### Results

#### Literature Search

The search strategy and selection process are summarised in Figure 1. There were 1656 unique articles found through the literature search. Of these, 266 abstract and full-text reviews were completed, resulting in a total of 35 eligible randomised controlled trials at low risk of bias enrolling 13 566 patients^{[45–79]} who were included in the systematic review and meta-analysis. (Figure 1) The kappa statistic for the final selection of studies was 0.94. Table S1 contains details of each trial. Chey 2014, Chey (b) 2014,^{[65]} Tack 2015^{[66]} and Tummala (a)^{[67]} 2015 all studied naloxegol 25 mg daily, and Chey (a, c) 2014,^{[65]} Tack (a) 2015^{[66]} and Tummala 2015^{[67]} studied naloxegol 12.5 mg daily. To ensure there was no duplication of patient enrolled in the placebo arms of the two doses of naloxegol, we cut the placebo group to half.

Figure 1.

Schematic representation of all studies evaluated and the studies that were included in the systematic review and meta-analysis based on medication

#### Bowel Function (Continuous Outcomes)

The continuous outcomes assessing bowel function were changes in the Bowel Function Index, spontaneous bowel movements, complete spontaneous bowel movements and the Bristol Stool Form Scale. The data from these pooled results are represented as a numerical increase at the end of the trial as compared to baseline. If the 95% confidence interval crosses 0, it was not significant.

Table 1 summarises the change in bowel functions for each individual medication.

1. Bowel Function Index—Only five studies evaluated the Bowel Function Index; four utilised fixed-dose oxycodone/naloxone^{[60,62–64]} or naloxone alone as treatment of opioid-induced constipation.^{[61]} (Figure S1)

a. When including all studies, there was a 11.2 unit (95% CI 8.53–13.78), *I* ^{2} 98% improvement in the Bowel Function Index with active treatment compared to placebo.

b. Subgroup analysis—studies that utilised oxycodone/naloxone as a fixed-dose combination demonstrated a 13.2 unit improvement (95% CI 11. 08–15.33), *I*^{2} 96%^{[60,62–64]} as compared to studies that utilised naloxone alone 3.9 unit improvement (95% CI 2.0–5.8), *I*^{2} NA.^{[61]}

2. Increase in spontaneous bowel movements per week (Figure 2)

a. Overall analysis included 22 studies and demonstrated an increase of 1.1 spontaneous bowel movements/week (95% CI 0. 78–1.41), *I* ^{2} 99%.^{[45,48–52,54–56,59,65,66,69–71,73–79]}

b. Subgroup analysis demonstrated the greatest improvement in spontaneous bowel movements (≥1 increase in spontaneous bowel movements/week) with naldemedine, naloxone, naloxegol 25 mg, linaclotide, alvimopan and axelopran.

(i) Naldemedine—increase 1.55 spontaneous bowel movements per week (95% CI 1.03–2.06), *I* ^{2} 76%.^{[45,48–51]}

(ii) Naloxone—increase 3.81 spontaneous bowel movements per week (95% CI 1.35–6. 27), *I*^{2} N/A. However, this is based on a single trial.^{[59]}

(iii) Naloxegol 25 mg—increase 1.05 spontaneous bowel movements per week (95% CI 0.78–1.33), *I*^{2} 0%.^{[65,66]}

(iv)Linaclotide—increase 1.33 spontaneous bowel movements per week (95% CI 0.45–2.21), *I*^{2} NA.^{[71]}

(v)Alvimopan—increase of 1.15 spontaneous bowel movements per week (95% CI 0. 45–1.84), *I* ^{2} 95%.^{[75–78]}

(vi)Axelopran—increase of 1. 8 spontaneous bowel movements per week (95% CI 0. 86–2.74), *I*^{2} NA.^{[79]}

3. Increase in number of complete spontaneous bowel movements (Figure S2)

a. Among all 11 studies with this reported outcome, there was a significant increase of 1.47 complete spontaneous bowel movements per week (95% CI 0.97–1.97), *I* ^{2} 100%.^{[45–47,49–51,62–64,74,79]}

b. Subgroup analysis demonstrated greatest improvement in complete spontaneous bowel movements (≥1 increase in complete spontaneous bowel movements per week) with naldemedine, oxycodone/naloxone and axelopran.

(i) Naldemedine results in an increase of 1.24 complete spontaneous bowel movements per week (95% CI 0.98–1.5), *I* ^{2} 28%.^{[45–47,49–51]}

(ii) Oxycodone/naloxone resulted in an increase of 1.9 complete spontaneous bowel movements per week (95% CI 1. 09–2. 71), with *I* ^{2} 100% indicating high heterogeneity.^{[62–64]}

(iii) Axelopran resulted in an increase of 1.7 complete spontaneous bowel movements per week (95% CI 0. 74–1.97), *I*^{2} NA.^{[79]}

4. Bristol stool form scale—based on the 7-point scale (Figure S3).

a. Overall analysis demonstrated an improvement of 0. 29 units (95% CI 0. 19–0.40), *I* ^{2} 47%.^{[56,65,69–71,73,77,78]}

b. Linaclotide resulted in a 0. 75 unit (95% CI 0. 32–1. 18), *I*^{2} NA improvement in the Bristol Stool Form Scale.^{[71]}

c. c.Lubiprostone resulted in a 0.24 unit (95% CI 0.11–0.37), *I* ^{2} 0% improvement in the Bristol Stool Form Scale.

Figure 2.

Forest plot of change in spontaneous bowel movements improvement with each medication used in treatment of opioid-induced constipation

#### Bowel Function (Binary Outcomes)

The remaining bowel function analyses were based on binary data and the results are reported as odds ratios. If the 95% confidence interval crossed 1, the result was not statistically significant.

1. FDA endpoint for spontaneous bowel movements (>3 spontaneous bowel movements per week + increase 1 spontaneous bowel movements per week compared to baseline; Figure 3)

a. Overall, the entire analysis demonstrated that medications had 2.02 OR (95% CI 1.76–2. 32), *I* ^{2} 42% of meeting the FDA endpoint for spontaneous bowel movements.^{[45–47,49–53,55,65,66,70,71,77,78]}

b. Naldemedine (OR 2. 48 [95% CI 1. 87–3.3], *I*^{2} 63%),^{[45–47,49–51]} methylnaltrexone (OR 2. 08 [95% CI 1. 53–2. 81], *I*^{2} 64%),^{[52,53,55]} naloxegol 25 mg (OR 1. 87 [95% CI 1.42–2. 47], *I*^{2} 0%),^{[65,66]} naloxegol 12.5 mg (OR 1.56 [95% CI 1.18–2.06], *I*^{2} 0%)^{[65,66]} and axelopran (OR 3.63 [95% CI 1.63–8.1], *I*^{2} NA),^{[79]} resulted in the highest odds ratio of meeting the FDA spontaneous bowel movements endpoint. Odds ratios for lubiprostone (1 trial OR 1.65 [95% CI 1.01–2.51])^{[70]} was significant, but that for linaclotide (1 trials OR 1. 35 [95% CI 0. 71–2. 54])^{[71]} and alvimopan (OR 1.93 [95% CI 0.94–3.92], *I* ^{2} 80%)77, 78 were not significant compared to placebo.

2. FDA endpoint for complete spontaneous bowel movements (>3 complete spontaneous bowel movements per week + increase 1 complete spontaneous bowel movements per week compared to baseline).

a. Overall, there were fewer studies that assessed this outcome. Overall, all medications resulted in OR 3. 38 (95% CI 2.33–4.9), *I*^{2} 0% of achieving the complete spontaneous bowel movements FDA endpoint as compared to placebo.^{[49,51,62]} (Figure S4).

b. Naldemedine 25 mg had an OR 3.9 (95% CI 2. 28–6. 69), *I*^{2} 0%^{[49,51]} and naloxone had an OR 2.95 (95% CI 1. 76–4.9), *I*^{2} NA^{[62]} of achieving the complete spontaneous bowel movements FDA endpoint as compared to placebo.

Figure 3.

Forest plot of individual medications associated with achieving the FDA endpoint for spontaneous bowel movements (ie at least 3 spontaneous bowel movements per week in 9 of 12 weeks' treatment and increase of 1 spontaneous bowel movements per week over baseline) in treatment of opioid-induced constipation

#### Subgroup Analysis

We conducted meta-regression of continuous and dichotomous bowel functions at the two doses of naloxegol. Meta-regression found a nonsignificant effect for the spontaneous bowel movements FDA endpoint (*P* = 0.36) and the Bristol Stool Form Scale (*P* = 0.23), indicating no difference in efficacy between naloxegol 12.5 and 25 mg. On the other hand, there was a significant increase in spontaneous bowel movements per week with naloxegol 25 mg compared to naloxegol 12.5 mg (*P* = 0.01). However, both doses of naloxegol did result in clinically significant increase in spontaneous bowel movements per week. Studies with naloxegol did not assess complete spontaneous bowel movements or the FDA complete spontaneous bowel movements endpoint.

#### Safety Based on Treatment Emergent Adverse Events, Serious Adverse Events and Major Adverse Cardiovascular Events

Overall, the rates of treatment emergent adverse events (Figure S5)^{[45,49–52,54,57–63,65–70,72,74–78,80,81]} were elevated across all studies, OR 1.23 (95% CI 1.11–1.36), *I*^{2} 31%. Among the individual medications, only naloxegol 25 mg (OR 1.36 [95% CI 1.00–1.84], *I*^{2} 32%) and lubiprostone (OR 1.25 [95% CI 1.06–1.48], *I*^{2} 0%) demonstrated a significant increase in treatment emergent adverse events compared to placebo. There was no increase in major adverse cardiovascular events^{[45,52,65,71,73,78,80]} (Figure S6) among all medications.

Table 2 summarises the major adverse event data found within this systematic review and meta-analysis.

#### Individual Adverse Events

1. Serious adverse events were defined based on trial physician assessment or Medical Dictionary for Regulatory Activities (MeDRA) (Figure 3).

a. Overall analysis of all studies demonstrated no significant increase in serious adverse events, OR 0.95 (95% CI 0. 76–1. 18), *I* ^{2} 14%.^{[45,49–52,57–63,65–71,73,75–78,80,81]}

b. Oxycodone/naloxone was the only medication that had higher rates of serious adverse events compared to placebo, OR 2. 22 (95% CI 1. 08–4.53), *I*^{2} 0%.^{[59–63]} Upon review of the literature, all studies noted a positive association with the adverse events and trial drug. Only one trial detailed the serious adverse events, which included pneumonia, bile duct obstruction, cholecystitis, grand mal convulsion and fall complicated by skin laceration.^{[62]}

2. Abdominal pain (Figure 4)

a. Overall, PAMORAs, promotility agents and secretory agents had an OR 2.24 (95% CI 1.79–2.8), *I* ^{2} 30% of resulting in abdominal pain as compared to placebo.^{[45,50–53,57,58,60,62,64–66,68–70,72,74–81]}

b. Subgroup analysis—Lubiprostone [OR 5.56 (95% CI 2–15.41), *I*^{2} 21%],^{[69,70,72]} methylnaltrexone (OR 2.55 [95% CI 1.27–5.13], *I*^{2} 65%),^{[52,53,57,58,81]} naloxegol 25 mg (OR 3.02 [95% CI 1.87–4.88], *I*^{2} 0%) and naldemedine (OR 3.19 [95% CI 2.07–4.92], *I*^{2} 0%)^{[45,50,51,80]} reported the highest rates of abdominal pain compared to placebo.

c. The lowest rates of abdominal pain were seen with combination oxycodone/naloxone (OR 1.28 [95% CI 0.28–5.87], *I* ^{2} 71%)60, 62, 64 and prucalopride (OR 1.38 [95% CI 0.45–4.22], *I* ^{2} NA%).74

3. Diarrhoea was seen with the highest proportions with the secretory medications (Figure 5).

a. Overall, all medications had a significantly higher rate of diarrhoea compared to placebo, OR 2. 07 (95% CI 1. 71–2.5), *I* ^{2} 17%.^{[45,49,50,52,53,57,58,60–66,68–71,73,75–81]}

b. Secretory medications (particularly lubiprostone) had the highest rates of diarrhoea (OR and 95% CI >2.0) compared to placebo; note that the 95% CI for naldemedine and naloxegol is below 2.0.

(i) Lubiprostone—OR 3. 51 (95% CI 2.34–5.27), *I*^{2} 0%.^{[69,70,73]}

(ii) Naldemedine—OR 2.44 (95% CI 1.8–3.3), *I* ^{2} 0%.

(iii) Naloxegol 25 mg—OR 2.4 (95% CI 1.37–4.19), *I* ^{2} 0%.

4. Medication Discontinuation (Figure 6)

a. All medications evaluated had nonsignificant discontinuation rates (OR 1. 27 [95% CI 0.99–1.63], *I* ^{2} 1313%),^{[49,50,52,57,59,61,65–67,74–81]} except for naloxegol 25 mg (OR 1.94 [95% CI 1.16–3.23], *I* ^{2} 0%).59, 61

5. Opioid Withdrawal—None of the medications evaluated resulted in significant opioid withdrawal symptoms (OR 1. 14 [95% CI 0. 57–2. 25], *I*^{2} 0%).^{[45,62,65,79,80]}

a. Rates of opioid withdrawal were not significant in studies that assessed oxycodone/naloxone (OR 0.1 [95% CI 0.01–1.95], *I* ^{2} NA).62

Figure 4.

Forest plot demonstrating serious adverse events (SAE) among the different medications included in the systematic review and meta-analysis. Oxycodone/naloxone was the only medication with a significantly higher risk of SAE compared to placebo

Figure 5.

Forest plot of adverse effect of abdominal pain in treatment of opioid-induced constipation

Figure 6.

Forest plot of adverse effect of diarrhoea in treatment of opioid-induced constipation

#### Risk of Bias and Trial Quality Using GRADE Criteria

All the included trials were deemed to have low risk of bias. All trials have used adequate randomisation procedures and blinded patients and trial personnel. Most trials (28/37; 76%) had low attrition rate, below 10%. Among those with higher attrition rates, the cause and rates of attrition were similar in the treatment and placebo arms. Details regarding risk of bias assessment are available in Supporting Information.

GRADE criteria was applied for increase in spontaneous bowel movements per week, FDA endpoint of spontaneous bowel movements, serious adverse events and medication discontinuation (Figure 7) for studies with greater than one study for each endpoint. Overall, most of the medications (16/24) met high GRADE criteria for the trial endpoints. Moderate GRADE was assigned to three studies that had high heterogeneity. Only lubiprostone trial was assigned low GRADE based on the width of the confidence interval estimate and heterogeneity. Details for GRADE assessment for the primary outcomes per study drug are shown in Table 3.

Figure 7.

Forest plot of adverse effect of medication discontinuation from treatment of opioid-induced constipation

Aliment Pharmacol Ther. 2020;52(1):37-53. © 2020 Blackwell Publishing