Reports from animal studies provide strong evidence to support the analgesic effects of CBs; however, studies in human volunteers and patients are at a much earlier stage and the evidence is, at present, weak. Nevertheless, patient surveys report the use of cannabis for pain relief: in a recent survey of 2,969 UK patients, medicinal use of cannabis is reported by those with chronic pain (25%) multiple sclerosis (22%), arthritis (21%) and neuropathy (19%) (Ware et al. 2005). Of a sample of 209 Canadian chronic non-cancer pain sufferers,
15% reported having used cannabis at least once for the control of their pain, with approximately 38% using cannabis at least daily (Ware et al. 2003). There is also evidence that a signiﬁcant proportion of people living with human immunodeﬁciency virus (HIV) in London use cannabis for symptom control (Woolridge et al. 2005).
Evidence from Volunteer Studies
There is a relatively small literature on the effects of CBs in human vol- unteer models of pain. Application of HU210 to human skin has been re- ported to have inhibitory effects on histamine-evoked itch (Dvorak et al.
2003) and capsaicin-evoked pain responses (Rukwied et al. 2003). In the latter study, the rate of increase in pain intensity following capsaicin appli-
cation was lower in the HU210-treated compared to the placebo (ethanol)- treated group, and a similar temporary slowing in the development phase
of capsaicin-associated mechanical allodynia was reported. However, there was no overall difference in pain intensity. Anti-nociceptive properties of smoked marijuana have also been reported in a noxious thermal stimulus withdrawal test (Greenwald and Stitzer 2000). In contrast, a study on 12
volunteers reported that a single does of oral THC (20 mg) did not have signiﬁcant anti-nociceptive effects on responses to thermal, mechanical or
electrical stimuli. Interestingly, 30 mg of morphine was effective in most of the tests, and some analgesic effects were reported when THC was ad- ministered with 30 mg of morphine before electrical stimulation (Naef et al. 2003).
Evidence from Randomised Controlled Clinical Trials
The data from clinical trials to date indicate that the efﬁcacy of currently available CBs in humans is modest and that their effectiveness is hampered by an unfavourable therapeutic index.
A qualitative systematic review of trials published up to 1999 identiﬁed nine clinical trials of CBs of sufﬁcient quality for inclusion in the analysis (Campbell et al. 2001). Five of these trials used cancer pain as a model, two used chronic non-malignant pain and two acute pain. Most of the trials examined
the effects of either Δ9THC or levonantradol. The analgesic effect of these CBs was estimated to be approximately equi-analgesic to codeine 50–120 mg, but adverse effects were common, being reported in all studies. One study showed
the adverse effects to be dose-related and it is possible that these obfuscated a greater analgesic effect at the higher dose of Δ9THC which was examined (Noyes et al. 1975).
Since 1999, several clinical trials have been published in which the anal-
gesic effects of CBs in chronic, especially neuropathic, pain conditions were investigated. One trial examined the efﬁcacy of CBs in alleviating neuropathic symptoms in multiple sclerosis (Wade et al. 2003). In this study, a self-titration
regimen was used in which either plant-derived Δ9THC, cannabidiol, a 1:1 mix- ture of Δ9THC and cannabidiol, or placebo were administered by sublingual spray. Data on a range of symptoms were collected and modest analgesic effects
were evident. Whilst the ﬁxed ratio preparation (mean dose 22 mg/day) was not associated with a signiﬁcant reduction in baseline pain intensity scores, Δ9THC (23.5 mg/day) and cannabidiol (22 mg/day) were, when compared to the effect of placebo. Adverse effects were common; 30%–67% of subjects
reporting more than one adverse event, and 17% of patients withdrew from the trial because of adverse effects. A cross-over trial of 24 multiple sclero- sis patients with central pain investigated the effects of a 3-week course of
daily treatment with oral dronabinol (maximum 10 mg dose) (Svendsen et al.
2004). The authors report a modest but signiﬁcant analgesic effect of the active treatment on median spontaneous pain intensity scores, calculating a number needed to treat (NNT) value for 50% pain relief at 3.5. A study of 48 patients with central neuropathic pain resulting from brachial plexus avulsion injury reported that short-term treatment with oral cannabis extract (as an adjunct to existing medication) revealed no improvement in the primary pain efﬁcacy measure (Berman et al. 2004). Furthermore, a complicated trial of 21 patients with chronic neuropathic pain, in which patients received two daily doses of
ajulemic acid (the synthetically modiﬁed metabolite of THC) for a week, did reveal evidence of analgesic efﬁcacy for this compound (Karst et al. 2003). No signiﬁcant effects of the drug on mechanical hypersensitivity were reported in this trial; however, there was evidence of an analgesic effect. Pain relief from conventional CBs was also measured as a secondary outcome in a very large study in multiple sclerosis patients (Zajicek et al. 2003): although there was no effect of CBs on the primary efﬁcacy measures, a modest analgesic effect was reported.
In addition to efﬁcacy, adverse effects are important in determining the clin- ical effectiveness of any novel therapy, and an acceptable therapeutic index for short-term adverse events will have to be proved for CBs. Furthermore, concerns relating to the long-term risk of developing mental illness in regular cannabis users have obvious relevance for the patients electing to undergo long-term, regular medication with CBs in conditions such as chronic pain. For example, a 27-year cohort study of 50,000 Swedish military conscripts found a dose-dependant increased risk of schizophrenia (once confounding factors such as cannabis use in prodromal schizophrenia and concomitant drug abuse had been excluded) in regular cannabis users (Zammit et al. 2002). Similar ﬁndings for depression and anxiety (Patton et al. 2002) and psychosis (Arseneault et al. 2002; D’Souza et al. 2004; Henquet et al. 2005) have also been reported in smaller cohort studies. There is also evidence that the risk of cannabis-induced psychosis is substantially greater in those individuals who have a predisposition to developing psychosis (Henquet et al. 2005) or genetic functional polymorphisms (Caspi et al. 2005). There are also data indicating cumulative, dose-dependent deﬁcits in cognitive function in regular cannabis users (Solowij et al. 2002).
There are compelling laboratory data supporting the analgesic effects of CBs; however, before CB-based drugs can be used therapeutically in humans, they must be shown to be both effective and safe in long-term regular use. Well- designed clinical trials are therefore required, but are perhaps premature until suitable CBs with a satisfactory therapeutic index for analgesia and proven bioavailability when administered by a practical route of administration are available for human study. Current strategies to circumvent side-effects but retain analgesic efﬁcacy include the development of non-psychoactive or pe- ripherally active CBs, such as cannabidiol, ajulemic acid and CB2 receptor
selective agonists (Fride et al. 2004; Salim et al. 2005; Ibrahim et al. 2005). Whilst new drugs are being trialled, the prospective clinical uses for CBs could lie with chronic pain patients in whom fear contributes to abnormal pain be- haviour (Marsicano et al. 2002) or in situations where the adverse effects of CBs might confer additional beneﬁt over existing therapies, for example by exploiting their anti-inﬂammatory or anti-emetic properties.
Acknowledgements The authors receive support from the Wellcome Trust.