ORIGINAL ARTICLE

A double-blind, randomized, placebo-controlled, parallel group

study of THC/CBD spray in peripheral neuropathic
pain treatment
M. Serpell1, S. Ratcliffe2, J. Hovorka3, M. Schofield4, L. Taylor5, H. Lauder5, E. Ehler6
1 Pain Clinic Office, Gartnavel General Hospital, University of Glasgow, UK
2 MAC Clinical Research, Trafford Park, Manchester, UK
3 Neurology Department, Na Fratisku Hospital, Prague, Czech Republic
4 West Suffolk Hospital, Bury St Edmunds, UK
5 GW Pharma Ltd, Porton Down Science Park, Salisbury, UK
6 Neurologické odd, Krajská nemocnice Pardubice, Pardubice, Czech Republic

Correspondence Abstract
Mick Serpell
E-mail. [email protected] Background: Peripheral neuropathic pain (PNP) associated with
allodynia poses a significant clinical challenge. The efficacy of Δ9-
Sativex, a THC/CBD fixed dose combination tetrahydrocannabinol/cannabidiol (THC/CBD) oromucosal spray, a novel
oromucosal spray, does not have an INN.
cannabinoid formulation, was investigated in this 15-week randomized,
Nabiximols is the FDA US Adopted Name
(USAN)
double-blind, placebo-controlled parallel group study.
Methods: In total, 303 patients with PNP associated with allodynia were
ClinicalTrials.gov registration number: screened; 128 were randomized to THC/CBD spray and 118 to placebo, in
NCT00710554. addition to their current analgesic therapy. The co-primary efficacy
endpoints were the 30% responder rate in PNP 0–10 numerical rating scale
Funding sources (NRS) score and the mean change from baseline to the end of treatment in
This study was sponsored by GW Pharma Ltd
this score. Various key secondary measures of pain and functioning were
(GW). GW was involved in the study design,
data collection and analysis, as well as in the
also investigated.
preparation of this manuscript and Results: At the 30% responder level, there were statistically significant
publication decisions. treatment differences in favour of THC/CBD spray in the full analysis
(intention-to-treat) dataset [p = 0.034; 95% confidence interval (CI):
Conflicts of interest 1.05–3.70]. There was also a reduction in mean PNP 0–10 NRS scores in
M. Serpell, S. Ratcliffe, J. Hovorka, both treatment groups that was numerically higher in the THC/CBD spray
M. Schofield and E. Ehler were all
group, but which failed to reach statistical significance. Secondary
investigators in this study and received
investigator fees from GW accordingly for
measures of sleep quality 0–10 NRS score (p = 0.0072) and Subject Global
their participation in the study. GW medical Impression of Change (SGIC) (p = 0.023) also demonstrated statistically
writers L. Taylor and H. Lauder undertook the significant treatment differences in favour of THC/CBD spray treatment.
initial compilation and quality control review Conclusions: These findings demonstrate that, in a meaningful
of the manuscript. Together with the other proportion of otherwise treatment-resistant patients, clinically important
authors, the target journal was then agreed improvements in pain, sleep quality and SGIC of the severity of their
and all authors reviewed and contributed to
condition are obtained with THC/CBD spray. THC/CBD spray was well
the content of the manuscript, and agreed
upon the final submitted version. All
tolerated and no new safety concerns were identified.
Intellectual Property Rights arising out of the
current clinical study are vest in or exclusively
licensed to GW.

Accepted for publication

22 November 2013

doi:10.1002/j.1532-2149.2013.00445.x

© 2014 European Pain Federation - EFIC® Eur J Pain 18 (2014) 999–1012 999
Efficacy of THC/CBD spray in peripheral neuropathic pain M. Serpell et al.

What’s already known about this topic? a limited effect on PNP, and the side-effect problems
• Neuropathic pain is a debilitating form of chronic associated with each are well known.
pain and can be difficult to treat, with only The endocannabinoid system modulator, Δ9-
approximately half of sufferers achieving partial tetrahydrocannabinol (THC)/cannabidiol (CBD) oro-
relief, often requiring the use of novel analgesics mucosal spray, is formulated from plant-based extracts
due to the ineffectiveness of conventional prepared from genetically distinct chemotypes of Can-
pharmacotherapies. nabis sativa L. and contains an approximately 1:1 ratio
• Cannabinoids, including Δ9-tetrahydrocanna- of THC : CBD, plus smaller amounts of other com-
binol/cannabidiol (THC/CBD) spray, have dem- pounds, including minor cannabinoids and terpenes
onstrated efficacy in addressing this unmet need. (Russo, 2011). It was recently licensed for use in
A previous randomized controlled trial in neuro- various European countries for the relief of spasticity
pathic pain patients demonstrated positive effects in multiple sclerosis (MS) (MHRA Public Assessment
in pain and allodynia at 5 weeks. Report, 2010), as well as outside the European Union
(in Canada, Israel, New Zealand). THC/CBD spray is
also licensed for use in Canada for the treatment of
What does this study add?
central neuropathic pain (CNP) in MS patients.
• The study demonstrates that THC/CBD spray can
Cannabinoids are thought to act primarily via spe-
provide clinically relevant improvements in pain,
cific receptors, designated cannabinoid receptor-1
sleep quality and patient global impression of the
(CB1) and cannabinoid receptor-2 (CB2). CB1 receptors
change in their condition in a meaningful pro-
are predominantly distributed throughout the nervous
portion of usually treatment-resistant patients.
systems, while CB2 receptors are primarily located in
• This supports the hypothesis that THC/CBD
the periphery, especially the immune system (Howlett
could be a useful candidate for peripheral neu-
et al., 2002).
ropathic pain treatment, demonstrating efficacy
Cannabinoids are postulated to offer a new thera-
in a few key outcomes over a much longer period
peutic approach to neuropathic pain treatment. Previ-
of time (15 weeks compared to 5 weeks).
ous studies using synthetic THC and a synthetic
metabolite of THC demonstrated effects in patients on
CNP (Svendsen et al., 2004) and PNP associated with
allodynia (Karst et al., 2003), respectively. Further-
1. Introduction
more, in a previous randomized controlled trial (RCT)
Neuropathic pain is a chronic, debilitating and wide- (Rog et al., 2005) and in an open-label extension
spread condition with an estimated prevalence of over study (Rog et al., 2007), GW has shown that THC/
1% (Backonja and Serra, 2004). Two recent CBD spray has pain relieving effects in neuropathic
population-based studies in Europe estimated the pain associated with MS and in difficult to treat pain
prevalence of chronic neuropathic pain, or pain with following brachial plexus avulsion (Berman et al.,
neuropathic characteristics, to be 8% and 7%, respec- 2004). In addition, a previous 5-week GW study of
tively (Torrance et al., 2006; Bouhassira et al., 2008). THC/CBD spray in the treatment of PNP concluded
Neuropathic pain can be triggered by a variety of dis- that THC/CBD spray is an effective treatment, which
eases and conditions, but the mechanisms that estab- provided a rapid clinically relevant improvement
lish and maintain it are specific to the characteristics of (Nurmikko et al., 2007).
the damage and/or dysfunction of the nervous system. The objectives of this study were to investigate the
Allodynic pain, characterized as pain evoked by a nor- therapeutic benefits of 15-week THC/CBD spray treat-
mally non-nociceptive stimulus (such as tempera- ment on PNP associated with allodynia, as well as
ture), is a subgroup of peripheral neuropathic pain associated sleep disturbance and patient quality of life.
(PNP) and can be very difficult to treat.
A mechanistic approach to neuropathic pain is cur-
rently believed to represent the optimal means of 2. Methods
symptom management (Jensen et al., 2001; Woolf
and Max, 2001). However, there is little clinical proof 2.1 Study design
that this approach is the most effective strategy. Exist- This was a 15-week (1-week baseline and 14-week treat-
ing therapies for PNP include tricyclic and related anti- ment period), multi-centre, double-blind, randomized,
depressants, anti-epileptic agents and opioids (Attal placebo-controlled, parallel group study to evaluate the effi-
et al., 2006). However, these therapies may have only cacy of THC/CBD spray in patients with PNP associated with

1000 Eur J Pain 18 (2014) 999–1012 © 2014 European Pain Federation - EFIC®
M. Serpell et al. Efficacy of THC/CBD spray in peripheral neuropathic pain

allodynia. The study took place at 21 centres in the United psychiatric, renal, hepatic, cardiovascular or convulsive dis-
Kingdom (UK), seven centres in Czech Republic, six centres orders, or with a known hypersensitivity to the study medi-
in Romania, four centres in Belgium and one centre in cation. Those with CRPS type 1, cancer-related PNP or pain
Canada. The study was approved by the relevant Institution resulting from diabetes mellitus were excluded. Patients
Review Board or Ethical Committee in each country and was receiving a prohibited medication [including cannabis or
conducted in accordance with the principles of the Declara- cannabinoid-based medications (in the last year), any anal-
tion of Helsinki and the International Conference on Har- gesics taken on a ‘PRN’ (when required) basis, the introduc-
monization Good Clinical Practice guidelines. All patients tion of any new analgesic medication, or any alteration to
provided written informed consent to take part in the study. the dosage of the patient’s concomitant analgesic medication
All visits took place at study centres. Following eligibility (other than the rescue analgesia provided), or all
screening, patients completed a 7-day baseline period. paracetamol-containing medications (stopped on the day the
Patients were then assessed, randomized and received dose patient entered the baseline period)], who were unwilling to
introduction. Visits occurred at the end of weeks 2, 6, 10 and abstain for the study duration were also excluded, as were
at the end of the study (treatment week 14) or earlier if they those with a known history of alcohol or substance abuse.
withdrew. A follow-up visit occurred 28 days after study Women of child-bearing potential or their partners were
completion or withdrawal. Patients were then given the excluded unless willing to ensure effective contraception was
opportunity to enrol in an open-label extension study. used throughout the study, as were those who had received
Results from the open-label extension study will not be an investigational medicinal product within 12 weeks of
presented in this report. screening. Pregnant or lactating women and those planning
At each visit, the following information was recorded: a pregnancy were excluded. Patients with any physical
adverse events (AEs), vital signs, intoxication 0–10 numeri- abnormality at screening (i.e., any abnormalities that, in the
cal rating scale (NRS), sleep quality 0–10 NRS, PNP 0–10 opinion of the investigator, would prevent the patient from
NRS, neuropathic pain scale (NPS), use of rescue analgesia, safely participating in the study), or those intending to travel
any changes in current medical conditions, dose of regular or donate blood during the study were also ineligible to take
maintenance analgesic, changes in concomitant medication, part.
current dose of study medication and medication compli-
ance. Clinical laboratory sampling (haematology, biochemis-
try and urinalysis) was carried out at screening and at the 2.3 Study medication and procedures
end of treatment.
A pump action oromucosal spray was used to deliver study
medication. Each 100 μL spray of THC/CBD delivered 2.7 mg
2.2 Inclusion and exclusion criteria of THC and 2.5 mg of CBD to the oral mucosa, and each
spray of placebo delivered the excipients plus colorants. Both
2.2.1 Inclusion criteria THC/CBD spray and placebo contained peppermint oil to
blind the smell and taste. Patients self-administered the
Eligible patients were aged 18 or older, had mechanical allo-
medication to their optimal dose, but were restricted to a
dynia within the territory of the affected nerve(s) (confirmed
maximum of eight sprays in a 3-h period up to a maximum
by either a positive response to stroking the allodynic area
of 24 sprays per 24-h period. Initially, patients began at a
with a SENSELABTM Brush 05 (Somedic AB, Hörby, Sweden)
maximum of one spray per 4-h period. Thereafter patients
or to force applied by a 5.07 g Semmes-Weinstein monofila-
were advised to self-titrate their medication to symptom
ment), at least a 6-month history of PNP, and were receiving
relief or maximum dose, but increases were limited to a
the appropriate treatment for their PNP. Eligible patients had
maximum of 50% of the previous day’s dose.
at least one of the following underlying conditions, which
caused their PNP: post-herpetic neuralgia, peripheral neu-
ropathy, focal nerve lesion, radiculopathy or Complex 2.3.1 Concomitant medications
Regional Pain Syndrome (CRPS) type 2. Patients also had a
sum score of at least 24 on a pain 0–10 NRS for more than 6 As would be expected in this group of patients, many were
days (baseline days 2–7) during the baseline period (average receiving concomitant medications for analgesia and were
0–10 NRS score of 4/10), and pain that was not wholly allowed to continue their concomitant analgesic medication,
relieved by their current therapy. In addition, their analgesic with the exception of paracetamol (acetaminophen), pro-
regimen was stable for at least 2 weeks preceding study entry vided that a stable dose was maintained throughout the
and they were willing for the responsible authorities (i.e., study. Patients were not permitted to take analgesics on a
primary care consultant or physician) to be notified of their ‘PRN’ (when required) basis, and the introduction of any
participation in the study. new analgesic medication or any alteration to the dosage of
the patients’ concomitant analgesic medication (other than
the rescue analgesia provided) was prohibited during the
2.2.2 Exclusion criteria
study. The rescue analgesia provided contained paracetamol
Patients with severe pain from other concomitant conditions Ph Eur 500 mg. The maximum single dose was two 500 mg
were excluded, as were those with a history of significant tablets, and the maximum total daily dose was 4 g (i.e., 8

© 2014 European Pain Federation - EFIC® Eur J Pain 18 (2014) 999–1012 1001
Efficacy of THC/CBD spray in peripheral neuropathic pain M. Serpell et al.

tablets per day). A single dose was not to be taken more

2.4.2.2 Sleep quality 0–10 NRS
frequently than every 4 h, with no more than four doses in
any 24-h period. Sleep quality was assessed at all study visits on a 0–10 NRS,
with the main variable for analysis being the change from
baseline to the end of treatment in sleep quality 0–10 NRS
2.4 Study endpoints score. The sleep quality 0–10 NRS was completed at the same
time each day, i.e., bedtime in the evening. The patient was
2.4.1 Primary efficacy endpoints asked ‘on a scale of “0 to 10”, please indicate how your pain
disrupted your sleep last night’, with the anchors: 0 = ‘did
In this study, a 0–10 NRS was used as the primary measure
not disrupt sleep’ and 10 = ‘completely disrupted (unable to
of pain severity. The efficacy endpoints for analysis were the
sleep at all)’.
proportion of patients showing a 30% or more improvement
from baseline to the end of treatment in PNP 0–10 NRS
score, and the mean change in PNP 0–10 NRS score from 2.4.2.3 SGIC
baseline to the end of treatment. End of treatment PNP 0–10 At baseline, patients wrote a brief description of their pain
NRS scores were the average of all scores during the last 7 caused by peripheral neuropathy, which was used at the end
days of the evaluable treatment period. of treatment to aid their memory regarding their symptoms
The PNP 0–10 NRS was recorded daily by patients in their at the start of the study. The SGIC was completed at the end
diary books. Each patient was instructed to complete their of treatment. A 7-point Likert-type scale was used to evalu-
PNP 0–10 NRS score by reviewing their day’s pain at the end ate the patients’ perception of their condition, and patients
of every day. Patients were asked, ‘On a scale of “0 to 10”, were asked, ‘Please assess the status of your pain due to
please indicate the average level of your nerve pain over the peripheral neuropathy since entry into the study using the
last 24 h’, with the anchors: 0 = ‘no pain’, 10 = ‘worst pos- scale below’, with the anchors: ‘very much improved’,
sible pain’. The assessment reviewed the entire day’s pain, ‘much improved’, ‘slightly improved’, ‘no change’, ‘slightly
and therefore, the perception of pain was less likely to be worse’, ‘much worse’ or ‘very much worse’.
influenced directly by sleep, compared with an assessment
made on waking. Patients were instructed to relate ‘no pain’
to the time prior to their onset of their PNP associated with
2.4.2.4 BPI-SF
allodynia. The BPI-SF (Cleeland and Ryan, 1994) was performed twice,
once at baseline and once at the end of treatment, with the
change in score between these time points being the variable
2.4.2 Secondary efficacy endpoints for analysis. The BPI-SF consists of nine questions, each of
Secondary endpoints included the mean changes from base- which consists of a single response apart from question 9,
line to the end of treatment in the following scores: NPS, which is sub-divided into seven parts (9A–9G). Questions
sleep quality 0–10 NRS, Subject Global Impression of Change 3–6 ask patients to rate pain on a 0–10 scale over the prior
(SGIC), Brief Pain Inventory (short form) (BPI-SF), dynamic week (where 0 = ‘no pain’ and 10 = ‘pain as bad as you can
and punctate allodynia tests, quality of life (EQ-5D) health imagine’). Severity is measured as worst pain, least pain,
questionnaire, as well as the proportion of patients showing average pain and pain right now. The severity composite
a 50% or more improvement in PNP 0–10 NRS score, and score was calculated as the arithmetic mean of the four
the use of rescue analgesia. severity items (range 0–10). The minimum value is zero and
maximum is 10.
The BPI-SF also records the degree to which pain interferes
2.4.2.1 NPS with activities on a 0–10 scale (where 0 = ‘does not interfere at
all’ and 10 = ‘pain completely interferes with activity’). As
The NPS (neuropathic pain scale PDF) was collected weekly such, a higher score represents a poorer outcome.
in the patient diaries during the whole length of the study. Two composite scores were calculated from the BPI-SF:
The variable for analysis was the change in mean NPS score (1) The pain severity composite score: the arithmetic mean
from baseline (mean of two assessments during the baseline of the four pain scores (questions 3–6) and represents the
period) to the end of the study (mean of last two assessments pain intensity.
during the evaluable period). (2) The pain interference composite score: the arithmetic
The NPS consists of 10 individual items. Nine of these mean of the seven interference items (questions 9A–9G) and
provide a total of ten 0–10 NRS responses and there is a represents the effect of pain.
multi-part free text question. The NPS score to be used for
the analysis was the sum of the ten 0–10 NRS responses. If
2.4.2.5 Dynamic allodynia test
up to three individual items were missing, then an NPS score
was imputed by multiplying the mean of the completed The dynamic allodynia test was performed twice, once at
items by 10. If more than three individual items were baseline and once at the end of treatment, with the change in
missing, then the whole score was missing. score between these time points being the variable for analy-

1002 Eur J Pain 18 (2014) 999–1012 © 2014 European Pain Federation - EFIC®
M. Serpell et al. Efficacy of THC/CBD spray in peripheral neuropathic pain

sis. At each time point, dynamic allodynia was assessed by post-treatment, vital signs, oral examination and intoxica-
stroking the skin over the affected area five times with a tion 0–10 NRS.
SENSELAB Brush 05, designed specifically for sensory testing
at 5-s intervals, and recording the pain severity on a 0–10
NRS, where 0 = ‘no pain’ and 10 = ‘most pain imaginable’. All 2.4.4 Sample size
strokes were of the same length, minimum 2 cm. The mean of
Based upon previous GW studies, it was believed that this
the five scores for the identified allodynic area only was
study would result in a difference in the primary endpoint
calculated to define the dynamic allodynia pain score.
between THC/CBD spray and placebo patients of at least 0.9
points on the PNP 0–10 NRS. Also based on previous GW
2.4.2.6 Punctate allodynia test studies and the literature, it was estimated that the standard
The punctate allodynia test was performed twice, once at deviation of the changes from baseline in the primary end-
baseline and once at the end of treatment, with the change point would be approximately 2.1 points (Rowbotham et al.,
in score between these time points being the variable for 1998; Rice et al., 2001; Serpell and Neuropathic Pain Study
analysis. Punctate allodynia was measured using an in-house Group, 2002; Boureau et al., 2003). Taking this into account,
built pressure algometer comprising a strain gauge connected for a significance level of 5% and 80% power, we would need
to a metal filament with a diameter of 1 mm and blunt tip at a total of 174 evaluable patients (87 in each group) to detect
baseline and end of study. The filament was manually a difference of 0.9 points in the PNP 0–10 NRS. Allowing for
directed against the skin at an angle of 90° and a steadily 20% of randomized patients to be unevaluable, then 218
increasing pressure was applied until the patient verbally patients (109 in each group) would need to be randomized.
indicated that they perceived pain (punctate pressure pain
threshold). Patients were asked to verbally rate the intensity
of the pain elicited, choosing a number between 0 = ‘no pain’
2.5 Method of assigning patients to treatment
and 10 = ‘most intense pain imaginable’. The average of the groups and blinding
ascending pain threshold forces, as available, for the identi- Patients were randomized to receive either THC/CBD spray
fied allodynic area only was calculated to define the punctate or placebo. Randomization was carried out using a predeter-
allodynia pain threshold force. mined computer-generated randomization code, produced
by the GW Biometrics Department, in which treatment allo-
2.4.2.7 EQ-5D questionnaire cation was made using permuted blocks of four. Study medi-
cation was pre-packed by the GW Clinical Trial Supplies
The EQ-5D questionnaire (The Euroqol Group, 1990) was Department and dispatched to the investigator centres
completed twice during the study, once at baseline and once labelled with patient numbers. The randomization scheme
at the end of treatment. involved patient numbers being assigned sequentially by the
The EQ-5D questionnaire provided two outcomes: investigator staff.
(1) A weighted health state index visual analogue scale Study medication was provided in 5.5-mL type I amber
(VAS). glass vials labelled with the GW name, study code, patient
(2) A self-rated health status VAS. number, visit number and the expiry date. The investigator
The self-rated health status VAS anchors were: 0 = ‘worst staff, pharmacy and GW Clinical Department held sealed
health state imaginable’ to 100 = ‘best health state imagin- code break envelopes for each patient. Since THC/CBD spray
able’. The weighted health state index used the same VAS as is a plant-based extract in alcoholic solution with a distinc-
above but was calculated for each assessment without impu- tive smell, taste and colour, both THC/CBD spray and
tation to account for missing values, i.e., if one or more placebo contained peppermint oil to blind the smell and
individual items were missing, then the whole index was taste. The placebo also contained quinoline and sunset
missing. yellow, to match the colour of the plant extract. As such,
The change from baseline to the end of treatment was participants, investigators are caregivers were all blinded to
calculated for both VASs. the treatment allocation.

2.4.2.8 Use of rescue analgesia

2.6 Statistical methods
Use of breakthrough medication was recorded daily during
the study as the number of paracetamol tablets taken. The All randomized patients who received at least one dose of
change in mean daily quantities of tablets used was calcu- test treatment and had on-treatment efficacy data were
lated from baseline to the last 7 days of treatment. included in the intention-to-treat (ITT) analysis set. The per
protocol (PP) analysis set included those with evaluable data
for the primary parameter with no protocol deviations,
2.4.3 Safety endpoints
which were considered to affect the comparison between
The safety endpoints were the incidence of AEs and serious treatments for this endpoint. All summaries and statistical
adverse events (SAEs), clinical laboratory sampling pre- and analyses were performed using SAS Version 9.1 (SAS Insti-

© 2014 European Pain Federation - EFIC® Eur J Pain 18 (2014) 999–1012 1003
Efficacy of THC/CBD spray in peripheral neuropathic pain M. Serpell et al.

tute Inc., Cary, NC, USA). Statistical comparisons of efficacy and 52 withdrew. Six patients (one taking placebo and
data between treatments used two-sided statistical tests at five taking THC/CBD spray) were not included in the
the 5% significance level. PNP 0–10 NRS scores were evalu- analysis as they had no on-treatment efficacy data. A
ated by analysis of covariance (ANCOVA), with baseline summary of the flow of the trial can be found in Fig. 1.
values as covariate and treatment group and centre group as
The mean duration of the underlying neuropathic con-
main effect. These tests were performed at the 10% signifi-
dition in these patients was similar between treatment
cance level as a possible indicator of an interactive effect. An
additional analysis was performed on the PNP 0–10 NRS
groups at approximately 6 years with the minima and
dataset to assess the time course of the treatment effect using maxima also being similar at 0.6–38.1 years for THC/
repeated measures. A multivariate linear model was used CBD spray and 0.4–39.3 years for placebo groups,
with a separate unstructured covariance matrix in each respectively. The duration of their treatment-resistant
treatment arm. The mean (fixed effects structure) incorpo- neuropathic pain was also similar and no notable dif-
rated full treatment-by-(categorical) time interaction. Base- ferences in the proportions of patients with each type of
line was included as a covariate, together with baseline-by- underlying condition were seen between treatment
time interaction. Grouped centre was included as a groups, the most common of which was focal nerve
categorical covariate. The fitted model was also used to lesions for both groups. These and other study popula-
produce a final time point comparison.
tion demographics are displayed in Table 1. Overall, the
Changes from baseline to the end of treatment were
mean daily dose of THC/CBD spray was 8.9 sprays and
compared between treatment groups using ANCOVA for
the following secondary endpoints: NPS, dynamic allodynia
for placebo was 14.2 sprays, and the median duration of
pain score, punctate allodynia pain score, BPI-SF, sleep treatment was 78.2 days for THC/CBD spray and 86.4
quality 0–10 NRS and EQ-5D. Models included treatment days for placebo.
and centre group as factors and baseline mean usage as a
covariate.
3.1 Concomitant medication
The change from baseline in mean daily quantity of rescue
analgesia usage was analysed in a fashion similar to the PNP The majority of patients (90% overall) continued to
0–10 NRS. take analgesics during the study. The most commonly
In the SGIC outcome, the two treatment groups were reported classes of analgesic were non-selective mono-
compared using ordinal logistic regression and the propor-
amine reuptake inhibitors (tricyclic antidepressants)
tional odds model, incorporating centre group.
taken by 26% of patients, anti-epileptics (pregabalin)
taken by 20% of patients and other anti-epileptics
2.7 Amendments during trial (gabapentin) taken by 23% of patients. In addition,
The following inclusion criterion was removed: ‘Subject has at 19% and 18% of patients, respectively, took natural
least moderate PNP, which is defined as the total of the two opium alkaloids (such as dihydrocodeine) and other
NPS scores before randomization being at least 80’. After opioids (mostly tramadol). The most commonly
ethics approval had been granted for the study, the Commit- reported classes of non-analgesic concomitant medi-
tee for Medicinal Products for Human Use (CHMP) Guideline cation were proton pump inhibitors (18%), HMG
on Clinical Investigation of Medicinal Products Intended for Co-A reductase inhibitors (statins, 15%), angiotensin-
the Treatment of Neuropathic Pain were finalized and issued converting enzyme inhibitors (14%) and beta block-
(CPMP guideline, 2004). The CHMP guidance notes clearly ing agents (13%).
recommended that the 0–10 NRS should be used as the
primary efficacy endpoint. Therefore, to have an entry crite-
rion of the two NPS scores before randomization being at least 3.2 Primary endpoint: 30% responder analysis
80 in addition to the minimum 0–10 NRS pain scores was and change from baseline to the end of
considered futile. The NPS was still collected as a secondary treatment in PNP 0–10 NRS
outcome measure and analysed and reported accordingly.
A total of 34 patients (28%) receiving THC/CBD spray
were classified as responders at the 30% level com-
3. Results
pared with 19 patients (16%) on placebo. Responder
The study took place between 27 September 2005 and analysis at this level showed a statistically significant
18 October 2006. In total, 303 patients were recruited treatment difference in the evaluable period for the
and 246 were randomized and analysed at 39 study ITT population with an odds ratio of 1.97 (p = 0.034;
centres. Of these, 128 received THC/CBD spray, 118 95% CI: 1.05–3.70), in favour of THC/CBD spray
received placebo and 57 were withdrawn before ran- treatment (Table 2). This finding was supported by the
domization. A total of 173 patients completed the PP analysis set, in which 27 (36%) of patients in the
study, 21 ceased treatment but remained in the study, THC/CBD spray treatment group achieved at least a

1004 Eur J Pain 18 (2014) 999–1012 © 2014 European Pain Federation - EFIC®
M. Serpell et al. Efficacy of THC/CBD spray in peripheral neuropathic pain

Screened
(n = 303)

Excluded (n = 57)
Did not meet entry
criteria: 43
7-day baseline period Withdrew consent: 11
Other: 3

Randomized
(n = 246)

THC/CBD
Placebo
spray
(n = 118)
(n = 128)

Withdrawal (n = 24)
Withdrawal (n = 49)
Adverse event: 7
Adverse event: 24
Withdrew consent: 3
Withdrew consent: 7 98 days of treatment Lost to follow-up: 1
Lost to follow-up: 7
Lack of efficacy: 12
Lack of efficacy: 11
Other: 1

Completed Completed
(n = 79) (n = 94)

Intention-to-treat analysis
set (n = 240)
Per protocol analysis set
Figure 1 Breakdown of patients enrolled in (n = 163)
Safety set (n = 246)
the study.

30% improvement in 0–10 NRS pain scores compared responders was observed to increase much more
with 18 (20%) in the placebo treatment group, with quickly in relation to the dose of THC/CBD spray com-
an odds ratio of 2.27 (p = 0.021; 95% CI: 1.12–4.57) pared with placebo, as illustrated in Fig. 2. At a point
(Table 2). For 30% responders, the proportion of of around 14–15 sprays per day, the response rate in

Table 1 Demographics and baseline characteristics for all patients who took part in the study.

THC/CBD spray Placebo Total

(n = 128) (n = 118) (n = 246)

No. of patients (%)

Gender
Male 43 (34) 53 (45) 96 (39)
Female 85 (66) 65 (55) 150 (61)
Ethnic origin
White/Caucasian 127 (99) 116 (98) 243 (99)
Black/African American 0 2 (2) 2 (1)
Other 1 (1) 0 1 (< 0.5)
Previous cannabis use in the last year 13 (10) 12 (10) 25 (10)
Type of underlying condition causing neuropathic pain
Post-herpetic neuralgia 34 (27) 30 (25) 64 (26)
Peripheral neuropathy 35 (27) 25 (21) 60 (24)
Focal nerve lesion 44 (34) 52 (44) 96 (39)
Complex regional pain syndrome-II 17 (13) 14 (12) 31 (13)

Mean (SD)

Age (years) 57.6 (14.4) 57.0 (14.1) 57.3 (14.2)

Body mass index (kg/m2) 28.4 (6.5) 27.3 (4. 9) 27.9 (5.8)
Duration of neuropathic condition (years) 6.3 (6.7) 6.3 (6.4) 6.3 (6.6)
Duration of peripheral neuropathic condition (years) 5.7 (6.3) 5.2 (5.4) 5.5 (5.9)

CBD, cannabidiol; THC, Δ9-tetrahydrocannabinol.

© 2014 European Pain Federation - EFIC® Eur J Pain 18 (2014) 999–1012 1005
Efficacy of THC/CBD spray in peripheral neuropathic pain M. Serpell et al.

Table 2 Summary of the analysis of all primary and secondary efficacy endpoints (ITT and PP analysis sets). Treatment differences between THC/CBD
spray and placebo are presented using change from baseline to the end of treatment data for each endpoint, unless otherwise stated.

Endpoint ITT analysis set PP analysis set

Primary endpoints

Odds ratio 95% CI p-value Odds ratio 95% CI p-value

30% responder analysis (PNP 0–10 NRS) 1.970 1.049 to 3.702 0.034 2.266 1.124 to 4.568 0.021

Treatment Treatment
difference (SE) 95% CI p-value difference (SE) 95% CI p-value

PNP 0–10 NRS −0.34 (0.230) −0.79 to 0.11 0.139 −0.48 (0.303) −1.08 to 0.12 0.116

Secondary endpoints

Treatment Treatment
difference (SE) 95% CI p-value difference (SE) 95% CI p-value

NPS −2.86 (2.211) −7.22 to 1.50 0.198 −5.26 (2.873) −10.94 to 0.41 0.069
Sleep quality 0–10 NRS −0.83 (0.306) −1.43 to −0.23 0.007 −0.91 (0.369) −1.63 to −0.18 0.015
BPI-SF (pain severity composite score) −0.25 (0.236) −0.72 to 0.21 0.288 −0.27 (0.291) −0.85 to 0.30 0.349
BPI-SF (average pain) −0.34 (0.237) −0.81 to 0.12 0.148 −0.47 (0.299) −1.06 to 0.13 0.122
BPI-SF (worst pain) −0.30 (0.265) −0.82 to 0.22 0.255 −0.39 (0.322) −1.02 to 0.25 0.234
BPI-SF (pain interference composite score) −0.32 (0.241) −0.80 to 0.15 0.183 −0.39 (0.304) −0.99 to 0.21 0.204
Dynamic allodynia test 0.08 (0.305) −0.52 to 0.68 0.795 −0.27 (0.359) −0.98 to 0.44 0.460
Punctate allodynia test −0.14 (0.118) −0.37 to 0.09 0.233 −0.06 (0.150) −0.35 to 0.24 0.701
EQ-5D (weighted health status index VAS) −0.01 (0.024) −0.06 to 0.04 0.617 – – –
EQ-5D (self-rated health status VAS) −0.75 (2.459) −5.60 to 4.09 0.760 – – –
Use of rescue analgesia −0.38 (0.237) −0.85 to 0.09 0.112 0.40 (0.316) −1.02 to 0.23 0.211

Odds ratio 95% CI p-value Odds ratio 95% CI p-value

50% responder analysis (PNP 0–10 NRS) 1.699 0.645 to 4.476 0.280 2.045 0.750 to 5.576 0.157
SGIC (end of treatment only) 1.762 1.080 to 2.876 0.023 2.988 1.661 to 5.378 0.0003

BPI-SF, Brief Pain Inventory (short form); CBD, cannabidiol; CI, confidence interval; ITT, intention-to-treat; NRS, numerical rating scale; PNP, peripheral
neuropathic pain; PP, per protocol; SGIC, Subject Global Impression of Change; THC, Δ9-tetrahydrocannabinol; VAS, visual analogue scale.

patients receiving THC/CBD spray slowed, while for ITT and PP datasets, the adjusted mean reduction in
those taking placebo, the proportion of responders was PNP 0–10 NRS score gave respective estimated treat-
still increasing maximally. ment differences of −0.34 points (p = 0.14; 95% CI:
In the co-primary endpoint of change from baseline −0.79 to 0.11 points) and −0.48 points (p = 0.12; 95%
to the end of treatment in PNP 0–10 NRS score, for the CI: −1.08 to 0.12 points), in favour of a benefit with

Figure 2 Cumulative percentage of respond-

ers at the 30% level by mean sprays.

1006 Eur J Pain 18 (2014) 999–1012 © 2014 European Pain Federation - EFIC®
M. Serpell et al. Efficacy of THC/CBD spray in peripheral neuropathic pain

Table 3 Sleep quality ratings by study visit, ITT and PP datasets.

Adjusted mean change from baseline

THC/CBD spray Placebo Treatment difference Lower and upper limits

(n = 122) (n = 117) (THC/CBD spray vs. placebo) 95% CI

Time point ITT

Visit 3 (day 15) −1.44 −0.73 −0.70 −1.22, −0.19

Visit 4 (day 43) −1.45 −0.74 −0.71 −1.31, −0.11
Visit 5 (day 71) −1.39 −0.66 −0.74 −1.34, −0.13
Visit 6 (day 99) −1.47 −0.69 −0.78 −1.36, −0.21
Final visit (day 127) −1.57 −0.74 −0.83 −1.43, −0.23

PP

(n = 73) (n = 89)

Visit 3 (day 15) −1.46 −0.81 −0.65 −1.30, −0.01

Visit 4 (day 43) −1.62 −0.83 −0.78 −1.58, 0.01
Visit 5 (day 71) −1.52 −0.71 −0.81 −1.58, −0.03
Visit 6 (day 99) −1.49 −0.58 −0.91 −1.63, −0.18
Final visit (day 127) −1.49 −0.58 −0.91 −1.63, −0.18

CBD, cannabidiol; CI, confidence interval; ITT, intention-to-treat; PP, per protocol; THC, Δ9-tetrahydrocannabinol.

THC/CBD spray treatment. However, these failed to Decreases (improvements) in favour of the THC/
reach statistical significance. CBD spray group were also observed in the following
parameters: NPS total score, mean number of tablets of
3.3 Secondary efficacy analysis rescue medication administered, BPI-SF scores (pain
severity composite score, average pain, worst pain and
At the 50% responder level in the PNP 0–10 NRS score pain interference composite score) and EQ-5D ques-
analysis, the treatment difference was also in favour of tionnaire scores (both weighted health status index
the THC/CBD spray treatment group in both the ITT VAS and self-rated health status VAS). These results
and the PP populations, but did not reach statistical applied to both ITT and PP population analysis sets, but
significance in either population (Table 2). none reached statistical significance (Table 2). The
For the ITT complete period, the adjusted mean dynamic allodynia test score increased (improved) in
sleep quality 0–10 NRS score decreased (improved) by the ITT analysis set but was not in favour of active
1.57 points from a mean baseline score of 5.4 points in treatment in the PP analysis set (Table 2).
the THC/CBD spray group, compared with an adjusted Interestingly, there was an apparent treatment by
decrease of 0.74 points from a baseline of 5.8 points in centre interaction in the changes from baseline to the
the placebo group. The estimated treatment difference end of treatment in sleep quality 0–10 NRS (p = 0.016)
was −0.83 points, in favour of THC/CBD spray, a and BPI-SF scores (p = 0.079) (in the domain of ‘pain
highly statistically significant result compared with interference composite’), with an apparent treatment
placebo (p = 0.0072; 95% CI: −1.43 to −0.23 points) effect in the UK but not elsewhere (data not shown).
(Table 3). In the PP population, the treatment differ-
ence was slightly greater, in favour of THC/CBD spray,
3.4 Safety and tolerability
and was also statistically significant compared with
placebo (−0.91 points, p = 0.015; 95% CI: −1.63 to All AEs experienced by patients with an incidence of
−0.18 points) (Table 3). 3% or greater during this study are displayed in
In the secondary efficacy analysis of SGIC, there was Table 4. The most common system organ classes
a statistically significant treatment difference in favour (SOCs) affected for treatment-related AEs were
of THC/CBD spray in the ITT dataset, compared with ‘nervous system disorders’, ‘gastrointestinal disorders’,
placebo (odds ratio: 1.76; p = 0.023; 95% CI: 1.08– ‘general disorders and administration site conditions’,
2.88) that was mirrored in the PP population, with the ‘infections and infestations’ and ‘psychiatric disorders’.
odds ratio in favour of THC/CBD spray increasing to ‘Psychiatric disorders’ were experienced by 36 (28%)
2.99 compared with placebo (p = 0.0003; 95% CI: patients receiving THC/CBD spray versus only 11
1.66, 5.38). The proportion of patients selecting each (9%) receiving placebo. By preferred term, dissocia-
category is presented in Fig. 3. tion [nine (7%) THC/CBD spray patients affected vs.

© 2014 European Pain Federation - EFIC® Eur J Pain 18 (2014) 999–1012 1007
Efficacy of THC/CBD spray in peripheral neuropathic pain M. Serpell et al.

70

60
Proportion of patients (%)

50

40

30 THC/CBD spray
Placebo
20

10

0
Much Slightly No change Slightly Much Very much Figure 3 Subject global impression of change,
worse worse improved improved improved intention-to-treat complete period.

no placebo patients] and disorientation [eight (6%) therapy faces enormous challenges to modify signifi-
THC/CBD spray patients affected vs. no placebo cantly the changes established within the nervous
patients] were the most commonly reported AEs in system. Despite these limiting factors, this study
this SOC (Table 4). Additionally, other SOCs were confirms the results previously reported, showing THC/
more commonly affected in the THC/CBD spray versus CBD spray to produce a clinically relevant improve-
placebo arms, notably ‘nervous system disorders’, ment (30% or more) in mean daily pain in a
‘gastrointestinal disorders’ and ‘general disorders and significantly greater proportion of patients than
administration site conditions’ (Table 4). placebo when administered in addition to existing
The majority of treatment-emergent AEs were mild medication (Nurmikko et al., 2007). Furthermore,
to moderate in severity across both treatment groups. since the evidence base is considered to be poor for
Ten patients (8%) receiving THC/CBD spray experi- medicines currently licensed for the treatment of
enced SAEs, none of which was considered to be evoked neuropathic phenomena, these findings
treatment-related. Six patients (5%) receiving placebo suggest that THC/CBD spray is a promising new candi-
experienced a treatment-emergent SAE, one of which date for treating mixed neuropathic pain characterized
was considered related to treatment. A total of 33 by allodynia (Rowbotham et al., 1998). An additional
patients stopped receiving study medication due to advantage of THC/CBD oromucosal spray is the simple
AEs, 25 in the THC/CBD spray arm and 8 in the handling and fast action of the medicament.
placebo group. No obvious trends were shown for A greater than 30% improvement in pain intensity,
biochemistry, haematology or urinalysis, and no mean considered to signify a clinically meaningful improve-
changes in blood pressure and pulse rate were ment (Rasmussen et al., 2004), was reported by 28%
observed from baseline to final visit. Furthermore, no of patients receiving THC/CBD spray compared with
patients died during the course of this study. 16% of patients taking placebo. This finding was sta-
tistically significant in favour of THC/CBD spray and,
considering the patient population in the study, is
4. Discussion
encouraging. The co-primary analysis of the mean
Neuropathic pain is one of the most difficult types of change from baseline to the end of treatment in PNP
pain to treat (The Committee for Medicinal Products for 0–10 NRS score also showed a treatment difference in
Human Use (CHMP), 2004), with fewer than half of favour of THC/CBD spray, but this did not reach sta-
treated patients receiving meaningful benefit from any tistical significance.
pharmacological drug (Attal et al., 2006). The current The importance of sleep in chronic pain states has
study patients represented an especially resistant treat- been well established (Casarett et al., 2001; Turk and
ment group as they had not responded adequately to Dworkin, 2004), and improved sleep is considered a
existing therapies, had a mean pain 0–10 NRS score of significant treatment objective by patients (Dworkin
4 or above, despite the majority currently taking anal- et al., 2005), especially as neuropathic pain tends to be
gesics for their neuropathic pain, and had a median worse at night (Stacey et al., 2010). Here, we demon-
duration of neuropathic pain of more than 3 years. In strate a statistically significant improvement in sleep
the face of such prolonged neuropathic pain, a new with THC/CBD spray treatment, a finding that sup-

1008 Eur J Pain 18 (2014) 999–1012 © 2014 European Pain Federation - EFIC®
M. Serpell et al. Efficacy of THC/CBD spray in peripheral neuropathic pain

Table 4 Number of patients with at least one all-causality or treatment-related AE with an incidence of 3% or greater by primary system organ class and
preferred term (as medically encoded using the Medical Dictionary for Regulatory Activities [MedDRA] version 8.1).

All-causality Treatment-related

THC/CBD spray Placebo THC/CBD spray Placebo

(n = 128) (n = 118) (n = 128) (n = 118)
System organ class
Preferred term No. of patients (%) No. of patients (%)

Total subjects with at least one AE 109 (85) 83 (70) 97 (76) 56 (47)
Nervous system disorders 79 (62) 34 (29) 73 (57) 20 (17)
Dizziness 52 (41) 12 (10) 50 (39) 11 (9)
Dysgeusia 14 (11) 2 (2) 14 (11) 2 (2)
Headache 13 (10) 9 (8) 8 (6) 7 (6)
Disturbance in attention 8 (6) 2 (2) 8 (6) 1 (1)
Neuropathy peripheral 6 (5) 4 (3) 3 (2) 0
Tremor 6 (5) 0 4 (3) 0
Somnolence 5 (4) 2 (2) 5 (4) 2 (2)
Balance disorder 4 (3) 2 (2) 4 (3) 2 (2)
Memory impairment 4 (3) 2 (2) 4 (3) 2 (2)
Sedation 4 (3) 0 4 (3) 0
Gastrointestinal disorders 60 (47) 43 (36) 48 (38) 30 (25)
Nausea 23 (18) 14 (12) 22 (17) 9 (8)
Vomiting 13 (10) 7 (6) 6 (5) 3 (3)
Diarrhoea 12 (9) 6 (5) 8 (6) 2 (2)
Dry mouth 11 (9) 4 (3) 11 (9) 4 (3)
Abdominal pain upper 6 (5) 1 (1) 4 (3) 0
Dyspepsia 6 (5) 4 (3) 1 (1) 3 (3)
Constipation 4 (3) 2 (2) 2 (2) 0
Mouth ulceration 4 (3) 6 (5) 4 (3) 6 (5)
Oral pain 4 (3) 3 (3) 4 (3) 3 (3)
General disorders and administration site conditions 45 (35) 30 (25) 38 (30) 23 (19)
Fatigue 20 (16) 8 (7) 19 (15) 5 (4)
Feeling drunk 8 (6) 3 (3) 8 (6) 3 (3)
Application site pain 7 (5) 2 (2) 7 (5) 2 (2)
Psychiatric disorders 36 (28) 11 (9) 30 (23) 4 (3)
Dissociation 9 (7) 0 9 (7) 0
Disorientation 8 (6) 0 8 (6) 0
Depression 6 (5) 0 3 (2) 0
Anxiety 4 (3) 1 (1) 3 (2) 1 (1)
Panic attack 4 (3) 1 (1) 3 (2) 0
Infections and infestations 35 (27) 26 (22) 1 (1) 3 (3)
Nasopharyngitis 9 (7) 8 (7) 1 (1) 1 (1)
Gastroenteritis 4 (3) 1 (1) 0 0
Lower Respiratory Tract Infection 4 (3) 3 (3) 0 0
Metabolism and nutrition disorders 15 (12) 6 (5) 10 (8) 5 (4)
Increased appetite 6 (5) 1 (1) 6 (5) 1 (1)
Anorexia 4 (3) 1 (1) 1 (1) 1 (1)
Respiratory, thoracic and mediastinal disorders 15 (12) 16 (14) 7 (5) 5 (4)
Pharyngolaryngeal pain 7 (5) 5 (4) 2 (2) 5 (4)
Dyspnoea 4 (3) 3 (3) 1 (1) 0
Musculoskeletal and connective tissue disorders 11 (9) 8 (7) 2 (2) 1 (1)
Injury, poisoning and procedural complications 9 (7) 6 (5) 2 (2) 0
Skin and subcutaneous tissue disorders 9 (7) 9 (8) 2 (2) 2 (2)
Rash 5 (4) 4 (3) 1 (1) 0
Eye disorders 7 (5) 6 (5) 5 (4) 3 (3)
Ear and labyrinth disorders 6 (5) 1 (1) 5 (4) 1 (1)
Vertigo 5 (4) 0 5 (4) 0
Vascular disorders 4 (3) 5 (4) 3 (2) 2 (2)
Investigations 3 (2) 3 (3) 2 (2) 2 (2)
Neoplasms benign, malignant and unspecified (incl. cysts and polyps) 3 (2) 1 (1) 0 0
Renal and urinary disorders 3 (2) 2 (2) 0 1 (1)
Cardiac disorders 2 (2) 2 (2) 1 (1) 0
Reproductive system and breast disorders 2 (2) 1 (1) 0 0
Immune system disorders 1 (1) 0 0 0
Blood and lymphatic system disorders 0 2 (2) 0 0

AE, adverse effect; CBD, cannabidiol; THC, Δ9-tetrahydrocannabinol.

© 2014 European Pain Federation - EFIC® Eur J Pain 18 (2014) 999–1012 1009
Efficacy of THC/CBD spray in peripheral neuropathic pain M. Serpell et al.

ports the consistent improvements in sleep seen in trial would not be ethical. Moreover, the use of combi-
other clinical studies of this drug (Rog et al., 2005, nation treatments in clinical practice is becoming more
2007; Attal et al., 2006; Nurmikko et al., 2007). This commonplace due to the understanding that multiple
provides further evidence for the efficacy of THC/CBD pain mechanisms contribute to neuropathic pain
spray. Additionally, these improved sleep quality find- (Woolf, 2004; Wade et al., 2010). Adding THC/CBD
ings are also consistent with recent studies with spray to a mixture of pain treatments, which work by
smoked cannabis (Ware et al., 2010) and synthetic different mechanisms, should not impede the activity
THC (Toth et al., 2012). of THC/CBD spray. However, if the other treatments are
Analysis of the SGIC parameter evolution in the providing partial pain relief, this could reduce the mag-
current study demonstrated a statistically significant nitude of benefit derived from THC/CBD spray. The
treatment difference in favour of THC/CBD spray, with patients recruited for this trial were often very resistant
the most pronounced difference observed in the ‘No to pharmacological therapy, so to show a 30%
Change’ category, selected by a relatively high propor- improvement in pain intensity in a proportion of
tion of patients in the placebo group. The SGIC tool is patients was a clinically significant achievement.
considered the ‘gold standard’ measure of patient The self-titration regimen used was chosen for a
outcome in chronic pain trials (Dworkin et al., 2005). number of reasons, including the variable threshold of
Based on this, our findings suggest that overall, individual patients to the pharmacodynamic effects of
patients can achieve important changes in quality of THC/CBD spray (Rog et al., 2005; Attal et al., 2006).
life with THC/CBD spray treatment. Having a self-titration schedule allowed patients to
Interestingly, other cannabinoid trials in which optimize their dose based on their own efficacy and
evoked pain was assessed reported some similar ben- tolerability.
efits to the current study (Svendsen et al., 2004; In terms of safety, THC/CBD spray was well toler-
Abrams et al., 2007; Ware et al., 2010; Toth et al., ated in this study, with low levels of intoxication expe-
2012). Two RCTs that evaluated the effects of smoked rienced, and no evidence of tolerance developing,
cannabis on post-traumatic, post-surgical neuropathic since there was a stable dose pattern following initial
pain (Ware et al., 2010) or HIV-associated sensory pain titration. The most common treatment-emergent,
(Abrams et al., 2007) both demonstrated benefits in treatment-related events were dizziness, nausea,
levels of pain intensity with active treatment. A fatigue and dysgeusia (distortion of sense of taste).
further two trials that investigated different synthetic These AEs have been observed in other clinical studies
forms of THC, dronabinol (Svendsen et al., 2004) and with THC/CBD spray and are recognized as having a
nabilone (Toth et al., 2012) in the treatment of evoked possible causal relationship to the study medication
pain, again demonstrated benefits in levels of pain (Rog et al., 2005; Nurmikko et al., 2007; Wade et al.,
intensity, as well as improvements in the quality of life 2010). The increased incidence of AEs in certain SOCs
and overall patient status, which is similar to the with THC/CBD spray treatment compared with
current study. placebo (i.e., ‘psychiatric disorders’, ‘nervous system
All other secondary endpoints that directly mea- disorders’, ‘gastrointestinal disorders’ and ‘general dis-
sured pain intensity showed improvements from base- orders and administration site conditions’) have also
line to the end of treatment, with treatment been previously reported in other clinical trials with
differences in favour of THC/CBD spray compared THC/CBD spray (Rog et al., 2005; Nurmikko et al.,
with placebo treatment, with only one exception. The 2007; Wade et al., 2010). Psychiatric events such as
punctate allodynia test score was found to improve dissociation and disorientation are known to be
with THC/CBD spray treatment, but the treatment common in clinical trials with THC/CBD spray and are
difference was in favour of placebo. The analysis of representative of a cannabis ‘high’ (Wade, 2012). A
rescue analgesia use also showed a tendency for review of 805 THC/CBD spray patients versus 741
reduced use in the THC/CBD spray treatment group placebo patients found that 4% taking THC/CBD spray
compared with placebo, which could have impacted versus 0.5% taking placebo experienced disorienta-
the pain questionnaire outcomes. tion, while 1.7% taking THC/CBD spray versus 0.1%
Throughout this study, existing analgesia was main- taking placebo experienced dissociation (Wade, 2012).
tained based on ethical and clinical considerations. A While the incidence of these two specific AEs was
variety of treatments for neuropathic pain have dem- higher in this study, this may have been due to
onstrated efficacy and are in widespread use based on the titration regimen adopted. Indeed, a slower
existing guidelines (Attal et al., 2006). To deprive a up-titration administration regimen for THC/CBD
patient of these treatments during a placebo-controlled spray (over a 10-day period) was associated with a

1010 Eur J Pain 18 (2014) 999–1012 © 2014 European Pain Federation - EFIC®
M. Serpell et al. Efficacy of THC/CBD spray in peripheral neuropathic pain

lower number of AEs in later studies (Collin et al., a ‘placebo effect’, which may have diminished the
2010; Novotna et al., 2011). In clinical trials of THC/ positive results seen with THC/CBD spray. While the
CBD spray using a slow up-titration schedule, the treatment difference in favour of THC/CBD spray
incidence of psychiatric AEs is reduced from 15% to increased with increasing daily doses of study medica-
8% compared with the original more aggressive tion, this effect appeared to drop off at a dose of
regimen adopted in this study (Wade, 2012). around 14–15 sprays per day. At a similar dose,
A total of 10 SAEs were experienced by patients however, the proportion of responders in the placebo
receiving THC/CBD spray; however, none was consid- treatment group was still increasing markedly with
ered to be treatment-related. There were no consistent increasing numbers of daily sprays. This suggests that
patterns of difference between THC/CBD spray and patients who took higher mean daily doses of placebo
placebo for haematology, biochemistry and urinalysis perceived a benefit in the subjective pain severity
parameters. Furthermore, changes in vital signs for score. The consequence of this effect is an apparent
pulse rate and systolic blood pressure were unremark- decrease in the true treatment advantage of THC/CBD
able compared with baseline. spray over placebo, observed at lower daily doses.
These findings suggest that future studies would
benefit from a reduction in the current dose ceiling of
4.1 Study limitations
24 sprays per day, thus allowing comparison of the
The presence of a substantial proportion of non- two treatment groups at similar mean doses.
responders in this study suggests that the analysis of
mean changes may not be the most appropriate means 5. Conclusions
of identifying whether the medication has a clinically
useful effect, since the lack of improvement in the In conclusion, this study has shown that in a mean-
non-responders would dilute the improvement seen ingful proportion of otherwise treatment-resistant
in responders. In clinical practice, non-responders to patients, clinically important improvements in their
treatment would be unlikely to remain on a non- pain, sleep quality and global impression of change in
effective drug and would therefore not contribute to the severity of their condition were obtained by taking
understanding the utility of the medicine in the popu- THC/CBD spray. There is also a possibility that these
lation of patients for whom it is suitable. This dilemma results may have been more strongly in favour of
has been discussed by McQuay et al. (2008). THC/CBD spray if the upper dose level had been
Another potential study limitation was the inclusion capped to below 24 sprays daily, and a slower titration
of multiple aetiologies of PNP leading to considerable regimen had been adopted in an attempt to improve
clinical trial heterogeneity. The issue of clinical trial the overall tolerability and its effect of early withdraw-
heterogeneity in patients with neuropathic pain has als and, secondarily, to reduce the placebo response.
been well-documented, and several other controlled Reassuringly, there was no evidence of tolerance
trials of promising new therapeutic candidates have developing and few patients reported experiencing
been negative (Baron et al., 2012). By contrast, a severe AEs. Taken together, these finding are encour-
variety of neuropathic pain studies in heterogeneous aging and suggest that treatment of PNP associated
populations such as the current study have reported with allodynia with THC/CBD spray could bring sig-
positive results in terms of pain scores (Serpell, 2002; nificant benefit to patients.
Rowbotham et al., 2003), including studies in which
vaporized cannabis (Wilsey et al., 2013) and cannabis Author contributions
cigarettes (Wilsey et al., 2008) were used, although
All authors made a substantial contribution to the acquisi-
slightly different pain scales were adopted than those tion and interpretation of the data, critically reviewed the
used in the current study. Several clinical trials and article and approved the final version for publication.
post-hoc analyses have shown greater efficacy of the
study drug when patients are sub-grouped based on References
baseline sensory symptoms and/or pain thresholds
(Edwards et al., 2006; Simpson et al., 2010; Campbell Abrams, D.I., Jay, C.A., Shade, S.B., Vizoso, R.N., Reda, H., Press, S.,
Kelly, M.E., Rowbotham, M.C., Petersen, K.L. (2007). Cannabis in
et al., 2012). As such, future studies that incorporate painful HIV-associated sensory neuropathy. A randomised placebo-
sensory profiling may reveal specific subgroups of controlled trial. Neurology 68, 515–521.
patients in which THC/CBD spray is efficacious. Attal, N., Cruccua, G., Haanpää, M., Hansson, P., Jensen, T.S., Nurmikko,
T., Sampaio, C., Sindrup, S., Wiffen, P., EFNS Task Force. (2006). EFNS
A potential drawback of the maximum dose of 24 guidelines on pharmacological treatment of neuropathic pain. Eur J
daily sprays adopted in this study was the potential for Neurol 13(11), 1153–1169.

© 2014 European Pain Federation - EFIC® Eur J Pain 18 (2014) 999–1012 1011
Efficacy of THC/CBD spray in peripheral neuropathic pain M. Serpell et al.

Backonja, M., Serra, J. (2004). Pharmacologic management part 1: Better Rice, A.S., Maton, S.; Postherpetic Neuralgia Study Group. (2001). Gaba-
studied neuropathic pain diseases. Pain Med 5, S28–S47. pentin in post herpetic neuralgia: A randomised, double blind, placebo
Baron, R., Förster, M., Binder, A. (2012). Subgrouping of patients with controlled study. Pain 94(2), 215–224.
neuropathic pain according to pain-related sensory abnormalities: A first Rog, D.J., Nurmikko, T.J., Friede, T., Young, C.A. (2005). Randomized
step to a stratified treatment approach. Lancet Neurol 11(11), 999–1005. controlled trial of cannabis based medicine in central pain due to mul-
Berman, J.S., Symonds, C., Birch, R. (2004). Efficacy of two cannabis tiple sclerosis. Neurology 65, 812–819.
based medicinal extracts for relief of central neuropathic pain from Rog, D.J., Nurmikko, T.J., Young, C.A. (2007). Oromucosal delta-9-
brachial plexus avulsion: Results of a randomized controlled trial. Pain tetrahydrocannabinol/cannabidiol for neuropathic pain associated with
112, 299–306. multiple sclerosis: An uncontrolled, open-label, 2-year extension trial.
Bouhassira, D., Lanteri-Minet, M., Attal, N., Laurent, B., Touboul, C. Clin Ther 29(9), 2068–2079.
(2008). Prevalence of chronic pain with neuropathic characteristics in Rowbotham, M., Harden, N., Stacey, B., Bernstein, P., Magnus-Miller, L.
the general population. Pain 136, 380–387. (1998). Gabapentin for the treatment of post herpetic neuralgia. JAMA
Boureau, F., Legallicier, P., Kabir-Ahmadi, M. (2003). Tramadol in post- 280, 1837–1842.
herpetic neuralgia: A randomised, double-blind, placebo-controlled Rowbotham, M.C., Twilling, L., Davies, P.S., Reisner, L., Taylor, K., Mohr,
trial. Pain 104, 323–331. D. (2003). Oral opioid therapy for chronic peripheral and central neu-
Campbell, C.M., Kipnes, M.S., Stouch, B.C., Brady, K.L., Kelly, M., ropathic pain. N Engl J Med 348(13), 1223–1232.
Schmidt, W.K., Petersen, K.L., Rowbotham, M.C., Campbell, J.N. Russo, E.B. (2011). Taming THC: Potential cannabis synergy and
(2012). Randomized control trial of topical clonidine for treatment of phytocannabinoid-terpenoid entourage effects. Br J Pharmacol 163(7),
painful diabetic neuropathy. Pain 153(9), 1815–1823. 1344–1364.
Casarett, D., Karlawish, J., Sankar, P., Hirschman, K., Asch, D.A. (2001). Serpell, M.G., Neuropathic Pain Study Group. (2002). Gabapentin in
Designing pain research from the patients’ perspective: What trial end- neuropathic pain syndromes: A randomised, double-blind, placebo-
points are important to patients with chronic pain? Pain Med 2, 309–316. controlled trial. Pain 99, 557–566.
Cleeland, C.S., Ryan, K.M. (1994). Pain assessment: Global use of the Simpson, D.M., Schifitto, G., Clifford, D.B., Murphy, T.K., Durso-De Cruz,
Brief Pain Inventory. Ann Acad Med Singapore 23(2), 129–138. E., Glue, P., Whalen, E., Emir, B., Scott, G.N., Freeman, R., 1066 HIV
Collin, C., Ehler, E., Waberzinek, G., Alsindi, Z., Davies, P., Powell, K., Neuropathy Study Group. (2010). Pregabalin for painful HIV neuropa-
Notcutt, W., O’Leary, C., Ratcliffe, S., Nováková, I., Zapletalova, O., thy: A randomized, double-blind, placebo-controlled trial. Neurology
Piková, J., Ambler, Z. (2010). A double-blind, randomized, placebo- 74(5), 413–420.
controlled, parallel-group study of Sativex, in subjects with symptoms Stacey, B.R., Dacosta DiBonaventura, M., Martin, S., Bell, C.F. (2010).
of spasticity due to multiple sclerosis. Neurol Res 32(5), 451–459. Chronological Characteristics of Painful Diabetic Peripheral Neuropa-
Dworkin, R.H., Turk, D.C., Farrar, J.T., Haythornewaite, J.A., Jensen, M.P., thy. American Pain Society ASM, Abstract 23.
Katz, N.P., Kerns, R.D., Stucki, G., Allen, R.R., Bellamy, N., Carr, D.B., Svendsen, K.B., Jensen, T.S., Bach, F.W. (2004). Does the cannabinoid
Chandler, J., Cowan, P., Dionne, R., Galer, B.S., Hertz, S., Jadad, A.R., dronabinol reduce central pain in multiple sclerosis? Randomised
Kramer, L.D., Manning, D.C., Martin, S., McCormick, C.G., McDermott, double blind placebo controlled crossover trial. BMJ 329, 253–260.
M.P., McGrath, P., Quessy, S., Rappaport, B.A., Robbins, W., Robinson, The Committee For Medicinal Products For Human Use (CHMP). Guide-
J.P., Rothman, M., Royal, M.A., Simon, L., Stauffer, J.W., Stein, W., line on Clinical Investigation of Medicinal Products Intended for the
Tollett, J., Wernicke, J., Witter, J. (2005). Core outcome measures for Treatment Of Neuropathic Pain London, 18 November 2004 CHMP/
chronic pain clinical trials: IMMPACT recommendations. Pain 113, 9–19. EWP/252/03.
Edwards, R.R., Haythornthwaite, J.A., Tella, P., Max, M.B., Raja, S. The Euroqol Group. (1990). Euroqol – a new facility for the measurement
(2006). Basal heat pain thresholds predict opioid analgesia in patients of health-related quality of life. Health Policy 16, 199–208.
with postherpetic neuralgia. Anesthesiology 104(6), 1243–1248. Torrance, N., Smith, B.H., Bennett, M.I., Lee, A.J. (2006). The epidemi-
Howlett, A.C., Barth, F., Bonner, T.I., Cabral, G., Casellas, P., Devane, W.A., ology of chronic pain of predominantly neuropathic origin. Results
Felder, C.C., Herkenham, M., Mackie, K., Martin, B.R., Mechoulam, R., from a general population survey. Pain 7, 281–289.
Pertwee, R.G. (2002). International Union of Pharmacology. XXVII. Toth, C., Mawani, S., Brady, S., Chan, C., Liu, C., Mehina, E., Garven, A.,
Classification of cannabinoid receptors. Pharmacol Rev 54(2), 161–202. Bestard, J., Korngut, L. (2012). An enriched-enrolment, randomized
Jensen, T.S., Gottrup, H., Sindrup, S.H., Bach, F.W. (2001). The clinical withdrawal, flexible-dose, double-blind, placebo-controlled, parallel
picture of neuropathic pain. Eur J Pharmacol 429, 1–11. assignment efficacy study of nabilone as adjuvant in the treatment of
Karst, M., Salim, K., Burstein, S., Conrad, I., Hoy, L., Schneider, U. (2003). diabetic peripheral neuropathic pain. Pain 153(10), 2073–2082.
Analgesic effect of the synthetic cannabinoid CT3 on chronic neuro- Turk, D.C., Dworkin, R.H. (2004). What should be the core outcomes in
pathic pain. A randomized controlled trial. JAMA 290(13), 1757–1762. chronic pain clinical trials? Arthritis Res Ther 6, 151–154.
McQuay, H.J., Derry, S., Moore, R.A., Poulain, P., Legout, V. (2008). Wade, D. (2012). Evaluation of the safety and tolerability profile of
Enriched enrolment with randomised withdrawal (EERW): Time for a Sativex: Is it reassuring enough? Expert Rev Neurother 12(4 Suppl), 9–14.
new look at clinical trial design in chronic pain. Pain 135, 217–220. Wade, D.T., Collin, C., Stott, C., Duncombe, P. (2010). Meta-analysis of
MHRA Public Assessment Report. (2010). Nabiximols Oromucosal Spray the efficacy and safety of Sativex (nabiximols) on spasticity in people
(delta-9-tetrahydrocannabinol and cannabidiol) – PL 18024/0009. with multiple sclerosis. Mult Scler 16, 707–714.
UK/H/2462/001/DC. http://www.mhra.gov.uk/home/groups/par/ Ware, M.A., Wang, T., Shapiro, S., Robinson, A., Ducruet, T., Huynh, T.,
documents/websiteresources/con084961.pdf (accessed January 2012). Gamsa, A., Bennett, G.J., Collet, J.P. (2010). Smoked cannabis for
Novotna, A., Mares, J., Ratcliffe, S., Novakova, I., Vachova, M., chronic neuropathic pain: A randomized controlled trial. CMAJ
Zapletalova, O., Gasperini, C., Pozzilli, C., Cefaro, L., Comi, G., Rossi, P., 182(14), E694–E701.
Ambler, Z., Stelmasiak, Z., Erdmann, A., Montalban, X., Klimek, A., Wilsey, B., Marcotte, T., Deutsch, R., Gouaux, B., Sakai, S., Donaghe, H.
Davies, P., Sativex Spasticity Study Group. (2011). A randomized, (2013). Low-dose vaporized cannabis significantly improves neuro-
double-blind, placebo-controlled, parallel-group, enriched-design study pathic pain. J Pain 14(2), 136–148.
of nabiximols (Sativex®), as add-on therapy, in subjects with refractory Wilsey, B., Marcotte, T., Tsodikov, A., Millman, J., Bentley, H., Gouaux,
spasticity caused by multiple sclerosis. Eur J Neurol 18, 1122–1131. B., Fishman, S. (2008). A randomized, placebo-controlled, crossover
Nurmikko, T.J., Serpell, M.G., Hoggart, B., Toomey, P.J., Morlion, B.J., trial of cannabis cigarettes in neuropathic pain. J Pain 9(6), 506–
Haines, D. (2007). Sativex successfully treats neuropathic pain charac- 521.
terised by allodynia: A randomised, double-blind, placebo-controlled Woolf, C.J. (2004). Dissecting out mechanisms responsible for peripheral
clinical trial. Pain 133, 210–220. neuropathic pain: Implications for diagnosis and therapy. Life Sci 74,
Rasmussen, P.V., Sindrup, S.H., Jensen, T.S., Bach, F.W. (2004). Thera- 2605–2610.
peutic outcome in neuropathic pain: Relationship to evidence of Woolf, C.J., Max, B.M. (2001). Mechanism-based pain diagnosis. Anaes-
nervous system lesion. Eur J Neurol 11, 545–553. thesiology 95, 241–249.

1012 Eur J Pain 18 (2014) 999–1012 © 2014 European Pain Federation - EFIC®