Peninsula Health Technology Commissioning Group

Peninsula Health Technology Commissioning Group Health Technology Assessment and Commissioning Decision Sativex for treatment of spasticity in multiple sclerosis Clinical case: Generally supportable Cost-effectiveness case: Not cost-effective Commissioning decision: This treatment will not be routinely commissioned See text for definitions Date issued: December 2010 Rationale for Commissioning Decision Trial evidence for Sativex shows that there is only a small difference in the proportion of patients receiving Sativex who achieve a clinically significant response compared with patients receiving placebo. However, the PHTCG recognised that Sativex is intended for a patient group who have not responded to other anti-spasticity agents and for whom there are few other treatment options. Randomised controlled trial evidence which includes a trial period reflecting the manner in which Sativex is licensed for use in clinical practice is limited to one trial. Data on quality of life collected directly from patients in this trial shows an improvement in quality of life in patients who responded to Sativex during the trial period. Sativex is intended for long term use but trial data are limited to four months of treatment so there is uncertainty over benefit in the long term. The manufacturer of Sativex collected the quality of life data using the EQ-5D tool used nationally to assess quality of life benefits in relation to cost of medicines. The manufacturer has priced Sativex at a cost which does not fall in with the range usually considered to represent value for money to the NHS for the benefit it delivers. This makes the improvement in quality of life produced by this treatment poor value for money to the NHS. NOT TO BE USED FOR COMMERCIAL OR MARKETING PURPOSES. PRODUCED TO INFORM LOCAL DECISION MAKING. 1

Executive summary Sativex is an oromucosal spray containing cannabis-based medicinal products which is indicated as add-on treatment for symptom improvement in patients with moderate to severe spasticity due to multiple sclerosis who have not responded adequately to other anti-spasticity medication. Physicians should assess patients after four weeks of use. Only patients who achieve significant clinical benefit should continue treatment. There are three key randomised controlled trials. In each trial, Sativex was compared with placebo in patients maintained on their current anti-spasticity agents. The primary end-point was mean percentage change from baseline on a patient-reported numeric rating scale (NRS) of spasticity. This outcome measure was devised for Sativex trials and has been accepted as a valid measure of spasticity. Analysis of two parallel group studies showed that there was a significant difference between Sativex and placebo in the proportion of patients who achieved a clinically significant response on the NRS (35% vs 24%). The difference of 11% is small but considered to be clinically relevant in a population of patients who are difficult to treat. Evaluation of Sativex is hindered by the lack of published information on the third study which was designed to identify responders with a four week trial period reflecting use in clinical practice. At the end of 4 weeks of treatment with Sativex, 42% of patients satisfactorily met the threshold ( 20% reduction in NRS score) for predicting patients expected to achieve a clinically significant response in the long term and were randomised to receive Sativex or placebo for 3 months. Analysis of secondary outcomes from this 3 month period showed a significant difference in favour of Sativex with improvements in spasm frequency, sleep disruption and the Barthel index. Using the EQ-5D tool, an increase in utility of 20% was reported for responders to Sativex from the start of the study. There are no robust long term efficacy data. The longest trial is four months duration. Adverse events such as dizziness, fatigue, somnolence, vertigo and disorientation were reported commonly during clinical trials. There has been no previous published cost-effectiveness analysis assessing Sativex for the treatment of spasticity due to MS. We developed a decision-analytic model to estimate the cost-effectiveness of Sativex plus oral medicines compared to standard treatment in a UK cohort of adults who have moderate to severe spasticity due to MS and have achieved insufficient benefits from oral anti-spasticity medicines. In the base case, there was little difference in quality-adjusted life years (QALYs) gained between Sativex plus oral anti-spasticity agents and oral anti-spasticity agents alone. There was a large difference in costs with Sativex costing more. This resulted in an incremental cost-effectiveness ratio (ICER) of 48,588 per year. A probabilistic sensitivity analysis which examined all of the uncertainties of the data simultaneously was undertaken. The probability of Sativex being cost-effective at a chosen threshold of 30,000 per QALY was estimated at 11.08%. The price of Sativex has a large impact on the ICER. For Sativex to be considered cost-effective at a threshold of 30,000 per QALY, the cost per prescription of three bottle should be 229 or less, a reduction of 146 (38.9%) compared with the current price of 375. 2

Abridged Health Technology Assessment 1. The Technology Sativex is an oromucosal spray containing the cannabis-based medicinal products, delta- 9-tetrahydrocannabinol (27mg) and cannabidiol (25mg). 1 Sativex was licensed in the UK in June 2010 following the second licensing application to the MHRA and was launched shortly after. Sativex has been available on a named patient basis in the UK since December 2005. Sativex is indicated as add-on treatment for symptom improvement in adults with moderate to severe spasticity due to multiple sclerosis who have not responded adequately to other anti-spasticity medication and who demonstrate clinically significant improvement in spasticity related symptoms during an initial trial of therapy. 1 The SPC for Sativex states that treatment must be initiated and supervised by a physician with specialist expertise in treating this patient population. 1 A titration period, of up to two weeks, is required to reach the optimal dose of Sativex. The median dose in clinical trials for patients with multiple sclerosis was eight sprays per day. Doses of greater than 12 sprays per day are not recommended. The patient s response should be reviewed after four weeks of treatment. 1 If a clinically significant improvement in spasticity related symptoms is not seen during this initial trial period, treatment should be stopped. In clinical trials, this was defined as at least a 20% improvement in spasticity related symptoms on a 0-10 patient-reported numerical rated scale (NRS). 2. Background and clinical need Spasticity is common in patients with multiple sclerosis and is a major contributor to disability. Spasticity occurs during the progressive stage of multiple sclerosis which generally begins in the forties. 2 The severity of spasticity increases as the disease progresses. 3,4 Patients initially develop spasms which may be painful and can occur at night leading to sleep disruption. Over time the increase in muscle tone becomes more frequent and eventually permanent leading to difficulties in nursing care, in maintaining hygiene and can lead to bed sores. The clinical course of multiple sclerosis evolves over several decades. Treatment of spasticity is long term. The prevalence of MS in Devon was estimated in a prevalence study of MS patients in Plymouth in 2001 as 118 per 100,000 based on a case definition of definite and probable cases. 5 There is an unmet need for anti-spasticity agents for patients who are not achieving adequate benefit from oral anti-spasticity agents. 3. Strategic direction NICE clinical guidance on multiple sclerosis (CG8) was issued in November 2003. 6 These recommendations were supported by a HTA of treatments for spasticity and pain in multiple sclerosis. 3 The HTA included a systematic analysis of anti-spasticity agents but a cost-effectiveness analysis was not conducted. The guidance gives recommendations on the treatment of spasticity but predates the marketing of Sativex and use of Sativex on a 3

named patient basis. A TAG for Sativex in the treatment of spasticity in MS is due to be issued in January 2012. CG8 gives the following recommendations for pharmacological treatment of spasticity: Initial treatment for bothersome regional or global spasticity or spasms should be with: baclofen or gabapentin. If treatment with baclofen or gabapentin is unsuccessful or side effects are intolerable the options include: tizanidine, diazepam, clonazepam or dantrolene. Options for patients who have troublesome spasticity and spasms unresponsive to simpler treatments include: intrathecal baclofen and phenol injections to motor points or intrathecally. Intramuscular botulinum toxin should not be used routinely, but can be considered for relatively localised hypertonia or spasticity that is not responding to other treatments. 4. Evidence of Clinical Effectiveness 4.1 Identification of key clinical trials The Cochrane Clinical Trials database and Pubmed were searched for relevant studies. A Pubmed search was conducted using the MESH terms tetrahydrocannabinol-cannabidiol combination, cannabidiol, tetrahydrocannabinol, multiple sclerosis, muscle spasticity and spasm and the free text terms Sativex and nabiximol. Other sources of data included the MHRA Public Assessment Report. Table One lists the studies discussed in this assessment. Two published RCTs were not included. The first trial was of treatment of detrusor overactivity in patients with MS. 7 The second trial investigated the effect of Sativex on a range of symptoms in patients with MS including spasticity. 8 This trial been described as exploratory. 9 Table One: Summary of Sativex studies discussed in the HTA Study type Purpose of study Source RCTs Study 106 Collin 2007 10 Study 403 Phase III - short-term efficacy and safety Collin 2010 11 MHRA assessment Study 604 report 9, Montalban 2009 12, Ambler 2009 13 RCT Phase III withdrawal study intended to Notcutt 2009 14 provide data on long term effects RCT Independent study examining Aragona 2009 15 Other studies Meta-analysis Post-hoc analysis of RCT data Prospective noncomparative study Non-comparative extension study psychopathological and cognitive effects Combines data on primary outcome from Wade 2010 16 three studies Farrar 2008 17 Validation of numeric-rating scale used to measure primary outcome Anwar 2009 18 Long term effects Wade 2006 19 Data have been extracted from published papers. The MHRA assessment report is used where there is no published paper or the report contains additional information. 4

4.2 Outcomes for trials of spasticity The primary end-point used a numerical rating scale (NRS) to measure spasticity which was devised specifically for Sativex trials. Validation of the numeric rating scale was critical to the regulatory approval of Sativex and is discussed below. A range of secondary outcome measures were used to show benefit in terms of functional improvement. Numerical rating scale (NRS) to measure spasticity Devised specifically for Sativex trials. An 11-point scale with scores ranging from 0 (no spasticity) to 10 (worst possible spasticity). Subjects were asked On a scale of 0-10, please indicate your level of spasticity over the last 24 hours. Patients were told that spasticity was defined as muscle stiffness. Scores are recorded on a daily diary card in clinical trials. The baseline NRS and changes during treatment were generally the mean of recorded assessments over a 7 or 14 day period. Ashworth or modified Ashworth scale Measures resistance encountered during passive stretching of specific muscle groups. Each joint is tested up to three times. Scale ranges from 0 (no increase in tone) to 4 (affected parts rigid in flexion or extension). Spasm frequency The method of measuring spasm frequency has differed between trials and has included a recognised spasm frequency scale, a 0-10 NRS and simply recording the number of episodes of spasm a day. Timed 10 metre walk Time taken to walk 10 metre in seconds. Ambulatory patients only. Barthel Activities of Daily Living Ten items that measure a persons daily functioning in terms of their daily living and mobility. The items include bathing, grooming, feeding, dressing, bowels, bladder, toilet use, stairs, transfers and mobility on level surface. Scores vary between items. Overall maximum score =100 or 20 depending on whether 0, 5, 10, 15 is replaced with 0, 1, 2, 3. Motricity index Measures muscle power in hand pinch and various joints. Index ranges from 0 (no movement) to 22 (normal power). Global Impression of Change physician, carer or subject Measures improvement in patient s condition. Scores range from 1 (very much improved) to 7 (very much worse). Validation of the numerical rating scale (NRS) The primary efficacy measure for measuring spasticity was changed from the Ashworth scale initially to a visual analogue scale in early clinical trials. 8 The change in end-point occurred because a study of cannabinoids for spasticity (CAMS study) failed to show improvement in the Ashworth scale and a recent Cochrane review suggested that the Ashworth scale lacked reliability and sensitivity to measure significant functional change in spasticity. 20,21 In subsequent clinical trials of Sativex, a numerical rating scale (NRS) has been used as a measure of spasticity. During the licensing process for Sativex, the company were required to provide evidence that the NRS was a valid measurement tool for assessing spasticity. 9 The scale was assessed using recognised criteria for validation of a new instrument. Features of the scale which were tested included correlation with other symptomatic measures and clinician scored measures of spasticity and test/retest reliability over study visits. 17,18 The MHRA has accepted the NRS as a valid measure of symptoms related to spasticity to support the indication for Sativex. 9 An advantage of the NRS is considered to 5

be that it measures the severity of spasticity over an extended period of time allowing for day to day fluctuations and reflects the patient s daily experience of spasticity. Determination of clinically important differences on the NRS A post-hoc analysis of data from study 106 was used to estimate the level of change from baseline on the 0-10 NRS scale that constituted a clinically important difference (CID) and a minimal CID (MCID). 17 Change in NRS score was compared with change in Patient Global Impression of Change (GIC) categories. A reduction in NRS of 29.5% from baseline correlated with scores of much improved or better and represented a clinically important difference. Scores of minimally improved or better correlated with a 18% reduction in NRS and were considered to represent the minimal change considered to be clinically significant. 4.3 Overview of phase III studies Three key phase III studies investigated Sativex for treatment of spasticity (Table Two) The first studies to be conducted were double-blind parallel group RCTs (studies 106 and 403). 10,11 These formed the basis of the first licensing application. The third study (604), which had an initial trial period, was proposed following analysis of the outcomes of studies 106 and 403 and formed the basis of the second licensing application. 9 Sativex is an add-on therapy and was compared with placebo in each of these studies. Patients were maintained on stable doses of their current anti-spasticity agents. Patients were instructed to titrate their daily dose of study medication based on efficacy and tolerability as required to a maximum number of sprays/day. The maximum daily dose of Sativex differed between the three studies. The licensed dosing schedule of a maximum of 12 sprays/day is based on study 604. After titration, patients were required to remain on a stable dose for the remainder of the study. Table Two: Study characteristics of phase III trials Study number Description 106 Double-blind parallel group RCT 403 Double-blind parallel group RCT 604 Phase A: single blind response assessment Phase B: Double-blind randomised parallel group RCT Sativex Comparator Maximum dose /day 48 sprays Placebo (add-on to current oral anti-spasticity agents) 24 sprays Placebo (add-on to current oral anti-spasticity agents) 12 sprays Phase A: no comparator Phase B: placebo (add-on to current oral anti-spasticity agents) Patient numbers 189 (124 Sativex: 65 placebo) 337 (167 Sativex: 170 placebo) Phase A: 572 Phase B: 241 Duration 6 weeks 15 weeks Phase A: 4 weeks Phase B: 12 weeks The characteristics of the study populations were similar across all studies. 9-11 Patients with any subtype of multiple sclerosis were eligible for recruitment. The mean age of patients ranged from approximately 47 to 49 years. Females accounted for 60% of patients in each study. The duration of MS ranged from 12 to 15 years across the study. The mean duration of spasticity symptoms was reported to be 7.73 years in study 403. Assessments at baseline indicated that patients had spasticity of at least moderate severity which was likely to be disabling. Patients were required to have failed to have 6

gained adequate relief on current treatment and for their anti-spasticity regimen to be stable for 30 days before entry to the study. Study 403 gave details of concomitant or previous anti-spasticity agents. The most frequently cited agents were: baclofen (80%), tizanidine (43%), benzodiazepines (28%) and gabapentin (15%). Botulinum toxin toxin was reported for 4% of patients. Exclusion criteria for studies included psychosis or severe psychiatric disorder other than depression, alcohol or substance abuse, severe cardiovascular disorder or history of seizures. In study 403, 24% of the study population had a history of cannabis use and 41.8% of the study population in study 106. No information on prior cannabis use was available for study 604. Outcome of parallel group studies Treatment with Sativex resulted in a significant mean decrease of 0.52 points on the NRS compared with placebo over 6 weeks in study 106 (Table Three). 10 The difference between treatment groups was of questionable clinical significance. 9 No information was provided on the difference in NRS score which the study was powered to detect. No significant difference was observed for the primary outcome of a reduction in 0.23 points in favour of Sativex over a 14 week period in study 403. This study was powered to detect a treatment difference of 0.75 points change from baseline compared with placebo. In both studies, small changes in secondary outcomes in favour of Sativex were reported but none achieved statistical significance. Table Three: Change in NRS from baseline to end of study for studies 106 and 403 Study Mean baseline NRS Sativex Adjusted mean change from baseline* Mean baseline NRS Placebo Adjusted mean change from baseline* Sativex vs placebo: treatment difference Study 106 (n=189) 5.49-1.18* 5.39-0.63* -0.52 (95% CI: -1.029, -0.004, P=0.048) Study 403 6.77-1.05 6.48-0.82-0.23 (p=0.219) (n=337) *Variables adjusted for included baseline NRS Further analysis of parallel group studies Analysis of study data by mean response was thought to have potentially masked the effect of Sativex in patients who did respond to treatment. 9 A combined analysis of studies 106 and 403 showed that 35% of patients receiving Sativex and 24% of patients receiving placebo achieved a clinically important reduction in NRS of at least 30%. 9 The difference of 11% (OR 1.63, 95% CI: 1.10, 2.24) was considered to be of sufficient clinical relevance to use an alternative study design to identify responders given that no patient characteristics had been identified that accurately predicted response to treatment with Sativex. A difference of 11% was also reported in the meta-analysis of Sativex studies. 16 Table Four: Percentage of patients with a 30% improvement in NRS from baseline Study Sativex Placebo Difference Odds ratio 95% CI P value 106 48/120 14/64 18% 2.38 1.19, 4.78 (40%) (22%) 403 51/166 42/169 6% 1.34 0.83, 2.17 (31%) (25%) Overall 99/286 (35%) 56/233 (24%) 11% 1.63 1.10, 2.41 P=0.015 7

A further analysis of study 403 was undertaken to establish the best predictor of patients who would achieve a reduction of at least 30% in NRS in the long term. 11 Using a threshold of 20% reduction in NRS score at week 4 (ie minimal clinically important difference) was found to provide the best balance between ensuring that the majority of patients who achieved a clinically important response to Sativex would continue treatment without subjecting a high proportion of non-responders to long term treatment. 9 Outcome of study designed to identify responders to Sativex The third study was designed to identify patients who responded to Sativex ( responders ) and subsequently randomised responders to receive Sativex or placebo for three months. This study has not been published. Details were obtained from the MHRA assessment report and conference abstracts. 9,12,13 Study 604 consisted of a 7-day screening period, a single-blind 4 week treatment period (phase A) and a 12 week double-blind, randomised, placebo-controlled treatment period (phase B). In Phase A all patients received Sativex. Patients who achieved at least a 20% reduction in NRS from baseline were considered to be a responder to Sativex and entered phase B to be randomised to Sativex or placebo. Patients were not told which drug they received during phase A and that treatment may change at week 5. In the randomised phase, subjects were to maintain a stable dose as established in the singleblind phase. A total of 660 subjects were screened, of which 572 were enrolled in the single-blind phase. Of these 538 (94%) completed the 4-weeks of single-blind treatment and 241 (42%) satisfactorily met the NRS threshold criteria for a responder and were randomised into the double-blind placebo-controlled phase. Of the subjects who were not randomised, approximately 50% achieved less than 5% reduction in their NRS score. For responders to Sativex, the greatest reduction in NRS score was seen during the first four weeks of treatment (phase A) with only a small change in NRS score observed for patients receiving Sativex during weeks 5-16 (phase B) (Table Five). The change in NRS score from phase B randomisation to end of study was significant for Sativex. Patients receiving Sativex had a small reduction in NRS score whereas an increase in NRS score for patients receiving placebo indicated that their spasticity was beginning to worsen. The study was powered to detect a 0.75 point difference in NRS score which was considered to be of clinical interest. Table Five: Change in NRS from double-blind randomisation (week 5) to end of study (week 16) for responders to Sativex Mean NRS at pretreatment baseline on entry to phase A (weeks 1-4) 6.81 3.87 (1.49) Sativex (n=124) Mean (SD) Adjusted NRS at mean NRS double-blind change at randomisation end of study* -0.04 3.92 (1.55) *Variables adjusted for included baseline NRS Placebo (n=117) Mean (SD) Adjusted NRS at mean NRS double-blind change at randomisation end of study* Sativex placebo: treatment difference vs 0.81-0.84 (p=0.0002, 95% CI: -1.29, -0.40) Of the patients who entered phase B, 74% of patients receiving Sativex achieved a 30% reduction in NRS from pre-treatment baseline (week 0) to the end of the study compared with 51% of patients receiving placebo (OR 2.73, 1.59, 4.69, p=0.0003). 8

Comparison of Sativex and placebo during the subsequent randomisation of responders showed significant differences in favour of Sativex for spasm frequency, sleep disruption and global impression of change subject, carer and physician scores. Spasm frequency was measured as number of episodes of spasm per day. At screening for entry to the study, the mean number of episodes a day was 10.73. At baseline for phase B, the mean number of episodes a day was 5.61 for patients receiving Sativex indicating that there was a substantial decrease in the number of episodes in responders to Sativex over the initial 4 weeks of treatment. There was a large difference between mean and median results but significance testing allowing for the skewed distribution showed a statistically significant difference for Sativex with an increase in spasm frequency reported in patients receiving placebo from phase B randomisation to end of study (p<0.0023). In addition, there was a significant difference between Sativex and placebo for the Barthel index, which measures activities of daily living. More patients receiving Sativex showed an improvement in their condition and fewer patients reported a worsening of their condition (p=0.0067). Statistical significance was not achieved for the modified Ashworth score and timed 10 metre walk but the trends favoured Sativex. Data on quality of life was collected using the EQ-5D tool. There was a 20% increase in utility from pre-treatment baseline (week 0) to end of study for patients who met the criteria for responder in phase A and entered phase B. Long term clinical evidence Two open-label non-comparative extension studies were conducted. The first study recruited patients with a range of symptoms and in the second study, only a subgroup of patients had MS. 8,9 These studies were not considered to provide reliable evidence of long term outcomes for the indication of spasticity. A placebo-controlled parallel group randomised double-blind withdrawal study of Sativex was conducted in patients with spasticity due to MS receiving long-term Sativex. 14 The mean duration of Sativex use prior to the trial was 3.62 years. The objective was to evaluate the maintenance of effect of Sativex compared with placebo as evidence of long term efficacy. At the end of a baseline period, 36 patients ceased treatment with openlabel Sativex and were randomised to receive Sativex (n=18) or placebo (n=18) for 4 weeks. The primary endpoint was time to treatment failure defined as day of premature cessation of study drug, first day of longest period when the NRS score had increased by at least 20% from baseline and day of clinically relevant increase in anti-spasticity or disease-modifying medication. Eight patients (44%) receiving Sativex were classed as failures and seventeen patients (94%) receiving placebo. Three of the patients receiving Sativex and seventeen patients receiving placebo withdrew from the study. The hazard ratio for time to treatment failure for Sativex vs placebo was 0.335 (95% CI: 0.16, 0.69). Conclusion on clinical evidence Evaluation of the efficacy of Sativex is hindered by the lack of published information on study 604 which is the only study to reflect use in clinical practice. Analysis of two parallel group studies showed that there was a significant difference between Sativex and placebo in the proportion of patients who achieved a clinically significant response but the difference was small (11%). However, this was considered to be a clinically relevant difference in a population of patients who are difficult to treat. In a trial designed to identify responders, at the end of 4 weeks of treatment with Sativex 42% of patients satisfactorily met the threshold ( 20% reduction in NRS score) for predicting patients expected to achieve a clinically significant response in the long term. In the same trial, secondary outcomes showed significant differences for Sativex compared with placebo for outcomes of relevance including spasm frequency, sleep disruption and 9

the Barthel index. Using the EQ-5D tool, responders to Sativex reported a 20% increase in utility score. None of the RCTs provided evidence on whether use of Sativex allowed for a reduction in number or dose of other anti-spasticity agents taken concomitantly because patients were required to maintain a stable regimen of anti-spasticity agents during the trial. Previous history of cannabis use was common in the parallel group studies and a low level of unblinding is a possibility. 9 Robust data on long term efficacy are lacking. Multiple sclerosis is a chronic condition but the longest trial is four months in duration. 5. Discussion of evidence on safety The pharmacokinetic profile of Sativex is very different from that of smoked cannabis. Sativex does not produce the rapid rise in plasma levels which results in the psychoactive effects seen with smoked cannabis and maximum plasma concentration achieved by Sativex is lower than for inhaled cannabis. 1 Adverse events reported in clinical trials Details of adverse events reported in study 604 which reflects clinical practice have not been provided in the MHRA assessment report. 9 Withdrawals due to adverse events are available for study 604. Nineteen of 538 patients (3.5%) taking Sativex during the four week single blind phase withdrew due to adverse events. During the double-blind phase from week 5 to week 16, 8 of 124 patients (6%) receiving Sativex withdrew due to adverse events compared with no patients receiving placebo. The MHRA assessment report has a summary of all adverse events reported in comparative clinical trials of Sativex for multiple sclerosis. 9 A total of 805 patients treated with Sativex were compared with 740 patients who received placebo. The mean duration of exposure to Sativex was 67 days. The excess over placebo of typical cannabinoid effects was reported as: dizziness (25% vs 8%), somnolence (8.2% vs 2.3%), disorientation (4.1% vs 0.8%), disturbance in attention (3.9% vs 0.1%), balance disorder (2.9% vs 1.8%), vertigo (6.5% vs 2%), fatigue (12.5% vs 8.4%), feeling drunk (3% vs 0.4%) and asthenia (5.6% vs 3.1%). The adverse event profile is reported to be broadly in line with that expected from the known pharmacology of cannabis and the high level of morbidity and polypharmacy in this patient population. In the SPC, psychiatric events reported to occur in 1% of patients and less than 10% of patients, with a plausible relationship to Sativex, include depression, disorientation, dissociation and euphoric mood. 1 No psychiatric events are listed as occurring in more than 10% of patients. Psychiatric events listed in the SPC as occurring at a frequency of 0.1% of patients to less than 1% of patients with a plausible relationship to Sativex include hallucination, paranoia, suicidal ideation and delusional perception. Sativex is contraindicated in patients with a known or suspected history or family history of schizophrenia, or other psychotic illness, history of severe personality disorder or other significant psychiatric disorder other than depression associated with their underlying condition. 1 This reflects the exclusion criteria applied during clinical trials. The SPC states that no increase in daily dosage has been observed in long term use, and patient self-reported levels of intoxication are low. 1 For these reasons, dependence on Sativex is unlikely. 10

Conclusions The most frequently reported adverse events in clinical trials are typical of cannabinoids. It is not possible to judge the effect of the slower titration schedule which will be used in clinical practice because comprehensive data on adverse events are not available from study 604. Sativex is intended for long term use but the maximum duration of the placebo-controlled trials for spasticity was four months. The high frequency of some adverse events means that it is highly likely that some non-responders who receive Sativex for the initial trial period will develop an adverse reaction. 6. Summary of the clinical case The clinical case is considered to be: generally supportable. The criteria of generally supportable requires robust, preferably reproduced, evidence of a benefit likely to be considered meaningful by patients and which is likely to be applicable to the majority of the patients. A clinical case which is generally supportable would support commissioning. Appendix One shows the criteria used to define the clinical case. In summary, there was a significant difference in favour of Sativex in the proportion of patients achieving a clinically significant response in combined analysis of the parallel group studies. In the study reflecting clinical practice, improvements in secondary outcomes for responders to Sativex, including the 20% change in utility score, indicated that Sativex resulted in a meaningful benefit to these patients. Unlike specialist options such as botulinum toxin, treatment with Sativex is not restricted to specific types of spasticity. 7. Cost Effectiveness There had been no previous published cost-effectiveness analysis assessing Sativex for the treatment of spasticity due to MS. We developed a decision-analytic model to estimate the cost-effectiveness of Sativex plus oral medicines compared to standard treatment in a UK cohort of adults who have moderate to severe spasticity due to MS and have achieved insufficient benefits from oral anti-spasticity medicines. The model takes an NHS perspective. A Markov model was developed that followed a hypothetical group of patients. The outcomes of study 306 were used as the basis for this evaluation as the SPC recommendations for use of Sativex are based on this study. 9,12,13 The model had a five year time horizon. A wide variety of costs were included (drug acquisition, drug wastage, administration cost, monitoring and treatment of adverse events). Costs and benefits were discounted at 3.5%. Sensitivity analyses were conducted to assess the effect of uncertainty around the model structure and the impact of variation in certain key parameters that were expected to have a strong influence on the cost effectiveness analysis. The effect of patients receiving expensive alternatives to Sativex was explored, the technologies considered were botulinum toxin and baclofen pump. The price of Sativex was 375 per pack which contains three spray bottles. Each bottle contains approximately 90 sprays. There was no price reduction patient access scheme. 11

Results In the base case scenario, there was little difference in the quality-adjusted life years (QALYs) gained between Sativex plus oral anti-spasticity agents and oral anti-spasticity agents alone. There was a large difference in costs with Sativex costing more. This resulted in an incremental cost-effectiveness ratio of (ICER) of 48,588 per year. A probabilistic sensitivity analysis which examined all of the uncertainties of the data simultaneously was undertaken. This included examining the effect of a response rate up to 100%. The probability of Sativex being cost-effective at a chosen threshold of 30,000 per QALY was estimated at 11.08%. Other sensitivity analyses were conducted to test changes in baseline characteristics of the cohort, time horizons, price for Sativex, daily dose, withdrawals due to adverse events, differences between health state utilities and alternative treatments. Changes in dose influenced the ICER at the minimum dose used in clinical trials the ICER was less than 25,000 but at the maximum dose the ICER was more than 60,000. The base case ICER of 48,588/QALY was estimated using the mean daily dose from clinical trials of 8 sprays a day. The price of Sativex has a large impact on the ICER. For Sativex to be considered cost-effective at a threshold of 30,000 per QALY, the cost per prescription of three bottle should be 229 or less, a reduction of 146 (38.9%) compared with the current price of 375. 8. Summary of the cost-effectiveness case The cost effectiveness case is: not cost effective. The criteria of not cost effective means that cost effectiveness considerations suggest that the technology is likely to have similar cost effectiveness to interventions that have been assessed nationally as not cost effective. Appendix One shows the criteria used to define the cost-effectiveness case. Our analysis estimates a deterministic ICER of 48,588/QALY for Sativex and an 11.08% probability of being cost effective at a willingness to pay threshold of 30k/QALY 9. Budget impact Budget impact was estimated using two different approaches. First, the budget impact model reported in the London New Drugs Group review of Sativex was applied using local prevalence data for MS. 5,22 The cost is calculated for the first year of treatment assuming that 42% of patients continue treatment long term and 58% of patients discontinue Sativex at the end of the four week trial period. Using this method, if an estimated 111 patients started treatment, the cost for the first year would be 220,651. Second, local specialists estimated that 50 patients across the south west peninsula population of 1.6 million would be considered for treatment with Sativex each year. This gives an estimated cost for the first year of 110,273. 12

10. Authors This HTA has been produced by the Peninsula Health Technology Operations Group. Clinical aspects of treatment of multiple sclerosis have been discussed with local specialists. Support for the cost effectiveness analysis has been provided by PenCLAHRC at the Peninsula College of Medicine and Dentistry through work partially funded by the National Institute of Health Research (NIHR). This collaboration gives access to an expert resource with experience of producing appraisals for NICE. The authors have no interests to declare. This work was partially funded by National Institute for Health Research (NIHR) through The Peninsula Collaboration for Leadership in Applied Health Research and Care (PenCLAHRC), which is based at the University of Exeter. This article presents independent research and the views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health. 12. References 1. Summary of Product Characteristics. Sativex Oromucosal Spray. GW Pharma Ltd. June 2010. 2. Compston A and Coles A. Multiple sclerosis. Lancet 2008; 372: 1502-1517. 3. Beard S et al. Treatments of spasticity and pain in multiple sclerosis: a systematic review. Health Technology Assessment 2003; vol 7: no. 40. 4. Crayton H et al. A multimodal approach to managing the symptoms of multiple sclerosis. Neurology 2004: 63 (11 Suppl 5): S12-8 5. Fox CM et al. The epidemiology of multiple sclerosis in Devon: a comparison of the new and old classification criteria. J Neurol Neurosurg Psych 2004 75: 56-60. 6. NICE. Multiple sclerosis. National Clinical Guideline for diagnosis and management in primary and secondary care. November 2003. National Collaborating Centre for Chronic Conditions. 7. Kavia RB, de Ridder D, Constantinescu CS et al. Randomized controlled trial of Sativex to treat detrusor overactivity in multiple sclerosis. Multiple Sclerosis 2010; 16 (11): 1349-59. 8. Wade DT et al. Do cannabis based medicinal extracts have general or specific effects on symptoms in multiples sclerosis? A double-blind, randomized, placebo-controlled study on 160 patients. Multiple sclerosis 2004; 10 (4): 434-41. 9. MHRA Public Assessment report on Sativex Oromucosal Spray. June 2010 10. Collin C, Davies P, Mutiboko IK, Ratcliffe S. Randomized controlled trial of cannabis-based medicine in spasticity caused by multiple sclerosis. European Journal of Neurology 2007, 14: 290 296. 11. Collin C, Ehler E, Waberzinek G, Alsindi Z and et al. A double-blind, randomized, placebocontrolled, parallel-group study of Sativex, in subjects with symptoms of spasticity due to multiple sclerosis. Neurological Research 2010, 32(5): 451-459 12. Montalbán X and Wright S. Trial period for new symptomatic treatments: lessons learnt from a Sativex in MS spasticity clinical trial. Mutiple Sclerosis 2009; 15: S272. 13

13. Ambler Z and et al. A two-phase study of Sativex in the relief of spasticity due to multiple sclerosis: phase A singleblind response assessment followed by phase B doubleblind, randomised, placebo-controlled, parallel-group study. Multiple Sclerosis 2009; 15: S258. 14. Notcutt W et al. Results of a randomised withdrawal study of subjects with spasticity due to multiple sclerosis who were receiving long-term Sativex. Multiple Sclerosis 2009; 15: S258 15. Aragona M et al. Psychopharmacology and cognitive effects of therapeutic cannabinoids in multiple sclerosis: a double-blind, placebo-controlled, cross-over study. Clin Neuropharmacol 2009; 32 (1): 41-7. 16. Wade DT et al. Meta-analysis of the efficacy and safety of Sativex (nabiximols) on spasticity in people with multiple sclerosis. Multiple Sclerosis 2010; 16 (6): 707-714. 17. Farrar J et al. Validity, reliability and clinical importance of change in a 0-10 numeric rating scale measure of spasticity: a post-hoc analysis of a randomised double-blind placebo-controlled trial. Clinical Therapeutics 2008 30: 5; 974-985. 18. Anwar K and Barnes MP. A pilot study of a comparison between a patient scored numeric rating scale and clinician score measure of spasticity in multiple sclerosis. Neurorehabilitation 2009; 24 94): 333-40. 19. Wade DT et al. Long term of a cannabis-based medicine in the treatment of spasticity and other symptoms in multiple sclerosis. Multiple Sclerosis 2006; 12 (5): 639-45. 20. Shakespeare D et al. Anti-spasticity agents for multiple sclerosis. Cochrane Database of Systematic Reviews 2003; issue 4. 21. Zajicek J et al. Cannabinoids for treatment of spasticity and other symptoms related to multiple sclerosis (CAMS study): multicentre randomised placebo-controlled trial. Lancet 2003; 362: 1517-1726. 22. London New Drugs Group. APC/DTC briefing document. Sativex. June 2010. 14

Appendix One: Criteria used to define the clinical and cost-effectiveness case 1. Clinical case Category Generally Supportable Supportable with restrictions Equivocal Not supportable Criteria There is robust, preferably reproduced, evidence of a benefit likely to be considered meaningful by patients and which is likely to be applicable to the majority of the patients. The clinical case would support commissioning. There is evidence of a benefit likely to be considered meaningful by patients with the condition to be treated but it is likely to be limited only to certain subgroups (which the Task Specific Group should define). Commissioning for use in the defined subgroup would be appropriate. There are data suggestive of a benefit but this is derived from studies which are not conclusive or where data are conflicting. Examples may also include The benefit may only be apparent in certain subgroups of patients which are not readily defined. Professional bodies may recommend treatment but this is based upon consensus rather than trial evidence of the direct benefit of the technology in the condition under consideration Data are available demonstrating evidence only of improvements in surrogate outcomes which may not translate into objective benefits likely to be considered meaningful by patients. There are significant difficulties and limitations to the data. Examples may include: Significantly different patient populations studied to that treated locally, significant doubts about the robustness of trial data. 2. Cost effectiveness case Category Generally cost effective Restricted cost effectiveness Not cost effective Criteria Cost effectiveness considerations suggest that under general use the technology compares favourably with other interventions that have been assessed nationally as a cost effective use of NHS resources. Cost effectiveness considerations suggest that the technology only compares favourably with other nationally assessed cost effective technologies under restricted circumstances. The task specific group should describe these circumstances. Such circumstances may include identifiable sub groups, or assumptions about outcomes that have not been proven. Cost effectiveness considerations suggest that the technology is likely to have similar cost effectiveness to interventions that have been assessed nationally as not cost effective. 15