A Randomized, Double-Blind, Multi-Site, Crossover, Placebo-Controlled Equivalence Study of Morning Versus Evening Once-Daily Sustained-Release Morphine Sulfate in People with Pain from Advanced Cancer

Article Outline

• Abstract
• Introduction
• Methods
  • Study Setting
  • Participation
  • Intervention
  • Data Collection
    • Primary and Secondary Outcome Measures
  • Data Analysis
    • Study Power
  • Null Hypothesis
  • Ethics Committee Approval
• Results
  • Baseline Characteristics
  • Study Progress
  • Primary Outcome Measure
  • Other Measures of Pain and Pain Control
  • Secondary Measures
• Discussion
  • Limitations of the Study
  • Generalizability
  • Future Directions
• Acknowledgments
• References
• Copyright

Abstract 

Diurnal variation in pain perception is recognized. The question of whether opioid prescribing should be adjusted to account for diurnal variation can be tested with the advent of once-daily sustained-release morphine. The study recruited 45 people with opioid-responsive pain on stable doses of analgesics and advanced cancer from five regional palliative care programs in Australia. Each participant took one placebo and a 24-hourly dose of sustained-release morphine daily, 12 hours apart—active dose in the morning for one week and in the evening for the other week. The order of the weeks was randomized in a double-blind manner. The primary outcome from the last two days (steady state) on both arms was averaged four-hourly pain scores while awake on a 100mm visual analogue scale (VAS). Secondary outcomes included VAS and categorical scales for other pain parameters, quality of sleep, nausea, vomiting, constipation, confusion, and somnolence. Twenty-six of 42 participants completed the study and provided adequate power for analysis. Mean VAS was 16mm for morning dosing and 14mm for evening dosing (P=0.76, difference of adjusted means 2mm, 95% confidence interval: −2, 6). No differences were found in pain control, pain during the day, pain disturbing sleep, or with breakthrough medication use. This study suggests that any difference between morning and evening dosing of once-daily sustained-release morphine in people with significant opioid-responsive pain and advanced cancer is small and unlikely to be clinically significant for most people.

Key Words: Palliative care, opioid prescribing, controlled clinical trial, pain relief, chronovariability, cancer pain

Back to Article Outline

Introduction 

There have been data for more than 30 years to support circadian and circamensual variation in the perception of pain both in healthy volunteers and in people with intractable pathological pain,¹, ² although patterns may vary with the underlying pathology causing the pain.³, ⁴ There are complex patterns observed depending on the stimulus and target participants.³ The relationship of pain over a 24-hour period to other factors, including work, becomes a complex interplay.¹

The potential clinical application of this important observation has been explored in open-label prospective studies in people with severe pain due to cancer. Data to date suggest that there is diurnal variation in patient-controlled analgesia (PCA), with much lower levels of use at night.⁵ How does this observation of an almost 50% difference in daytime to nighttime PCA use translate to the population who are adequately maintained on oral opioid analgesia and are blinded to their dosing schema in steady state? Given evidence of the persistence of chronovariability for pain even with adequate around-the-clock analgesia, should dosing patterns be modified to account for these changes?⁴

Observations of pain chronovariability are complemented by data to support chronopharmacokinetic variability in the metabolism of morphine that show that identical doses of morphine produce higher plasma levels when administered in the morning than the evening.⁶ Accepting that there are diurnal variations in pain sensitivity, chronopharmacokinetic effects, and chronopharmacodynamic effects, the current study sought to determine the clinical relevance of these observations in an adequately powered study.

The need for more rigorous, controlled studies in opioid analgesia to improve clinical care is recognized.⁷ The advent of controlled and subsequently once-daily sustained-released morphine sulfate allows this question to be explored in a blinded way for the first time.⁸

Maximum serum opioid levels occur between eight and 12 hours after administration of sustained-release morphine sulfate in steady state.⁸ This means that a morning dose of sustained-release morphine sulfate will have peak serum opioid levels that broadly coincide with peak circadian levels of pain.¹, ², ³ The null hypothesis is that there is no difference in pain relief giving either once-daily morning or evening sustained-release morphine sulfate. This hypothesis was formed despite observations that there is a diurnal pattern to pain perception. This study is an equivalence study to exclude clinically relevant differences in pain scores in people with adequate pain control on an already established regular opioid regimen for opioid-responsive nociceptive pain. Any difference observed needs to be of an order of magnitude that is of relevance to most patients being treated with opioid analgesia in the setting of advanced cancer.

Back to Article Outline

Methods 

Study Setting 

The study was conducted across five centers in metropolitan teaching hospitals and their affiliated community teams in Australia over a 12-month period. Between these services, there are more than 3,500 referrals per year, of whom more than 90% of people at the time of the study had cancer as their life-limiting illness.

Participation 

This was an effectiveness study, so the inclusion criteria were broad and the exclusion criteria were as narrow as possible (Table 1). People were sought in both the community and the inpatient setting.

Table 1. Inclusion and Exclusion Criteria

Inclusion Criteria	Exclusion Criteria
Pain from cancer requiring regular opioid dosing	Confusion severe enough to impair completion of the study measures
Aged >18	Uncontrolled nausea or vomiting
Able to fill out English language diaries	Known or suspected gastrointestinal tract obstruction
Serum creatinine<1.5X upper limit of normal	Known history of hypersensitivity to morphine
Regular dose of morphine for opioid-responsive pain between 40a and 400mg per 24 hours	History of opioid-induced respiratory depression
Stable dose of opioid (as defined by 2 or fewer doses of breakthrough medications in the two days prior to commencing the study)	History of substance misuse
Stable cancer (no chemotherapy, change in hormone therapy, or radiotherapy in the last month)
Stable coanalgesics (no change to coanalgesics in the 48 hours before commencing the study)

Intervention 

The participant took two doses each day—an active drug in the form of a once-daily (24-hourly) dose of sustained-release morphine and an identical placebo. Each participant took the active dose in the morning for one week and in the evening the other week. At no time was the participant without their full regular dose of morphine (Table 2).

Table 2. Study Plan

Randomization was based on whether the active drug in the first week was given in the first week as a morning or evening dose. This randomization was allocated from a central computer-generated random number sequence that created a list for the clinical trials pharmacists at each site for sequential assignment of the next enrolled patient. People then swapped to the opposite arm for the following week. The process was blinded at all times to participants and treating clinicians, and double blinding remained in place until the completion of the entire trial.

Data Collection 

Data were collected on key clinical and demographic factors at the beginning of the trial, including eligibility criteria. The participants were asked to fill out diary cards twice each day for the two days before the trial started and for the 15 days of the study (seven days on each arm and the crossover day in the middle).

The primary outcome was pain on a 100mm visual analogue scale (VAS) anchored on the left (no pain) and right (worst possible pain), and filled out every four hours while awake and averaged over the day. Gourlay et al.⁸ showed that patients receiving 24-hourly sustained-release morphine were at steady state on Day 5 (and should theoretically be reached between 40 and 60 hours); consequently our primary outcomes were based on data from the last two days of both arms (Days 6 and 7). Categorical scales for pain relief, pain since waking, and pain during the day were sought daily. Blinded preference was asked of participants at the end of the study for the week that gave better analgesia.

Secondary outcomes were about toxicity, with VAS and categorical scales for quality of sleep, nausea, vomiting, constipation, confusion, and somnolence.

Data Analysis 

Continuous measures, including all VAS reports, were analyzed using ANOVA for crossover design.⁹ The baseline score was included as a covariate. Analysis included treatment effects (morning versus evening once-daily sustained-release morphine), sequence effects (morning-evening versus evening-morning), and period effects (first versus second week).

Ordinal measures were categorized as being the same, worse, or better during Period 2 compared with Period 1. Given the small sample size and the potential for a period effect, the exact version of the Mainland-Gart test was used.⁹ All P-values quoted are two-tailed tests, with significance attributed at P<0.05.

A priori, it was calculated that 26 subjects would provide at least 80% power, at a two-tailed Type 1 error of 0.05, to detect a difference in mean pain score between morning and evening treatments of 0.6 times the within-subject standard deviation (SD). This was a threshold felt to reflect clinical relevance in people already stabilized on opioids with good analgesia.

Null Hypothesis 

The null hypothesis was that there was no difference between morning and evening doses of sustained-release morphine administered daily.

Ethics Committee Approval 

This study was approved by the Human Research Ethics Committees of all five institutions where recruitment occurred. All participants provided written informed consent before commencing the study.

Back to Article Outline

Results 

Baseline Characteristics 

A total of 42 people were randomized in the study (28 males), with a mean age of 63.8 (SD 10.1, median 65.5, range 36–82). Performance status at the time of enrollment was measured using the Eastern Cooperative Oncology Group (ECOG) scale: ECOG 0: 3 (7%), ECOG 1: 17 (41%), ECOG 2: 13 (31%), ECOG 3: 9 (21%). At baseline, mean VAS averaged over the day was 12mm±11 for the AM/PM sequence and 14mm±15 for the PM/AM sequence (P=0.76). At baseline, mean categorical pain rating averaged over the day was 1.82±0.53 for the AM/PM sequence and 1.58±0.66 for the PM/AM sequence (P=0.85).

Opioid medications used before entry into the study were Kapanol® (29 patients [69%]), of whom seven were already taking it once daily), or MS Contin® (13 patients [31%]). The mean total daily dose of morphine at entry into the study was 145.7mg (SD 82.7mg, median 140, range 40–320mg).

There were no significant baseline differences between the two groups in gender, age, height, weight, performance status, opioid used, frequency of dosing or opioid dose at the time of enrollment, baseline scores for pain as measured on VAS or categorical scales, pain relief, any of the secondary measures, or demographics.

Study Progress 

Of the 42 people randomized, 16 withdrew during the course of the study (Table 3). Nine participants withdrew from the AM/PM arm and seven from the PM/AM arm. For three participants in the AM/PM arm and one in the PM/AM arm, lack of efficacy was cited as the reason for withdrawal. None of the withdrawals were otherwise linked to the intervention. There was no difference in withdrawals between AM/PM and PM/AM (P=0.97).

Table 3. Reasons for Withdrawal from Study

Primary Outcome Measure 

The primary outcome measure was the difference between treatment arms on a 100mm VAS, which participants recorded every four hours while awake. For purposes of analysis, scores were averaged over 48 hours. Adjusted means for all people when taking morning once-daily sustained-release morphine over 24 hours were 16mm and while taking once-daily sustained-release morphine in the evening, 14mm (difference of means adjusted for baseline 2, 95% confidence intervals [CI]: −2 to 6mm). There were no significant differences between treatments (P=0.34) and period (first week versus second week; P=0.51) or sequence (AM/PM versus PM/AM; P=0.43) effects.

Other Measures of Pain and Pain Control 

Only 10 participants expressed a blinded preference for the effectiveness of pain control in one arm over the other, eight for PM once-daily sustained-release morphine. There was no significant treatment effect (P=0.21) or period effect (P=1.0). Patient satisfaction with pain medication disclosed no significant treatment effect (P=0.57) or period effect (P=1.0). There were no treatment (P=0.32) or period effects (P=0.14) with breakthrough medication dosing. There were no differences between the arms on participant rating of how well pain was controlled, with no significant treatment (P=0.47), period (P=0.15), or sequence (P=0.20) effects. The difference between means adjusted for baseline for pain control was 0.07 (95% CI: −0.12 to 0.25) on a five-point categorical scale.

Adjusted means for categorical pain ratings (0=no pain, 1=mild, 2=moderate, 3=severe, or 4=intolerable pain) reflect the same findings. There was no significant difference between the group taking morning sustained-release morphine (adjusted mean 1.8) and the group taking evening sustained-release morphine 1.7 (P=0.85, difference in adjusted means 0.1, 95% CI: −0.1 to 0.4).

Pain control (1=complete, 2=partial (acceptable), 3=partial (unacceptable), or 4=not controlled) was compared during the last two days after adjusting for baseline scores. There was no significant difference in pain control between the morning sustained-release morphine (adjusted mean 1.7) and evening sustained-release morphine (adjusted mean 1.6, P=0.46, difference in adjusted means 0.1, 95% CI: −0.1 to 0.3).

No significant differences were seen between morning ratings of pain since waking (on a four-point scale; P=0.22, mean difference adjusted for baseline 0.2, 95% CI: −0.1 to 0.4) and evening ratings for pain during the day (P=0.47, mean difference adjusted for baseline 0.1, 95% CI: −0.1 to 0.3).

In total, people were very satisfied or satisfied with morning sustained-release morphine in 24/26 cases and with evening sustained-release morphine in 25/26.

Secondary Measures 

For nausea, there was a significant treatment difference (P=0.03; adjusted means sustained-release morphine morning 1.17, evening 1.34), but there were no period (P=0.51) or sequence (P=1.0) effects. The mean difference after adjusting scores for baseline was −0.17 (95% CI: −0.30 to −0.03). There was, however, no difference in the use of antiemetic medications between morning and evening sustained-release morphine. There were no other significant treatment, period, or sequence effects noted for other side effects (Table 4).

Table 4. Secondary Outcome Measures

	Adjusted Mean Scores for Participants When Taking Once-Daily Sustained-Release Morphine		P-valuea	Confidence Interval
Anchored at 0=none	Morning	Evening		Lower	Upper
Quality of sleep	1.65	1.51	0.19	−0.07	0.36
Nausea or vomiting during the day	1.17	1.34	0.03b	−0.30	−0.03
Constipation	1.33	1.34	0.91	−0.17	0.15
Somnolencec	1.78	1.76	0.89	−0.19	0.22
Confusion	1.25	1.26	0.63	−0.08	0.05
Appetite	1.87	1.86	0.93	−0.17	0.19

Back to Article Outline

Discussion 

This adequately powered equivalence study has not disproved the null hypothesis and suggests that any difference between morning and evening dosing of daily sustained-release morphine in people with opioid-responsive pain and advanced cancer is at best small and unlikely to be of clinical significance for most people. It is likely that diurnal variations in pain are more important as a laboratory construct than in informing pharmaceutical interventions for pain in this patient population. These findings support earlier noncontrolled studies in a similar population with cancer-related pain whose use of “as needed” analgesia as their only source of opioids for 48 hours when transferring to “strong” opioids showed no clinically important diurnal variation.¹⁰

The findings do not support an earlier unblinded study in a similar patient population where there was an almost 50% decrease in PCA self-administration in opioid use at night.⁵ The current study of sustained-release morphine was powered to find a difference of a similar magnitude. The difference may have been that participants in the study using PCA slept at night, unable to activate their PCA until pain had woken them, whereas the study on sustained-release morphine provided participants with an established dose of opioid to maintain steady state at all times, including throughout the night.

The issue of nausea being a significant finding is probably of statistical rather than clinical significance. Given that several factors were being analyzed, the finding may be a Type I error. The other analyses and the evidence from the study of identical use of antiemetics support that this may be a spurious finding.

This study demonstrates that rigorous multi-site studies are feasible in the palliative care setting. Such studies are necessary if we are to continue to refine prescribing and translate laboratory observations into evidence-based clinical practice. The study also highlights the continuing challenges of both recruitment and retention into palliative care studies. Although it is estimated that 80% of people with advanced cancer will have moderate or severe pain, from the 2,800 potential participants referred to the participating services during the period when the study was open for recruitment, only a small number were eventually enrolled in the study. The attrition rate in this 17-day study was predictably high but independent of the study intervention, and did not itself mean that an adequately powered study could not be completed. Creating infrastructure for ongoing, multi-site, blinded intervention studies will continue to be crucial in palliative care clinical research given these fundamental difficulties with recruitment and retention rates.

Equivalence studies are pivotal to refining palliative practice where both benefits and burdens of different interventions for the same problem need to be prospectively evaluated and where currently there is clinical equipoise. In frequently encountered symptoms such as fatigue, pain, dyspnea, and nausea, this design can help refine clinical practice.

Limitations of the Study 

Should the study have been powered for a smaller difference between the two arms? Given that this was a population where analgesia had already been optimized on opioids with other agents and the baseline levels of pain were already very low, there would have been no advantage in defining a difference that was of smaller magnitude than the one chosen. Accepting the outer limits of the CIs for the primary outcome, a difference of less than 4mm on a 100mm VAS cannot be considered to be clinically significant.¹¹ Although using a smaller difference with a larger sample size may have achieved a statistically significant difference between the groups, such an outcome is unlikely to improve clinical prescribing.

Generalizability 

The study was conducted rigorously with outside data monitoring and analysis to ensure quality and consistency across sites. The study aimed to be an effectiveness study across sites with differing referral patterns, with broad inclusion and exclusion criteria so that the findings could be translated to the widest possible palliative population.

This study is of a group of people with characteristics typical of pain at the end of life. Whether there is a subgroup of people with poorly controlled pain, people on very high doses of opioids, or people who are still trying to function at work where diurnal variations become more clinically important will need to be answered by subsequent studies in these subgroups.³ There may be people, who clinicians need to identify, where there are larger individual diurnal variations in pain perception, who may also benefit from specifically tailored dosing schedules.³

Future Directions 

In this particular population, the next step would be to correlate these findings with people's level of function to determine whether the differences in the laboratory are magnified in people who are most mobile and with the highest functional states despite their pain.³, ¹⁰

Back to Article Outline

Acknowledgments 

The authors would like to thank the staff who so generously gave their time and commitment to improving the care all clinicians can offer, the staff of the units involved, and colleagues who have offered support. Thanks specifically go to Ms. Katrina Campion and Mr. Bruce Kellett for their respective support and contributions to the study.

Back to Article Outline

References 

1. Glynn CJ, Lloyd JW. The diurnal variation in perception of pain. Proc Royal Soc Med. 1976;69:369–372
   • View In Article
2. Strian F, Lautenbacher S, Galfe G, Holzl R. Diurnal variations in pain perception and thermal sensitivity. Pain. 1989;36:125–131
   • View In Article
   • Abstract
   • Full-Text PDF (733 KB)
   • CrossRef
3. Labrecque G, Vanier M-C. Biological rhythms in pain and in the effects of opioid analgesics. Pharmac Ther. 1995;68:129–147
   • View In Article
4. Odrcich M, Bailey JM, Cahill CM, Gilron I. Chronobiological characteristics of painful diabetic neuropathy and post-herpetic neuralgia: diurnal pain variation and effects of analgesic therapy. Pain. 2006;120:207–212
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (108 KB)
   • CrossRef
5. Citron ML, Kalra J, Seltzer VL, et al. Patient-controlled analgesia for cancer pain: a long-term study of inpatient and outpatient use. Canc Invest. 1992;10:335–341
   • View In Article
6. Gourlay GK, Plummer JL, Cherry DA. Chronopharmacokinetic variability in plasma morphine concentrations following oral doses of morphine solution. Pain. 1995;61:375–381
   • View In Article
   • Abstract
   • Full-Text PDF (656 KB)
   • CrossRef
7. Bell RF, Wisloff T, Eccleston C, Kelso E. Controlled clinical trials in cancer pain. How controlled should they be? A qualitative systematic review. Br J Cancer. 2006;94:1559–1567
   • View In Article
   • MEDLINE
8. Gourlay GK, Cherry DA, Onley MM, et al. Pharmacokinetics and pharmacodynamics of twenty four-hourly once daily sustained release morphine compared to twelve-hourly MS Contin in the treatment of severe cancer pain. Pain. 1997;69:295–302
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (225 KB)
   • CrossRef
9. Jones B, Kenward MG. Design and analysis of cross-over trials. London: Chapman and Hall; 1989;
   • View In Article
10. Klepstad P, Skogvoll E, Kaasa S, Borchgrevink PC. Circadian distribution of oral opioid consumption in cancer patients. Methods Find Exp Clin Pharmacol. 2000;22:753–755
    • View In Article
    • MEDLINE
    • CrossRef
11. Norman GR, Sloan JA, Wyrwich KW. Interpretation of changes in health-related quality of life: the remarkable universality of half a standard deviation. Med Care. 2003;41(5):582–592
    • View In Article
    • MEDLINE
    • CrossRef