects who formed the BR(safety) population 705 had at least one postsurgery efficacy assessment and comprised the BR(MITT) population. The reasons that 170 subjects were excluded from the BR randomization population were that they underwent rTAH (106), required surgery other than that specified in the protocol (20) or did not undergo surgery at all because of medical or other factors (44). The analysis populations for the primary and post hoc analyses are summarized in Table 1.

Table 1.   Summary of analysis populations
Analysis populationNumber of subjects
PlaceboAlvimopan 6 mgAlvimopan 12 mgTotal
  1. BR, bowel resection; MITT, modified intent-to-treat; PCA, patient-controlled analgesia.

  2. * Subjects who had received ≥1 dose of randomized medication and had ≥1 post-baseline safety assessment.

  3. † Subjects who underwent BR, had received ≥1 dose of randomized medication and had ≥1 post-baseline safety assessment.

  4. ‡ Subjects who were randomized, underwent BR, had received ≥1 dose of randomized medication and had ≥1 postsurgery GI efficacy assessment.

  5. § Includes subjects who received opioid treatment via this route during the first 48 h after surgery, 21 subjects received no post-operative opioids during this period and there was insufficient data to accurately calculate the opioid exposure for 49 subjects.

BR(MITT) non-PCA§104111103318
BR(MITT) i.v. PCA§105101111317

Of the 738 subjects in the BR(safety) population, 565 (77%) completed treatment and 582 (79%) completed the study. Protocol violation, adverse events and consent withdrawal were the most common reasons for both study withdrawal and treatment discontinuation with re-insertion of a nasogastric tube also leading to discontinuation of the study medication for a number of subjects. The disposition of subjects in each of the treatment groups is summarized in Figure 2.

Figure 2.

 Flow chart describing progress of participants through trial.

The demographic and baseline characteristics of the BR(safety) population were comparable for subjects in all three treatment groups (Table 2). The anatomical site of the varietal surgical procedures, which might impact the degree of bowel manipulation, was categorized (small bowel, large bowel – left, right, transverse, sigmoid, others); no differences in the distribution of these procedures between the treatment groups were observed. The most common reason for BR surgery was malignancy. In the post hoc analysis, the demographic characteristics were similar in subjects who received post-operative opioids via PCA and those who did not. In the BR(safety) population, the most common concomitant opioid medications were: fentanyl (483 subjects, 65%), morphine (335 subjects, 45%), pethidine (188 subjects, 26%), piritramide (46 subjects, 18%) and tramadol (105 subjects, 15%). Subject compliance with study medication was high across all treatment groups, with a mean of 94% for the alvimopan 6-mg or placebo group and 95% for the alvimopan 12-mg group [BR(safety) population]. The mean duration of treatment and mean number of doses of randomised medication were similar in subjects across the three treatment groups (6.5 days and 11.4 doses for the alvimopan 6-mg group; 6.5 days and 11.6 doses for the alvimopan 12-mg group; 6.4 days and 11.3 doses for the placebo group).

Table 2.   Subject demographics and baseline characteristics [BR(safety) population]
 Placebo (n = 242)Alvimopan 6 mg (n = 247)Alvimopan 12 mg (n = 249)
  1. Data are presented as number (%) of subjects unless otherwise stated

Mean (s.d.) age (years)62.4 (11.9)65.2 (12.5)64.5 (12.6)
White239 (99)243 (98)245 (98)
Female109 (45)110 (45)112 (45)
Mean (s.d.) BMI (kg/m2)26.6 (4.6)23.1 (4.2)26.4 (4.5)
 Small bowel resection 13 (5)15 (6)10 (4)
 Large bowel resection – left23 (10)35 (14)34 (14)
 Large bowel resection – right83 (34)87 (35)89 (36)
 Large bowel resection – transverse6 (2)6 (2)5 (2)
 Large bowel resection – sigmoid81 (33)76 (31)81 (33)
 Large bowel resection – other36 (15)28 (11)30 (12)
 Large bowel resection – all229 (95)232 (94)239 (96)
Mean duration of surgery (s.d.) (h)2.6 (1.1)2.6 (1.1)2.6 (1.1)
Surgical history
 Abdominal surgery91 (38)99 (40)85 (34)
 Pelvic surgery48 (20)46 (19)51 (20)
 Other65 (27)70 (28)74 (30)
GI history
 Malignancy18 (7)22 (9)31 (12)
 Inflammatory bowel disease18 (7)16 (6)14 (6)
 Gastro-oesophageal reflux disease21 (9)15 (6)23 (9)
 Irritable bowel syndrome3 (1)4 (2)6 (2)
 Other functional disorders19 (8)13 (5)15 (6)
 Other42 (17)40 (16)32 (13)
Primary reason for planned surgery
 Malignancy175 (72)190 (77)197 (79)
 Inflammatory bowel disease22 (9)24 (10)18 (7)
 Other45 (19)33 (17)34 (14)


Primary endpoint.  For the total BR(MITT) population mean time to recovery of GI function (GI3 – first tolerance of solid food and first passing of flatus or bowel movement) was reduced by 8.5 and 4.8 h in the alvimopan 6- and 12-mg groups, respectively, compared with placebo (P = 0.042 for alvimopan 6 mg vs. placebo; P = 0.20 for alvimopan 12 mg vs. placebo, not considered significant under Hochberg correction) (Table 3). There was no statistically significant effect of surgery type or surgery duration on time to GI3.

Table 3.   Mean time to recovery of GI function [BR(MITT) population]
 Placebo (n = 229)Alvimopan 6 mg (n = 237)Alvimopan 12 mg (n = 239)
  1. Primary endpoint = first tolerance of solid food and first passing of flatus or bowel movement.

  2. Secondary endpoint = first tolerance of solid food and first bowel movement, P-values from this analysis were not adjusted for multiple endpoints.

  3. * Nonsignificant; the threshold for statistical significance is P ≤ 0.025 due to application of Hochberg’s method to correct for multiple comparisons.

Primary endpoint (GI3)
 Mean (S.E.) (h)92.6 (3.06)84.2 (2.37)87.8 (2.68)
 Mean difference vs.  placebo (95% CI) (h) −8.5 (−16.0, −0.9)−4.8 (−12.8, 3.2)
  P-value 0.042*0.20
Secondary endpoint (GI2)
 Mean (S.E.) (h)109.5 (3.41)95.2 (2.40)98.8 (2.71)
 Mean difference vs.  placebo (95% CI) (h) −14.3 (−22.4, −6.1)−10.7 (−19.2, −2.2)
  P value (unadjusted) <0.0010.008

Secondary endpoints.  The mean time to the first tolerance of solid food and first bowel movement (GI2) for the total BR(MITT) population was reduced by alvimopan 6 mg (reduction of 14.3 h, P < 0.001 unadjusted for multiple endpoints) and 12 mg (reduction of 10.7 h, P = 0.008 unadjusted), compared with placebo.

The effects of treatment on other secondary endpoints are summarized in Table 4, and none was statistically significant.

Table 4.   Summary of secondary efficacy endpoints [BR(MITT) population]: time to hospital discharge
EndpointPlaceboAlvimopan 6 mgAlvimopan 12 mg
  1. * Calculated from Kaplan–Meier analysis.

  2. † Estimates produced from Cox-proportional hazards analysis, unadjusted for covariates. P-values calculated from Wald chi-squared tests for the treatment contrasts from the Cox model were not adjusted for multiple endpoints.

Time until ready for hospital discharge (on the basis of recovery of GI function)
 Mean time (difference from placebo) (h)*146.1135.8 (−10.4)139.8 (−6.3)
 HR (95% CI)†1.16 (0.96, 1.41)1.11 (0.92, 1.35)
Time until ready for hospital discharge (on the basis of medical fitness)
 Mean time (difference from placebo) (h)*183.6172.0 (−11.5)180.9 (−2.7)
 HR (95% CI)†1.16 (0.95, 1.42)1.02 (0.84, 1.24)
Time to authorization of hospital discharge
 Mean time (difference from placebo) (h)*220.7212.6 (−8.1)214.8 (−5.9)
 HR (95% CI)†1.08 (0.88, 1.31)1.07 (0.88, 1.30)

Post hoc analysis: opioid consumption.  The data on pre-, intra- and post-operative total opioid consumption for the BR(MITT) population from the post hoc analysis are summarized in Table 5. There were no differences between the alvimopan and placebo groups in the total opioid consumption in the pre-, intra- and post-operative periods.

Table 5.   Total opioid consumption (mg) during the study period [BR(MITT) population]: post hoc analysis
 PlaceboAlvimopan 6 mgAlvimopan 12 mg
nMean (s.d.)nMean (s.d.)nMean (s.d.)
  1. PCA, patient-controlled analgesia.

  2. Data are presented in MS morphine equivalent (mg).

  3. * Opioid consumption in first 48 h after surgery.

Preoperative21621.8 (17.4)23024.5 (19.9)23123.9 (19.1)
Intraoperative22032.1 (32.8)22933.1 (31.9)23032.9 (33.2)
Pre- and intraoperative21554.2 (39.1)22957.5 (41.0)22857.1 (40.0)
 Overall213103.8 (120.0)220109.6 (133.8)223106.0 (127.2)
 PCA*10392.1 (87.4)10091.1 (119.8)10984.3 (73.0)
 Non-PCA*10245.3 (41.7)10749.6 (56.4)10145.7 (35.5)

A majority of opioid consumption took place in the first 48-h postsurgery, corresponding to the period of most intense surgical pain. During this period, VAS pain scores did not differ between the alvimopan and placebo groups. In the overall population, only 45% of subjects used post-operative PCA opioids. Non-opioid analgesic use in the first 48 h after surgery was 69%. In general, opioid consumption decreased substantially in each treatment group during the first few postsurgery days and by postsurgery day 3, median post-operative opioid consumption had fallen to zero in all treatment groups. Although opioid consumption did not differ by treatment group at any time during the study, subjects in the PCA subgroup received approximately twice the quantity of opioids in the 48 h after surgery compared with those received by non-PCA subjects (Table 5). However, the analyses of efficacy by quartiles or deciles of opioid consumption showed no evidence of an effect of opioid dosage on efficacy (data not shown) and the VAS pain scores did not differ between the PCA and non-PCA subgroups (data not shown). Non-opioid analgesic use in the first 48 h after surgery was 72% in the PCA and 70% in the non-PCA subgroups.

Post hoc analysis: recovery of GI function in the PCA and non-PCA subgroups.  The differences in post-operative opioid analgesia between the PCA and non-PCA subgroups appeared to have an impact on the primary endpoint (GI3) of this study as shown in Table 6. In the PCA subgroup, the mean time to GI recovery was reduced by 14.7 h (P = 0.029) and 8.3 h (P = 0.23) in the alvimopan 6- and 12-mg groups, respectively, compared with placebo. In the non-PCA subgroup, almost no effect was observed in the mean time to GI recovery; a reduction of 1.5 h and an increase of 1.1 h in the alvimopan 6- and 12-mg groups, respectively, compared with placebo. Reduction in the mean time to GI recovery according to the definition of the secondary endpoint (GI2) appeared to be better than placebo in both alvimopan groups in the PCA subgroup [21.3 h (P = 0.001) and 15.1 h (P = 0.023) for alvimopan 6 and 12 mg, respectively, vs. placebo], but no reductions in mean time to GI recovery were observed in the non-PCA subgroup (Table 6).

Table 6.   Time to recovery of GI function [BR(MITT) population]: post hoc analysis
Mean (h)
Alvimopan 6 mg
HR (95% CI; P-value); mean difference (h)
Alvimopan 12 mg
HR (95% CI; P-value); mean difference (h)
  1. PCA, patient-controlled analgesia.

  2. Primary endpoint = first tolerance of solid food and first passing of flatus or bowel movement.

  3. Secondary endpoint = first tolerance of solid food and first bowel movement.

  4. All observed P-values are exploratory.

  5. * 318 subjects with non-PCA opioids in first 48 h postsurgery: placebo (n = 104), alvimopan 6 mg (n = 111) and alvimopan 12 mg (n = 103).

  6. † 317 subjects with PCA opioids in first 48 h postsurgery: placebo (n = 105), alvimopan 6 mg (n = 101) and alvimopan 12 mg (n = 111).

Primary endpoint (GI3)
 Overall92.61.22 (1.01,1.47; 0.042); −8.51.13 (0.94,1.37; 0.20); −4.8
 Non-PCA subgroup*85.91.04 (0.78, 1.37; 0.797); −1.51.01 (0.76, 1.34; 0.941); 1.1
 PCA subgroup†98.21.37 (1.03, 1.82; 0.029); −14.71.18 (0.90, 1.56; 0.234); −8.3
Secondary endpoint (GI2)
 Overall109.51.39 (1.15, 1.69; <0.001); −14.31.30 (1.07, 1.58; 0.008); −10.7
 Non-PCA subgroup*100.11.16 (0.88, 1.55; 0. 295); −4.51.17 (0.88, 1.56; 0. 289); −3.6
 PCA subgroup†118.71.60 (1.20, 2.15; 0.001); −21.31.39 (1.05, 1.84; 0.023); −15.1


In the BR(safety) population, the treatment groups had similar numbers (%) of subjects with treatment-emergent adverse events (TEAEs) [121 (49%), 137 (55%) and 136 (56%) for the alvimopan 6-mg, alvimopan 12-mg and placebo groups, respectively] or drug-related TEAEs [23 (9%) and 20 (8%) for the alvimopan 6- and 12-mg groups, and 25 [10%] in the placebo group]. The most commonly reported TEAEs (reported by 5% or more subjects in any treatment group) were nausea, vomiting and pyrexia (Table 7). Nausea, vomiting, diarrhoea and hypertension were the most common drug-related TEAEs, and nausea and vomiting were the only AEs reported by more than two subjects in any group that led to premature discontinuation of study drug in the BR(safety) population.

Table 7.   Most common treatment-emergent adverse events occurring in ≥5% of subjects in any treatment group [BR(safety) population]
 Number (%) of subjects
Placebo (n = 292)Alvimopan 6 mg (n = 294)Alvimopan 12 mg (n = 297)
Nausea19 (8)22 (9)25 (10)
Vomiting19 (8)16 (6)12 (5)
Pyrexia17 (7)11 (4)14 (6)

Serious adverse events were reported by 32 (13%) subjects in the alvimopan 6-mg group, 38 (15%) subjects in the alvimopan 12-mg group and 28 (12%) subjects in the placebo group. The most common serious adverse events were post-operative infection [4 (2%), 3 (1%), 1 (<1%)], anastomotic leak [3 (1%), 2 (<1%), 2 (<1%)], wound dehiscence [2 (<1%), 3 (1%), 2 (<1%)] and abdominal pain [0, 4 (2%), 0] for the alvimopan 6-mg, alvimopan 12-mg and placebo groups, respectively. The serious adverse events reported in the rTAH group were not different in nature or frequency: two (6%) in the alvimopan 6-mg group and five (39%) in the placebo group with no specific event occurring in more than one subject. No serious adverse events were reported in the alvimopan 12-mg rTAH group. A total of seven subjects died during the study overall, six (0.8%) undergoing BR (placebo – three: peritonitis, upper GI haemorrhage and one death for which no cause was established; alvimopan 6 mg – one: multi-organ failure; and alvimopan 12 mg – two: sudden death and cerebrovascular accident); none of the deaths were considered to be study drug-related.

The percentages of subjects in the BR(safety) population with abnormal laboratory or vital sign values were low, and were comparable across the three treatment groups. Subjects with 12-lead ECG findings considered that changed from normal or abnormal but not of clinical significance at baseline to clinically significantly abnormal at the day 14 assessment or at discharge if sooner were few in each treatment group: five (2%) in the placebo group, one (<1%) in the alvimopan 6-mg group and two (1%) in the alvimopan 12-mg group. QTc corrected for heart rate by the Fredericia formula24 was assessed in a subgroup of subjects (approximately 50 subjects/group) at baseline and at days 3 and 7 postsurgery. Two subjects, one in each of the alvimopan 6- and 12-mg groups had post-baseline QT prolongations which were not considered serious and resolved. Both of these subjects had confounding factors contributing to the QT prolongation and did not appear to indicate a relationship with alvimopan treatment.

Post hoc analysis: safety in the PCA and non-PCA subgroups.  There was no apparent difference between the PCA and non-PCA subgroups, in terms of the number of serious adverse events. There was a small numerical increase in adverse events, including nausea, hypotension and hypoxia, in the PCA compared with the non-PCA subgroup. However, there was no evidence of a clinically significant effect of alvimopan with respect to the incidence of these adverse events.


This was the first large trial investigating the impact of alvimopan on POI after open abdominal surgery outside North America. It demonstrated that alvimopan, administered at doses of 6 and 12 mg b.d., had a statistically nonsignificant effect on accelerating GI recovery following BR as measured by the primary composite endpoint GI3 (first tolerance of solid food and first passing of flatus or bowel movement).

Although there is no standardized and widely-used definition of ‘recovery of GI function’, recovery of the upper GI tract is generally believed to have occurred once a patient can chew and swallow solid food without sequelae. Assessment of recovery of lower GI tract function is more controversial and has utilized measures such as the appearance of bowel sounds and the passage of flatus which are open to subjective interpretation. A more objective measure is the passage of stool. The North American trials,17, 18, 20 on which the design of the international trial was based, initially selected GI3 as the primary endpoint, which was not always achieved as the primary analysis. However, careful interrogation of the placebo data from these trials led to the conclusion that variability was introduced when flatus was used in the GI3 composite and that this did not add to the clinical interpretation. GI2 was subsequently selected as the primary endpoint in a later North American trial as a more clinically relevant measure of GI recovery.22 In this study, the secondary endpoint GI2, provided an assessment of GI recovery that was not dependent on subject reports of flatus, and the mean times to this endpoint were significantly shorter in the alvimopan groups than in the placebo group.

On the basis of post hoc analysis, it would appear that the clinical benefits of alvimopan may be most apparent in the subgroup of subjects who received opioids for post-operative pain management via i.v. PCA for at least the first 48-h postsurgery. In this cohort of subjects, GI recovery was significantly faster than placebo as measured either by the primary GI3 endpoint with alvimopan 6 mg or the secondary GI2 endpoint with both doses of alvimopan. Indeed, the magnitude of the treatment response in the PCA cohort was similar to that seen in the earlier North American studies.17, 18, 20 There was no difference in VAS pain scores between the treatment groups in the first 48 h postsurgery, when pain was most intense, indicating that alvimopan did not reverse opioid analgesia. Moreover, alvimopan was well tolerated, with a safety profile comparable to that of placebo.

The opioid and non-opioid analgesic consumption data in this study are indicative of an overall opioid-sparing approach to pain management (in an effort to bring about an earlier resolution of POI4, 7, 25). In the first 48-h postsurgery when surgical pain was the most intense, non-opioid analgesics were used by 69% of subjects. This is substantially higher than in the earlier North American studies of alvimopan in POI, in which the consumption of non-opioid analgesics for the same period was <4%17, 18, 20 (Adolor Corporation, pers. comm.).

The results of this study suggest that time to GI recovery after surgery is affected by the use of opioids for analgesia. This is consistent with the role of opioids in the patho-aetiology of POI. Further exploratory analyses indicate that subjects receiving post-operative opioids via PCA experienced a longer time to recovery of GI function than those receiving pain relief via non-PCA methods of opioid administration. However, there is strong evidence that subjects prefer PCA to alternative forms of post-operative analgesia,26–28 as it allows them to control both the timing of administration of analgesia and the quantity received. It is noteworthy that in the setting of this clinical trial, the response of PCA-treated subjects to alvimopan with respect to recovery of GI function was consistent with the results reported in the North American trials where most of the subjects used PCA.

Despite growing recognition of POI as a clinical problem for which appropriate management strategies should be implemented, there is no agreed definition of what constitutes POI, and approaches to its minimization and management vary widely from country to country.20 Current approaches to the management of POI are either preventive or supportive. Techniques employed to prevent POI or limit its impact include minimally invasive surgery, preoperative preparation, and decreased use of opioid analgesics. Supportive measures adopted in clinical practice comprise nonpharmacological (e.g. nasogastric intubation, early enteral feeding and early mobilization) as well as pharmacological (mainly the use of prokinetic agents or laxatives) strategies. However, none of the current management approaches, used individually, has consistently been effective in accelerating the resolution of POI,29–38 while maintaining adequate pain relief. New effective treatments for POI are needed.

POI is a common cause of prolonged hospitalization after abdominal surgery and is a significant factor for hospital readmissions.6, 39, 40 A large cross-sectional study estimated that POI increased hospital length of stay (LOS) by approximately 3 days in subjects undergoing abdominal surgery.40 Increased hospital LOS caused by POI is associated with increased health care costs, lost productivity and increased complications.7, 40, 41 Accelerating GI recovery and decreasing hospital LOS are important objectives for the management of POI.

In this study, alvimopan had no benefit on all three measures of the time to hospital discharge. In contrast, patients treated with alvimopan in the North American studies17, 18, 20 were able to leave hospital (discharge order being written) in a significantly shorter time than those on placebo. We also noted that the time to hospital discharge order written in the placebo group in this study was some 3–4 days longer than in the placebo groups of the North American studies, and perhaps the observed differences in times to hospital discharge may be explained in terms of differences in social and financial pressures between Europe and North America with respect to factors that contribute to discharge decisions. Nevertheless, the times to discharge in this study were shorter than reported in BR subjects in a recent survey of several European countries.42

In conclusion, there was an indication that both alvimopan 6- and 12-mg doses had an effect on reducing the duration of POI. An exploratory analysis showed that this was more marked in subjects who received PCA, a result that was consistent with that shown in the North American studies of this novel compound.


Declaration of personal interests: Markus Büchler has served as a speaker, a consultant and an advisory board member for GlaxoSmithKline, and has received research funding from GlaxoSmithKline R&D. Yves Flamant and John Monson have served as advisory board members for Glaxo SmithKline. Margaret Byrne, Eric Mortensen and Russell Williamson are employees of GlaxoSmithKline. Jill Altman was an employee of GlaxoSmithKline. Jill Altman, Margaret Byrne, Eric Mortensen and Russell Williamson own shares in GlaxoSmithKline. Christoph Seiler has no conflicts of interests. Declaration of funding interests: This study was funded fully by GlaxoSmithKline R&D. Initial data analyses were undertaken by Gemma Joseph, Janet Perkins and Jacquie Christie who are employees of GlaxoSmithKline. Writing support was provided by Samantha Smith and Jackie Bannister of Gardiner-Caldwell, London and funded by GlaxoSmithKline R&D.


The authors were assisted in manuscript preparation and collation of author contributions by professional medical writers at Gardiner-Caldwell, London.

The following investigators and institutions participated in study 001: Australia: David Scott, St Vincent’s Hospital, Vic.; Tony Chow, Box Hill Hospital, Vic.; Larry McNicol, Austin Medical Centre, Vic.; Afif Hadj, Maroondah Hospital, Vic.; Richard Halliwell, Westmead Hospital, NSW. Belgium: Ignace Vergote, UZ Gasthuisberg, Leuven; Wim Ceelen, UZ Gent; Alex Kartheuser, Cliniques Universitaires Saint-Luc, Brussels. France: Yves Flamant, Hôpital Louis Mourier, Colombes; Marcel Chauvin, Hôpital Ambroise Pare, Boulogne Billancourt; Richard Villet, Hôpital des Diaconesses, Paris; Boris Bryssine, Hôpital Hôtel Dieu, Lyon; Rolland Parc, Hôpital Saint Antoine, Paris; Claude Meistelman, Hôpital Brabois, Vandoeuvre les Nancy; Patrick Van Box Som, Clinique Jeanne d’Arc, Lyon. Germany: Jonas Goehl, Kliniken der Universität Erlangen-Nürnberg, Erlangen; Emre Yekebas, Universitäts-Krankenhaus Hamburg-Eppendorf; Ralf Konopke, Universitätsklinikum Carl Gustav Carus, Dresden; Markus Büchler and Christoph Seiler, Universitätsklinikum Heidelberg; Wolf Bechstein, Universitätsklinikum Frankfurt; Wolfgang Schareck, Universität Rostock; Karl-Walter Jauch, Klinikum der Universität M?nchen; Peter Kujath, Universitätsklinikum Schleswig-Holstein, L?beck; Wolfram Trudo Knöefel, Universitätsklinikum D?sseldorf; Pompiliu Piso, Medizinische Hochschule Hannover; Thomas Mansfeld and Thomas Kienast, Allgemeines Krankenhaus Altona, Hamburg; Arved Weimann, Klinikum Sankt Georg, Leipzig. Greece: George Antzaklis, Sismanoglio Hospital, Athens; Chris Dervenis, General Hospital ‘Ag. Olga’, Athens; Dionisos Voros, Areteon Hospital, Athens; George Papastratis, Areteon Hospital, Athens; Ioannis Kanellos, General Hospital ‘Papanikolaou’, Thessaloniki. New Zealand: Mark Thompson-Fawcett, Dunedin School of Medicine; Michael Booth, North Shore Hospital, Auckland; Janet Ansell, Tauranga Hospital; Ralph Van Dalen, Waikato Hospital, Hamilton; Stephen Purchas, Lower Hutt Hospital, Wellington. Poland: Ireneusz Krasnodebski, Klinika Chirurgii Ogolneij, Warsaw; Mariusz Bidzinski, Instytut, im. M. Sklodowskiej – Curei w Warszawie, Warsaw; Marek Nowacki, Instytut, ul. W.K. Roentgena, Warsaw; Janina Markowska, Klinika Ginekologii Operacyjnej, Poznan; Zbigniew Friebe, Klinika Ginekologii, Poznan; Michal Drews, Klinika Chirurgii Ogólnej, Poznan. Portugal: Francisco Rocha Pires, Hospital Fernando da Fonseca, Amadora; Antonia Nazare, Hospital Fernando da Fonseca, Amadora; Nuno Pinheiro, Hospital Fernando da Fonseca, Amadora; Eduardo Barroso, Hospital Curry Cabral, Lisbon; Joao Giria, Hospital Garcia de Orta, Almada; Francisco Castro e Sousa, Hospitais da Universidade de Coimbra. Spain: Eloy Espin Basany, Hospital Valle de Hebron, Barcelona; Enrique Martinez Molina, Hospital Ramon y Cajal, Madrid; Hector Ortiz, Hospital Virgen del Camino, Pamplona; Javier de Oca Burguete, Hospital de Bellvitge, Barcelona; Jose Antonio Rodriguez Montes, Hospital La Paz, Madrid; Luis Grande Posa, Hospital del Mar, Barcelona; Albert Munoz-Calero, Hospital Gregorio Maranon, Madrid; Salvador Liedo Matoses, Hospital Clinico de Valencia. Sweden: Magnus Wattwil, Universitetssjukhuset, Örebro; Rolf Sandin, Länssjukhuset, Kalmar; Albert Sundberg, Höglandssjukhuset, Eksjö; Lars Gillberg, Centralsjukhuset, Kristianstad; Håkan Samuelsson, Borås lasarett; Per Brunkwall, Universitetssjukhuset, Malmo. UK: Mike Thomas, Bristol Royal Infirmary; Colin Johnson, Southampton General Hospital; Robert Martin, Billinge Hospital, Wigan; Kenneth Hosie, Derriford Hospital, Plymouth; John Monson, Castle Hill Hospital, Hull; Rupert Pullan, Torbay Hospital; Paul Rooney, Royal Liverpool University Hospital.