Introduction

Chronic pain is a symptom that accompanies many conditions including joint pain, fibromyalgia, and osteoarthritis [1, 2]. In industrialized countries, opioids are frequently used alternatives for the treatment of moderate-to-severe chronic nonmalignant pain, second only to nonsteroidal anti-inflammatory agents in terms of prescription frequency [3–6]. A total of 365 million opioid prescriptions were made globally in 2005 [1]. One in five people in Europe suffer from moderate-to-severe chronic non-cancer pain; 25–30% of those are prescribed opioids [3, 4, 7, 8]. One in four people in the United States (US) suffer from moderate-to-severe chronic non-cancer pain, and 15–20% of those groups are prescribed opioids [7, 9, 10]. Chronic non-cancer pain affects a quarter of the Canadian population, a third of which are prescribed strong opioids [11, 12].

Opioid-induced constipation (OIC) is the most common side effect of opioid treatment in patients with cancer and non-cancer pain, and all patients using opioids are at risk of developing OIC [1]. OIC typically begins within the first few days of starting an opioid treatment, and although some patients may experience some relief with dose reduction of the opioid, it typically persists for over the treatment duration [1, 13].

The effects of OIC can cause psychological distress, bodily suffering, and social isolation [1, 14, 15]. The symptoms of OIC include rectal and/or lower abdominal pain, abdominal distension, straining during evacuation, bloating, and flatulence [16–18]. Stool impaction is a potential outcome of OIC that occurs when retained stools form large, firm lumps that are impossible to defecate, which in turn may lead to bowel obstruction, stercoral ulcers, and urinary retention [19]. This stage of OIC necessitates manual disimpaction or even surgery. Psychological distress is caused by OIC as a result of the general distress about the constipation itself as well as depressive symptoms and anticipatory anxiety, which may increase over time [16, 20]. The symptoms of OIC have a considerable negative impact on activities of daily living [1, 14]. It has been reported that constipation is even more common source of distress than the pain originally being treated [21]. In some patients, OIC becomes so severe that patients may taper or even discontinue opioid use in an attempt to relieve their discomfort, as they prefer tolerating their pain rather than suffering from continued bowel dysfunction [1, 14, 15, 18, 22–24]. Moreover, in contrast to other opioid-induced side effects (e.g., nausea), patients rarely develop a tolerance to OIC [1, 14, 17].

Utilities are one particular measure of HRQoL used in economic evaluations, representing preferences for states of health, measured on a scale of 0 [death] to 1 [perfect health] [25]. A number of studies have demonstrated low health state utility values for patients with OIC. For instance, a cross-sectional survey of OIC patients in the US, Canada, Germany, and the United Kingdom (UK) estimated that the mean European Quality of Life (EuroQol)—Five Dimensions questionnaire (EQ-5D) index score was 0.49 and the mean EQ-5D Visual Analogue Scale score was 50.6, measured on a scale of 0 [death] to 100 [perfect health] [2]. These EQ-5D values suggest that patients with OIC have lower utility than patients with conditions, such as myocardial infarction, minor stroke, chronic obstructive pulmonary disease, some high-incidence cancers, and even some end-stage cancers, such as prostate and colorectal cancer (mean EQ-5D range of 0.55–0.76) [26–31].

The low health state utility values in patients with OIC are most likely a result of both the chronic pain and the side effects of opioid treatment including OIC. For instance, patients with chronic lower back pain report EQ-5D scores ranging between 0.53 and 0.64 [32, 33], and those with chronic cancer pain who required constipation therapy had a mean EQ-5D score of 0.49 [34]. To quantify the impact of OIC on a patients’ health state utility, a comparison between patients with and without OIC, but who previously had OIC, is required. There is, however, limited published evidence of the impact of OIC on health state utility. A paper by Penning van Beest et al. estimated the EQ-5D in patients with a severe non-curable disease treated with opioids, and found that utility was much higher in those who were constipated than not constipated (0.65 vs. 0.31, respectively) [26]. However, this difference seems unreasonably large, and the study suffers from a number of limitations. It focused on constipation, in general, in opioid users, rather than OIC specifically, and the definition of constipation adopted by the study relied on self-reports of constipation and laxative use. Most importantly, the study was cross-sectional and not designed to assess the causal impact of OIC on health state utility and did not control for other factors that may influence this.

The importance of research to generate quantitative estimates of the impact of OIC on health state utility is underscored by the limited efficacy of standard of care for OIC. Laxatives are typically used as an initial treatment option for OIC in clinical practice; however, studies report between 81% and 94% of patients who take laxatives to treat their OIC exhibited an inadequate response [2, 14, 35]. It is evident therefore that OIC remains a significant issue for pain management with the large proportion of patients who are laxative inadequate responders (LIR), showing a clear unmet need [2].

Patients are likely to have an inadequate response to laxatives because laxatives do not target the mu-opioid receptors to which opioids bind in the gastrointestinal (GI) tract. Opioids disrupt normal motility, sphincter function, and intestinal fluid secretion into the lumen, potentially leading to OIC [36]. This has led to the use of pharmacological agents that work to specifically block the effect of opioids on the GI tract and prevent OIC from occurring, including nonselective opioid antagonists and peripherally acting mu-opioid receptor antagonists (PAMORAs) [19].

Naloxegol is a once daily treatment and is the first oral PAMORA that has been approved in the European Union for OIC. Naloxegol was studied in two, three arm, randomized, controlled, double-blind pivotal trials, KODIAC 4 and KODIAC 5, comparing naloxegol 12.5 mg, naloxegol 25 mg, and placebo in chronic non-cancer pain patients. The primary outcome in both studies was the response to study drug, defined as 3 or more spontaneous bowel movements (SBMs) per week and a change from baseline of 1 or more SBM per week for at least 9 out of the 12 study weeks and 3 out of the preceding 4 study weeks. Naloxegol 25 mg, the dose for which naloxegol received approval, was associated with significantly higher response rates in LIR patients compared with placebo over 12 weeks trials (KODIAC 4, 48.7% vs. 28.8%; KODIAC 5, 46.8% vs. 31.4%) [37].

The objective of the present analysis was to generate rigorous estimates of the impact of once daily 25 mg dose of naloxegol on the health state utility of LIR patients, the population for whom naloxegol has received market authorization in the EU; to examine if this utility impact of naloxegol is driven primarily by the change in OIC status, or whether the treatment has an impact beyond its impact on OIC status; and lastly to examine the utility impact of relief of OIC.

Methods

Data Sources

The KODIAC trial program was designed to assess the efficacy and safety of naloxegol 12.5 and 25 mg in adult patients with OIC with the EQ-5D being collected in three studies in the program. The primary objective of both KODIAC 4 and 5 (ClinicalTrials.gov identifier, NCT01309841 and NCT01323790, respectively) was to evaluate the response of patients over a 12-week treatment period (see “Introduction”), and the EQ-5D as an exploratory analysis. For these analyses, the data from the naloxegol 25 mg and placebo arms in the two trials were pooled, as the studies were of replicate design; both enrolled outpatients with non-cancer pain, aged 18–84, and were designed to ensure that ≥50% of the patients randomized were LIR at baseline. Across the two trials all arms had similar demographic and clinical characteristics enabling pooling, and this consistency was present in both the overall and LIR populations [38]. Moreover, the EQ-5D scores across trials were similar. In KODIAC 4, the placebo arm had a mean (with standard deviation) of 0.55 (0.31) and for naloxegol 25 mg of 0.60 (0.26). In KODIAC 5, the placebo arm had a baseline EQ-5D of 0.57 (0.29) and in the naloxegol 25 mg arm of 0.54 (0.31). Across trials the baseline EQ-5D was 0.56 (0.29) for the placebo arm and 0.57 (0.29) for the naloxegol 25 mg arm. KODIAC 4 and 5 were supplemented with data from KODIAC 7 (ClinicalTrials.gov identifier, NCT01395524), a 12-week extension study of patients completing the KODIAC 4 study, designed to assess the maintained efficacy and safety of naloxegol 12.5 mg and 25 mg in comparison with placebo in patients with OIC and non-cancer related pain.

Measures

The analysis was based upon the population of patients who were LIR, a subset of intent-to-treat population [37], in correspondence with the EU label. These patients were those who had reported taking at least one laxative class for a minimum of four days within the two-week period prior to the screening visit concurrent with at least moderately severe symptoms of OIC (i.e., incomplete bowel movements, hard stools, straining, or false alarms).

In this analysis, OIC was defined as having fewer than 3 SBMs per week in any of the 2 weeks in the preceding 4-week rolling period. Non-OIC was defined as having 3 or more SBMs per week in at least 3 weeks out of the preceding 4-week rolling period. Patients with insufficient data where an OIC or non-OIC response could not be determined were excluded from the analyses. This definition varied from the primary end point used in the trials (see introduction), which required responders to not only achieve a certain number of SBMs per week, but also a minimum change in the number of SBMs per week. This simplified the analysis, allowing estimates of utility based on constipation status, rather than change in that status. Data to measure OIC status were available at each of Weeks 4–12 in the KODIAC 4 and 5 studies. Data on OIC status was not collected in KODIAC 7.

Health state utility was measured using the EQ-5D. First, the EQ-5D, a generic QoL instrument, was collected in all three trials. The EQ-5D is a self-administered, generic assessment tool developed by the EuroQol Group, a network of international research teams, and consists of questions on five dimensions of health (i.e., mobility, self-care, pain/discomfort, usual activities, and anxiety/depression). The questionnaire was completed by patients at baseline, Week 4, and Week 12 in the KODIAC 4 and 5 studies and Weeks 16 and 24 in the KODIAC 7 study. Patient responses to the EQ-5D questionnaire were converted into utility values on a 0–1 scale, where 0 represents death and 1 represents perfect health, using a tariff estimated from surveys of the UK general population [39] and in accordance with the recommendation of the National Institute for Health and Care Excellence.

Analyses

The objective of the analysis was to estimate the impact of naloxegol 25 mg (from here on referred to as ‘naloxegol’) on the health state utility of LIR patients. This encompasses estimating how EQ-5D utility index scores (from here on referred to as ‘utility score’) and EQ-5D domain responses varied with treatment (naloxegol vs placebo), OIC status, and over time. To compare variables, t tests were run on continuous variables and Fisher’s exact tests were run on categorical variables, with no multiple testing corrections being undertaken. For such analyses, we assessed the absolute score differences between the arms at week 4 and 12 so to see the added benefits over time of both naloxegol and in having relief from OIC. A mixed model repeated measures model with random intercepts was used to predict the change from baseline in utility score over time. Specifically, changes in utility score were estimates at Weeks 4 and 12. Predictors tested in the model were: time (weeks), age, gender, race, body mass index, duration of opioid use, treatment, baseline utility, and OIC status. To account for the possibility that the impact of treatment on utility may vary over time, a treatment-time interaction variable was also included in the model. The final model included variables relevant to the objectives of the paper; OIC status, treatment, and time-treatment interaction, plus other variables whose association with the change in utility score was statistically significant (p < 0.1) in univariate regression models. The model was performed using SAS version 9.4, SAS Institute Inc, Cary, NC.

The studies were conducted in accordance with the Declaration of Helsinki of 1964, as revised in 2013, and the International Conference on Harmonisation. An ethics committee or institutional review board at each study site approved the final study protocol and informed consent form. All patients provided written informed consent at screening, before any study procedures were performed.

Results

The baseline characteristics of LIR patients in KODIAC 4 and KODIAC 5 are reported in Table 1.

Table 1 Baseline patient and study characteristics (LIR population)
Full size table

Utility among LIR patients in the placebo arm was demonstrably lower than among patients in naloxegol arm at Week 12 and when pooling across Weeks 4 and 12 (Table 2). Relative to placebo, the improvement in health state utility associated with treatment with naloxegol was 0.08 (p < 0.05) at Week 12.

Table 2 Impact of treatment on utility score (KODIAC 4 and 5 trials)
Full size table

Table 3 shows how patient responses to the five separate dimensions of the EQ-5D varied with treatment in the LIR population. Higher proportions of patients with “Moderate” or “Severe” problems were consistently observed with placebo across the five dimensions with most notable improvements relative to placebo in the mobility and anxiety/depression domains.

Table 3 Impact of treatment on EQ-5D domains at week 12
Full size table

Figure 1 shows the impact of treatment on the domains of the EQ-5D over the 12 weeks of the KODIAC 4 and 5 trials combined. A general trend was demonstrated in both treatment and placebo arms, with increases in the proportion of patients reporting “no problems”, and a drop in the proportion of patients reporting “some problems”. The increase in the proportion of patients who had “no problems” was consistently greater for naloxegol patients than placebo patients.

Fig. 1
figure 1

Change in percent of patients in EQ-5D domains, baseline to week 12 (KODIAC 4 and 5)

Full size image

Table 4 evaluates if a patient’s OIC status results in a difference in health state utility. It shows that there is a substantial utility gain associated with relieving OIC of 0.07 (p = 0.0025).

Table 4 EQ-5D utility score by OIC status (KODIAC 4 and 5 trials, pooled naloxegol and placebo arms pooled across weeks 4 and 12)
Full size table

The effect of treatment and OIC status on utility score, which was identified in Tables 2 and 4, respectively, are likely to be related. Table 5, below, presents the mean utility score of patients when segregated by treatment, time and OIC status.

Table 5 EQ-5D utility scores by treatment, OIC status, and time (KODIAC 4 and 5)
Full size table

Table 5 supports the observation that by Week 12, health state utility improves as a result of treatment with naloxegol. There is a health state utility difference of 0.07 (0.67 vs. 0.60) between naloxegol and placebo patients who are non-OIC at week 12, and a health state utility difference of 0.06 (0.60 vs. 0.54) between naloxegol and placebo patients who are OIC at Week 12. However, this difference in utility between treatment groups is no longer statistically significant when the change in OIC status is considered. To further examine this time and treatment effect, a regression analyses was undertaken.

Table 6 shows how responses to the five separate dimensions of the EQ-5D varied with treatment and OIC status in the LIR population. In Table 3, which did not distinguish between OIC status following treatment, there was a statistically significant impact of treatment on anxiety/depression, and a borderline statistically significant impact on mobility. These treatment impacts are no longer observed in patients who are still constipated after treatment. However, the treatment impact is still observed in patients who are non-OIC after treatment, suggesting that naloxegol may reduce anxiety/depression and improve mobility, over and above any impact on these dimensions reflected in the measure of OIC status.

Table 6 Responses to EQ-5D dimensions by response status and treatment at week 12
Full size table

The analysis reported in Table 7 broadly supports the observations above that the key driver of health state utility is OIC status along with baseline utility, but a time and treatment effect does exist. When controlling for other relevant factors, it is estimated that relieving OIC is associated with a 0.05 health state utility improvement. The analysis also identifies a significant treatment-time interaction effect (p = 0.024), suggesting that naloxegol at Week 12 generates a utility improvement over and above its impact on OIC status.

Table 7 MMRM for change from baseline in utility score (KODIAC 4 and 5)a
Full size table

Impact of treatment on health state utility was also captured in KODIAC 7. Table 8 suggests that the impact of treatment observed at 12 weeks is maintained to Week 16, though this impact is minimally observed at Week 24. It is important to note that these observations should be treated cautiously; they are based on the subset of participants who completed the KODIAC 4 trial and were rolled into the KODIAC 7 extension study. Consequently, the sample is small and subject to selection bias.

Table 8 EQ-5D utility scores of patients in each treatment group
Full size table

Discussion

These analyses demonstrate the utility benefits of treating OIC in LIR patients with naloxegol 25 mg. On a scale of 0–1, where 0 represents death and 1 perfect health, patients treated with this, saw a 0.08 improvement in health state utility at 12 weeks after treatment, relative to placebo. Recent studies concluded that opioid antagonists improve OIC-related symptoms. For instance, there is evidence that naloxone in combination with oxycodone generates improvements in the Bowel Function Index [41]. The current study adds to this literature by observing health state utility improvements in the EQ-5D. As well as being a validated instrument, and providing data compatible with the needs of those undertaking cost-utility analysis, the EQ-5D is a generic instrument which has the advantage of generating utility estimates comparable with those generated for many other indications.

Although we have demonstrated an impact on utility levels, an appropriate, indication specific, estimate of a minimum important difference (MID) is not in the published literature. Analyses in patients with OIC need to be undertaken to define this. Although Coretti et al. [42] concludes that there is a great amount of heterogeneity in estimates of MIDs due to variations in study populations and methodologies and in the absence of such an estimate, it is useful to observe that a 0.08 improvement represents a 14% improvement on the baseline utility of patients in trial, and is equivalent to the utility loss from the long-term consequences of a moderate stroke, profound hearing loss with ringing, or a dislocated shoulder [43].

When a baseline utility score and OIC status were controlled for, the effect of treatment alone on health state utility disappears though treatment-time effect remains, suggesting that naloxegol has a lasting impact on utility over and above its impact on OIC status. This may point to the limitation of the binary measure of OIC improvement adopted in the study. The definition of OIC reflected recognized best practice, drawing on the Rome III diagnostic criteria for functional constipation [44]. However, this measure would potentially miss some of the impact of treatment on health state utility if there is an improvement in OIC symptoms and the number of SBMs beyond that captured by the measure of OIC status. This caveat is supported by the trial data. In both KODIAC 4 and KODIAC 5, patients who were not constipated experienced more SBMs when on naloxegol than on placebo. Further work should be undertaken to explore the impact of changes in numbers of SBMs on utility score.

The analyses also suggest that change in OIC status is the most impactful driver of the improvement of OIC treatment on utility. When other factors are controlled, relieving OIC is associated with a 0.05 improvement in health state utility. This estimate of the impact of OIC on utility is lower than those previously reported in the literature [26], reflecting the limitations in the methods adopted in previous studies and the lack of face validity of their findings. Nevertheless, even at this lower level, the impact of OIC on utility points to an important burden of illness and the need for effective treatments for patients with OIC.

The analysis of changes in the dimensions of the EQ-5D instrument suggest that the utility gain associated with naloxegol is primarily a result of improved mobility and reduced anxiety/depression, due to relief of OIC symptoms. No statistically significant change in the pain dimension of the EQ-5D instrument was observed, which was not unexpected given naloxegol being a treatment for OIC. However, a disadvantage of the EQ-5D is that its health state descriptive system is not always designed to be sensitive to certain medical conditions [45]. The EQ-5D describes only 5 dimensions, and has been shown to be less sensitive to conditions, such as vision problems [46–50], diabetes, and arthritis. This could potentially lead to underestimation of the impact of indications on utility. For instance, the EQ-5D may miss the impact of OIC on social functioning. This concern has often led to the use of disease-specific instruments and mapping exercises to create utility scores based upon these disease specific instruments. One disease specific instrument developed for constipation is the Patient Assessment of Constipation Quality of Life questionnaire. This provides a brief but comprehensive assessment of the burden of constipation on patients’ everyday functioning and well-being [51]. It is the most frequently used instrument to assess health related quality of life in patients with chronic constipation and is validated in this disease, but has not been specifically validated in patients with OIC [22].

Finally, it should be reiterated that the EQ-5D was not the primary endpoint in the KODIAC trials. Consequently, the trials were not prospectively powered for these endpoints, and we need to be cautious when using secondary and exploratory endpoints to assess the impact of treatment. However, this caution should be offset against the positive result obtained on the primary endpoint, response, and the fact that response is a key driver of the improvement in health state utility observed with naloxegol.

Conclusion

Treatment of chronic pain with opioids is associated with many challenges, with OIC being an important factor due to its impact upon daily activities and health state utility. These analyses suggest that treatment with naloxegol improves patients’ health state utility, and this is primarily driven by a change in the status of OIC.