Randomized, controlled trial of cannabis-based medicine in central pain in multiple sclerosis

Central pain, i.e., pain initiated or caused by a primary lesion or dysfunction of the CNS,¹ is estimated to occur in between 17% and 52% of people with multiple sclerosis (MS).² As many as 32% of patients with MS regard pain among their most severe symptoms,³ confirming it as a “frequent, disabling and inadequately managed symptom.”⁴ The most common form of central pain in MS is nonparoxysmal extremity pain, which shows large interindividual variation and may manifest with several, typically dysesthetic, qualities such as burning, aching, pricking, stabbing, or squeezing.² Painful extremity spasms have also been classed as central pain.⁵

In the past decade, cannabinoids and the endocannabinoid system have come under intense scrutiny following the discovery of CB1 and CB2 receptors and development of specific cannabinoid receptor agonist and antagonist ligands.⁶ The rationale for performing a randomized, controlled trial (RCT) of cannabis-based medicine (CBM) in MS-related neuropathic pain is based on encouraging results using cannabinoid receptor agonists in relieving symptoms of experimental allergic encephalomyelitis⁷ and preliminary studies demonstrating modest positive effects of a synthetic cannabinoid analogue on neuropathic pain of mixed etiologies and of whole plant–derived CBM on neurogenic symptoms, including pain, in patients with MS.^8,9 A systematic review of trials of cannabinoids in pain management concluded that cannabinoids may relieve neuropathic pain, but some authors have questioned the appropriateness of trials of cannabinoids using oral administration due to the variability in their gastrointestinal absorption and crossover designs because of their long half-lives.^10,11

This study was designed to evaluate the effect of oromucosal CBM in central pain associated with MS. CBM is derived from cannabis plant chemovars, developed to produce high and reproducible yields of specified cannabinoids and formulated to produce a CBM. Oromucosal administration is efficient and convenient in achieving accurate self-titration to overcome the wide variability of interindividual response known to occur with cannabis and cannabinoids. The study preparation contained a mixture of two principal ingredients of cannabis (delta-9-tetrahydrocannabinol [THC] and cannabidiol [CBD]) in approximately a 1:1 ratio, with small amounts (<10%) of other cannabis-based compounds, delivered via an oromucosal spray. We sought to compare the efficacy, safety, and tolerability of CBM THC:CBD with placebo in relieving central neuropathic pain in patients with MS.

Methods.

Adult patients with central neuropathic pain syndromes due to MS were invited to participate in this 5-week, four-visit, randomized, double-blind, placebo-controlled, parallel-group study conducted at the Walton Centre Clinical Trials Unit, Liverpool, U.K. Patients were identified predominantly from a previous study validating the Neuropathic Pain Scale¹² (NPS) in MS central pain¹³ and also from the regional MS clinic or by specialist referral. After written informed consent, eligible patients with MS diagnosed at least 6 months previously (Poser criteria¹⁴) were included. Central pain for which a nociceptive cause appeared unlikely was required to be of at least 3 months' duration and expected to remain otherwise stable during the study. As there is no gold standard for central pain, the diagnosis was made based on pain description and clinical examination that had to be compatible with a central mechanism and by exclusion.² Patients with dysesthetic pain perceived as an unpleasant abnormal sensation spontaneously occurring or evoked and often described using terms such as burning, aching, pricking, stabbing, and squeezing,² which is the most common form of chronic neuropathic pain associated with MS, were included as were those with painful tonic spasms. Those with chronic visceral pain, headache, spasticity-associated aching pain, secondary entrapment syndromes, or acute MS-related pains, e.g., optic neuritis or positive Lhermitte sign alone, were not included. Patients with dysesthetic pain described it predominantly in their legs and feet. Patients were excluded if their sensations were not subjectively deemed painful or if they had spasticity or painless spasms alone or another noncentral pain mechanism was considered more likely, e.g., musculoskeletal pain from postural changes or peripheral neuropathic pain from nerve entrapment. Patients taking amitriptyline or other tricyclic antidepressants were required to reduce to or maintain a maximum dose of 75 mg/day. A stable neuropathic pain medication regimen was maintained during the 2 weeks immediately before screening and throughout the study. Changes in medications or procedures expected to affect central MS pain were prohibited.

No cannabinoid use (cannabis, Marinol, or Nabilone) at least 7 days before screening or during the study was permitted. Patients consented to their details being notified to the British Home Office and agreed not to travel outside the United Kingdom or donate blood during the study.

Patients were ineligible if they had a history of major psychiatric disorder other than depression associated with their underlying condition; severe concomitant illness, seizures, history or suspicion of substance abuse; concomitant severe nonneuropathic pain or the presence of illness such as diabetes mellitus that could cause peripheral neuropathic pain; or scheduled procedures requiring general anesthesia during the study. Patients were also excluded if they were pregnant, lactating, taking levodopa therapy within 7 days of study entry or had known or suspected hypersensitivity to cannabinoids.

The Local Research Ethics Committee approved the study and stipulated that patients should not drive during the study. Participants underwent assessment including calculation of Expanded Disability Status Scale¹⁵ (EDSS) score. Patients localized their pains using a body map. A pain history and bedside examination was undertaken to establish whether the pain or pains were likely to be of central origin, based on their location, quality or qualities, and associated sensory abnormalities such as hypoesthesia, allodynia, and hyperpathia. The patients' most troublesome central neuropathic pain was thus identified, and they then estimated at what time of day this was expected to be at its maximum severity. An 11-point numerical rating scale (NRS-11) (in which 0 = no pain and 10 = worst possible pain) was completed daily for the identified pain at the identified time for the 7- to 10-day baseline screening period and throughout the study. A daily NRS-11 scale recording sleep disturbance due to neuropathic pain rating (in which 0 = did not disrupt sleep and 10 = completely disrupts unable to sleep due to pain) was also completed. The NPS was completed at the same time as the pain NRS-11 for three consecutive days in both the run-in week and the final week of treatment.

The following items were examined before first dosing and at study completion or withdrawal: cognitive function (Brief Repeatable Battery of Neuropsychological Tests¹⁶), mood (Hospital Anxiety and Depression Scale¹⁷ [HADS]), and MS-related disability (Guy's Neurological Disability Scale¹⁸). At the end of the study, a Patient's Global Impression of Change¹⁹ (PGIC), a 7-point rating scale of a patient's overall change in status since commencing study medication, was applied.

Patients were randomized using a predetermined randomization code drawn up by a statistician who remained unknown to study personnel throughout the duration of the trial. Treatment allocation was made using randomized permuted blocks of four (two active drug, two placebo), with treatments sequentially assigned to either a whole-plant CBM (CBM = GW 1000-02, Sativex) containing THC:CBD delivered via oromucosal spray or placebo. Each spray delivered 2.7 mg of THC and 2.5 mg of CBD or placebo. Placebo was designed to match the appearance, smell, and taste of the active formulation but contained no active components, in ethanol:propylene glycol (50:50) excipient. To facilitate blinding, patients completed pain and sleep assessments at home, the physician examined patients, gave dosing advice, and assessed them for adverse events (AEs); trials nurses completed all other secondary outcome assessments; and a trials pharmacist dispensed the study medication. The identity of study medication assigned to patients, to which all study personnel remained blinded, was contained in individually sealed envelopes retained in the hospital 24-hour pharmacy and with the sponsor's Pharmacovigilance Department.

Results.

Study population.

Recruitment from the regional MS clinic and referrals from consultant neurologists and pain specialists took place between March and July 2002, and patients attended the Trials Unit on four occasions over 5 weeks. Eighty-five patients were screened, of whom 66 were randomized (fig 1), 34 to CBM and 32 to placebo; 64 patients (96.9%) completed the study. Two women patients withdrew, both on CBM. One developed an AE of agitation with tachycardia and hypertension after four sprays, which settled with conservative management within 3 hours. She declined further study medication and withdrew 7 days later without completing further scores. The second patient developed paranoid ideation and was withdrawn from study medication at the investigator's discretion in the second treatment week but subsequently completed all study diaries and assessments. Two patients violated the protocol, one patient's concomitant pain medication changed in the run in period, and another patient commenced interferon treatment 3 days after commencing the study. Both patients were in the active treatment group and were included in the ITT analysis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Profile of patients at each stage of the study.

Of the 66 patients randomized, nine had primary progressive, 33 secondary progressive, 23 relapsing remitting and one benign MS. The treatment groups were well balanced in terms of gender, age, duration of MS since diagnosis, and baseline NRS-11 pain score (table 1). At baseline, the NPS items Intense, Unpleasant, and Deep were experienced by more than 80% of patients with mean severities of greater than four of ten. Conversely, less than 25% of patients experienced Cold and Itchy components with a similar magnitude and about 45% of patients did not experience these pain qualities at all.

View this table:

• View inline
• View popup
• Download powerpoint

Table 1 Baseline demographic details

Forty-three patients (65%) required support to walk or were wheelchair bound. Patients were taking a mean of about two other medications for pain, spasms, or spasticity. Forty-seven percent of patients had previous experience of using cannabis medicinally and 16.7% recreationally. Only four patients randomized to CBM and eight to placebo had taken cannabis within 3 months of study entry. The proportion of patients with any previous exposure to cannabis was not different between CBM and placebo (CBM 15/34, placebo 21/32; CBM-placebo −0.22, 95% CI: −0.45 to 0.02, p = 0.08). The mean number of daily sprays taken in week 4 was 9.6 of CBM (n = 32) (range 2 to 25, SD = 6.1), equivalent to 25.9 mg THC:24 mg CBD and 19.1 of placebo (n = 31) (range 1 to 47, SD = 12.9) (CBM-placebo –9.5, 95% CI: −14.6 to −4.4, p = 0.0004).

Efficacy measures.

NRS-11 and NPS total pain scores.

Significant mean reductions favoring CBM were found for the primary outcome NRS-11 of pain and the secondary outcome NPS (table 2 and figures 2 and 3). Of the total 65 patients included in the ITT analysis, 59 (89%) had dysesthetic pain and seven (11%) had painful spasms. Post hoc analysis demonstrated that the seven patients with painful spasms had higher baseline NRS-11 pain intensities than the patients with dysesthetic pain (dysesthetic pain 6.3 [SE = 1.6]; painful spasms 7.3 [SE = 1.5]; difference in means = 1.3, 95% CI: 0.1 to 2.6, p = 0.04) and that the changes from baseline to week 4 tended to be greater in patients with painful spasms. In the patients with dysesthetic pain, the mean changes were −2.4 (SD = 1.5, n = 30) for CBM and −1.3 (SD = 1.7, n = 28) for placebo, whereas these changes were −5.7 (SD = 3.5, n = 3) and −2.1 (SD = 1.6, n = 4) in the patients with spasm. In an analysis including only the patients with dysesthetic pain, the NRS-11 pain treatment effect was 1.1 (SE = 0.4, p = 0.012). An interaction between treatment and type of pain was not found on a 5% level (p = 0.07).

View this table:

• View inline
• View popup
• Download powerpoint

Table 2 Changes in pain and sleep by treatment (CBM = 33, placebo = 32)

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Mean 11-point numerical rating scale (NRS-11) pain scores (± SEM) for the cannabis-based medicine (CBM) (n = 33) and placebo group (n = 32). Week 0 refers to the run-in week. The patients were on test medication in weeks 1 to 4.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Mean neuropathic pain scale (NPS) scores (± standard error of mean) for the cannabis-based medicine (CBM) (n = 33) and placebo group (n = 32) at baseline (Visit 1) and end of treatment (Withdrawal/Completion).

Post hoc analysis of individual NPS items demonstrated treatment effects for all 10 items in favor of CBM that reach significance for Intense (treatment effect estimate –1.31, 95% CI: −2.21 to −0.40, p = 0.0054), Dull (treatment effect estimate –1.04, 95% CI: −2.05 to −0.03, p = 0.0433) and Sensitive (treatment effect estimate –1.01, 95% CI: −2.02 to −0.01, p = 0.0484).

Pain-related sleep disturbance.

The reductions in pain were reflected in similar reductions in mean daily pain-related sleep disturbance with a mean treatment difference favoring CBM (table 2, figure 4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4. Mean sleep disturbance 11-point numerical rating scale (NRS-11) scores (± SEM) for the cannabis-based medicine (CBM) (n = 33) and placebo group (n = 32). Week 0 refers to the run-in week. The patients were on test medication in weeks 1 to 4.

PGIC.

The proportion of patients rating themselves as “much” or “very much improved” in the CBM group (9/34) was not greater than those receiving placebo (4/32) (treatment difference 14% points, 95% CI: −4.8 to 32.7, p = 0.218). No patient felt “much worse” or “very much worse” (see table E-1 on the Neurology Web site at www.neurology.org). On the 7-point PGIC, those treated with CBM were 3.9 times more likely to rate themselves in any improved category than those receiving placebo (95% CI: 1.51 to 10.09, p = 0.005).

AEs.

Thirty patients (88.2%) on CBM developed at least one AE, compared with 22 patients (68.8%) on placebo (CBM-placebo 0.19, 95% CI: 0.00 to 0.39, p = 0.053). Common AEs are summarized in table 3; in addition, confusion, crying, low mood, disorientation, paranoia, hallucination, and logorrhea all occurred once in the CBM group. Fifty-three percent of the patients in the CBM group experienced dizziness at least once compared to 16% in the placebo group (CBM-placebo 0.37, 95% CI: 0.16 to 0.58, p = 0.002). Some of the psychiatric AEs occurred in the same patient. No serious AEs, i.e., fatal, life-threatening, or resulting in persistent or major disability/incapacity or prolonging hospitalization, occurred. However, two women patients in the CBM arm experienced AEs severe enough to warrant trial withdrawal (see “Study Population” section).

View this table:

• View inline
• View popup
• Download powerpoint

Table 3 Summary of common adverse events during parallel group treatment

No significant changes were seen in either group in blood pressure, weight, temperature, hematology, or blood chemistry. During the first dosing, there was an increase in mean heart rate in 1 hour in the cannabis-based medicine group by 3.2 beats per minute (95% CI: 2.3 to 4.1) compared with a mean of 1.6 beats per minute (95% CI: 0.8 to 2.5) in the placebo group, p = 0.016.

Neuropsychological outcomes.

No significant differences between mean changes on each treatment were found in the 10/36 Spatial Recall Test, Symbol Digit Modalities Test, Paced Auditory Serial Addition Test, or Word Generation List between treatment groups on neuropsychological testing. In the long-term component of the Selective Reminding Test (SRT), a difference was found because of a mean improvement in the placebo group (n = 32) of 5.7 (95% CI: −19 to 26) not matched in the CBM group (n = 33) of –0.9 (95% CI: −20 to 23) mean treatment difference −6.95 (95% CI: −12.12 to −1.77), p = 0.009 (see table E-2).

Other secondary outcomes.

No differences between mean changes on each treatment were found between treatment groups in the other secondary measures of HADS anxiety and depression and Guy's Neurological Disability Scale (see table E-3).

Discussion.

This randomized, placebo-controlled trial demonstrates a beneficial effect of CBM both in the relief of central pain associated with MS and pain-related sleep disturbance. Although our study's inclusion criteria allowed for any type of MS-related central pain of greater than 3 months' duration, the predominance of dysesthetic extremity pain in this study (59 patients [89%]) is in agreement with previous series.^2,5 The definition of and conditions encompassing “neuropathic” pain remain controversial.²² No universally accepted validated clinical diagnostic criteria for neuropathic pain exist,²² and assessment of patients based on clinical examination and bedside tests to decide what is and what is not neuropathic is difficult, even for experts.²³

Some authors view spasm-related pain in MS as being neuropathic,⁵ whereas others do not.²¹ A crossover trial of dronabinol in MS central pain specifically excluded patients with spasm-related pain.²¹ Painful spasms in MS feature sudden-onset, either unilateral or bilateral, dystonic posturing with a stereotyped pattern in the same patient. They are thought to be caused by “a transversely spreading ephaptic activation of axons within a partially demyelinated lesion.”^24,25 No study to date has linked painful spasms to dysfunction in either the motor or somatosensory system exclusively, and whether there is any advantage in separating the two for therapeutic purposes is uncertain. Although patients with MS and painful spasm must have a CNS lesion, the key question is whether the pain is generated primarily in the spasmodic muscles or the CNS. In this study, patients with painful spasms responded similarly to those with predominantly dysesthetic pain, suggesting that dichotomizing patients based on putative differences in central mechanisms of the two groups may be superfluous.

Patients in our study were taking, on average, two other medications, with limited efficacy given baseline NRS-11 pain scores of 6.5. Therefore, as adjunctive analgesic treatment, CBM had a significant treatment effect of −1.25, in the NRS-11, in excess of the −0.6 achieved by oral dronabinol in an MS study in which concomitant analgesia was restricted to paracetamol²¹ and comparable to treatment effects of approximately 0.9 and 1.25 to 1.45 in RCTs of peripheral neuropathic pain using tramadol and pregabalin.^26,27 The treatment difference for the NRS-11 did not reach that for which the study was powered, although this calculation was based on peripheral neuropathic pain studies.^19,20 A meta-analysis of more than 2,700 patients with various painful conditions suggested approximately a 30% or 2-point NRS-11 score reduction in pain as being clinically significant²⁸ but notably did not include patients with central neuropathic pain, in which “relatively small decreases in pain intensity are often highly valued by the patients.”²

In our study, the numbers needed to treat²⁹ to achieve a 50% reduction in central pain in at least one patient was 3.7 (95% CI: 2.2 to 13.0), similar to that obtained in the dronabinol trial of 3.5 (95% CI: 1.9 to 24.8).²¹ Numbers needed to harm²⁹ (NNH) is calculated as 1/risk difference, and for the probabilities of at least one AE, this is 1/0.19 = 5.13. Specifically, for CBM to cause dizziness, the NNH was 1/0.37 = 2.68. Current options for treating central pain conditions remain limited and are based mostly on the use of CNS drugs with known problems of tolerability.^30,31 CBM was well tolerated overall, despite a population including 25 of 34 patients (73.5%) in the treatment group requiring some walking aid and eight (24%) being wheelchair bound.

A systematic review identified that, until 2001, RCTs of cannabinoids were largely confined to single-dose trials.¹⁰ In this trial, patients could titrate to a maximum of THC 130 mg:CBD 120 mg. The mean dose achieved of 25.9 mg THC, and particularly 24 mg CBD, is in excess of that used in other cannabinoid RCTs.^21,32,33 CBD is thought to modulate the effects of THC and also to have analgesic properties of its own. These factors may contribute to the positive outcomes in this trial. To place these reductions in patient's pain in context, a quality-of-life instrument would have been beneficial, however, with a relatively short 3-week fixed treatment period. This was omitted from our study; however, an odds ratio of 3.9 favoring an improved global impression of change with CBM, without a corresponding significant change in mood, suggests that patients felt a benefit from reduction in pain, sleep improvement, or both and contrasts with a previous RCT using orally administered THC and whole-plant cannabis extract, which significantly reduced PGIC.³² The Cannabinoids in Multiple Sclerosis (CAMS) trial using orally administered cannabinoid capsule formulations identified no objective change in Ashworth scores but did note subjective improvements in pain and sleep, which concurs with our results, as well as in spasms and spasticity.³³

In our study, the NPS 10-item total responsiveness also shows a significant treatment difference favoring CBM of −6.58 on a 100-point scale, demonstrating convergence with the traditionally accepted NRS-11 outcome. Post hoc analysis demonstrated significant treatment effects favoring CBM in the Intense, Dull, and Sensitive NPS items, suggesting that further studies should examine whether more sophisticated methods of analyzing the NPS are required.

Although unusual in neuropathic pain trials, some RCTs involving cannabinoids have included a question to formally assess the degree of blinding and demonstrated an element of unblinding in patients receiving cannabinoids.^32,33 Our study did not include a blinding question. However, despite a number of our patients having previous exposure to cannabis, our placebo group experienced both a large reduction in pain and number of AEs, suggesting a degree of blinding was preserved.

CBM does not appear to have significant effects on MS-related disability, mood, or on four of the five neuropsychological outcomes measured. No corrections for multiple comparisons were applied to secondary outcomes. The significant treatment effect-favoring placebo in the long-term storage component of the SRT, perhaps reflects a learning effect not matched in the CBM group. Analyses of the consistent long-term retrieval score and delayed recall at 11 minutes were not defined a priori for analysis in our study. The neuropsychological outcomes of chronic (10.2 to 24 years) recreational marijuana users and general population controls have been compared.³⁴ On the Rey Auditory Verbal Learning Test, a significantly less steep learning curve and generally recall of fewer words were observed in long-term (mean 24 years) users of cannabis than in short-term (mean 10.2 years) users or controls. Long-term users also recalled fewer words than short-term users or controls. The preliminary results of the psychological substudy of the CAMS trial found a significant reduction in the Californian Adult Verbal Learning Test in those receiving cannabis extracts compared with placebo.³⁵ These results require further analysis and incorporation of psychological outcomes in future cannabinoid trials.

Acknowledgment

The authors thank all patients who took part in the study. They acknowledge the assistance of the staff of the Clinical Trials Unit at the Walton Centre for Neurology and Neurosurgery who helped in the conduct of the study and application of many of the secondary outcome measures, especially Lynne Owen, Dot Marshall, Dave Watling, Lynne Pickering, C. Saminaden, Linda Moss, and Ann Dennis (pharmacist).