EFNS guidelines on neurostimulation therapy for neuropathic pain

Neurostimulation therapy for neuropathic pain 953are in progress in other movement and psychiatricdisorders, epilepsy and migraine. The neurostimulationtechniques proposed for treating pain are: transcutaneouselectrical nerve stimulation (TENS), peripheralnerve stimulation (PNS), nerve root stimulation (NRS),spinal cord stimulation (SCS), deep brain stimulation(DBS), epidural motor cortex stimulation (MCS), andrepetitive transcranial magnetic stimulation (rTMS).These techniques vary greatly in their degree of invasiveness,stimulated structures and rationale, but theyare all adjustable and reversible.Our Task Force aimed at providing the neurologistwith evidence-based recommendations that may help todetermine when a patient with neuropathic pain shouldtry a neurostimulation procedure. To provide a betterunderstanding, the results are preceded by a descriptionof the procedure and its supposed rationale.Search methodsTask Force participants were divided into subgroupsand assigned the search for specific neurostimulationprocedures, with two persons carrying out an independentsearch for each procedure. A two-stage approachto the relevant literature search wasundertaken. First the MEDLINE, EMBASE andCochrane databases were searched for systematic reviews,from inception date to May 2006. Detailedsearches are listed in Appendix 1 (SupplementaryMaterial). Recent textbooks known to the authorswere also examined for relevant references. These reviewsand books were used to identify the primaryliterature. Secondly, given the search cut off dates ofprevious systematic reviews, an update search forprimary studies (randomized controlled trials, nonrandomizedcontrolled trials, observational comparativestudies and case series) was undertaken. Studiesidentified by this updated search were added to thebody of evidence for each neurostimulation procedureunder each indication heading.All study designs were included except case reportsand very small case series (

Neurostimulation therapy for neuropathic pain 955Table 1 Summary of efficacy and safety of peripheral stimulations (TENS, PNS, and NRS)Technique/condition Available evidenceNo.patients Summary of efficacyTENS/chronicpainTENS/NePPainful diabeticneuropathyOne meta-analysis [10],analysing 38 RCTs;only two onneuropathicpain: Thorsteinsson1977 and Rutgers1988 [16,17]– There is no evidence; butit is clear that effectincreases with dose(duration of the session· frequency ofsessions · total duration)Reichstein 2005 [11] 25 TENS compared with HFmuscle: 25% success withTENS and 69% with HFDiabeticneuropathyPainful diabeticneuropathyDiabeticneuropathyTraumaticneuropathyForst 2004 [14] 19 LF TENS reduced VAS by23%, significant differencefrom placeboHamza 2000 [13] 50 Painful PENS reduced VASby 60%, significantdifference from sham, andimproved QoLKumar 1997 [12] 35 LF TENS with biphasicstimuli exponentiallydecaying was significantlybetter than sham inreducing neuropathicsymptomsCheing 2005 [15] 19 Significantly better thanplaceboRadiculopathy Bloodworth 2004 [18] 11 Random TENS and TENSon the cervical back weresignificantly better thanplacebo, with R-TENSbetter than TENSMixed pains CT Tulgar 1991(internal control)8 Two did not get sufficientpain relief; one receivedprolongedpain relief; three wentbetter with ÔburstÕ, onewith high-rate and onewith low-ratemodulated TENSSummary ofharms Comparator Blind Random<strong>EFNS</strong>class CommentsPracticallynothingVarious Yes Yes IPracticallynothingPracticallynothingHigh-frequencymuscle stimulationYes Yes II Reichstein [11] has noplacebo, TENS goesfar worse than HFmuscle stimulationPlacebo Yes Yes IIIPracticallynothingSham Yes III Crossover with aninadequate comparator(needles with no current)PracticallynothingSham Yes III Sham inadequate andreport of improvementof allsymptoms.PracticallynothingPracticallynothingPlacebo Yes Yes IIPlacebo andRandom-TENSYes Yes III Few patients andcrossoverPracticallynothingFour modes ofTENSstimulationYes No IV Few patients who chosewhich mode of TENSthey preferredÓ 2007 <strong>EFNS</strong> European Journal of Neurology 14, 952–970

956 G. Cruccu et al.Table 1 (Continuted)Technique/condition Available evidenceNo.patients Summary of efficacySummary ofharms Comparator Blind Random<strong>EFNS</strong>class CommentsPHN RCT Rutgers 1988[17]PeripheralneuropathyRCT Thorsteinsson1977 [16]few Practicallynothing24 Significantly better thanplaceboPracticallynothingAcupuncture See McQuayet al. 1997 [10]Placebo Yes Yes IIIPNS/NeP No meta-analyses, noRCTsCRPS II Buschmann 1999 52 Successful in 47Peripheral (n) Nashold 1982 35 Successful in 15Post-traumatic Law 1980 22 Successful in 13Mixed/various Picaza 1977 37 Successful in 18Mixed/various Campbell 1976 33 Successful in eightPeripheral (n)radiculopathy,amputationPicaza 1975 23 Successful in 20Totals 202 Successful in 121 (60%)Sometimesneed forreoperationNone No No Class IV notenoughevidencefor anyrecommendationVery few and oldpapers, thistechnique does notseem to be gettingpopularNRS/NeP No meta-analyses,no RCTsNeuropathicpelvic painInterstitialcystitisCT Everaert 2001 26 Successful in 16Whitmore 2003,no control33 Significant improvementSometimesneed forreoperationNone No No Class IV notenoughevidencefor anyrecommendationTENS, transcutaneous electrical nerve stimulation; LF TENS, low-frequency, high-intensity TENS, also called acupuncture-like; PNS, peripheral nerve stimulation with implanted electrodes; NRS,nerve root stimulation with implanted electrodes; QoL, quality of life; CT, controlled trial; HF, high frequency; PENS, percutaneous electrical nerve stimulation.Ó 2007 <strong>EFNS</strong> European Journal of Neurology 14, 952–970

Neurostimulation therapy for neuropathic pain 959Table 2 (Continued)IndicationVolume of evidence[no. trials (no.patients)]Diabetic neuropathy One prospective caseseries [n ¼ 8]Tesfaye et al. 1996;Daousi et al. 2004One retrospectivemixed case series[n ¼ 14]Kumar et al. 2006Other peripheral neuropathy One retrospectivemixed case series[n ¼ 23]Kim et al. 2001Post-herpetic neuralgia Four retrospectivecase series (3 mixed)[10; 28; 8; 4 (50)]Kumar et al. 2006;Harke et al. 2002;Meglio et al. 1989;Sanchez-Ledesmaet al. 1989Intercostal neuralgia No evidence foundrelating to thisspecific diagnosisBrachial plexus damage/avulsion Two retrospectivecase series (2 mixed)[8; 8 (16)]Simpson et al. 2003Hood; Siegfried 1984<strong>EFNS</strong>class Summary of efficacy Summary of harms<strong>EFNS</strong>grade CommentsIVIVPain relief>50% pain relief in6/8 at 14 months(6/7: one died at2 months)>50% relief in 5/6 at3 years (backgroundand peak pain)>50% relief in 4/4 at7 years (backgroundpain)>50% relief in 3/4 at7 years (peak pain)Exercise toleranceIncreased by 150%in 6/6Pain relief>50% relief in 12/14Ôlong-termÕIV Pain relief>50% relief in 10/23(43.5%) at 1 yearLead migration in 2/8Superficial infection in 2/8Skin reaction in 1/8Not disaggregated DD Prospective VAS + McGillSep. pain elementsPreservation of large fibre(vibration and jointposition) function essentialFive outcome measuresThird party assessorFollow-up unclearIV Pain reliefSignificant long-termin 38/50 (pooled)Medication stopped in21/31Opioids stopped in18/19Lead fracture in oneReceiver failure in oneLeads replaced in three toimprove coverageD Success varies between seriesdue to variabledeafferentationIV Pain reliefSignificant relief in8/16Nil, or not disaggregated D Different scoring methodsEvidence of full avulsion(cf damage) of specificrelevant nerve roots notalways givenÓ 2007 <strong>EFNS</strong> European Journal of Neurology 14, 952–970

960 G. Cruccu et al.Table 2 (Continued)IndicationVolume of evidence[no. trials (no.patients)]<strong>EFNS</strong>class Summary of efficacy Summary of harms<strong>EFNS</strong>grade CommentsAmputation pain(phantom and stump)Three retrospectivecase series[25; 9; 61 (95)]Lazorthes et al. 1995;Simpson 1991Krainick + Thoden1989; Krainick et al.1975Facial pain (trigeminopathic) Insufficient evidenceCentral pain ofFive retrospective casespinal cord originseries (4 mixed) [19;11; 101; 9; 35 (175)]Kumar et al. 2006;Barolat et al. 1998;Lazorthes et al.1995; Cioni et al.1995; Meglio et al.1989; Tasker et al.1992Central pain of brain origin Two retrospectivecase series (1 mixed)[45; 10 (55)]Katayama et al. 2001;Simpson 1991 [39]IV Phantom painSignificant relief in 7/14Stump painSignificant relief in 5/9Mixed – stump/phantom notspecifiedKrainick’s series: 56% of 61 had>50% relief (early), droppingto 43% (late). Reduced drugintake correlatedLazorthes: 60% of 25 good orexcellent long-term (‡2 years)IV Pain reliefa) cord injury: 15/62 significantlong term pain relief overallincomplete: 11/33 significantreliefcomplete: 0/11 significant reliefb) MS: long-term pain relief onfive outcome measures in 15/19(Kumar)bowel/sphincter functionimproved in 16/28Gait improved in 15/19(no details)c) mixed incl trauma, tumoursurgery, viral etc: 34% good/excellent at ‡2 years (Lazorthes;n ¼ 101). Pain relief, analgesicdrug intake and activityIV Pain reliefSignificant in 6/55>60% reduction in VAS in 3/45Infection 1.6%Surgical revisions 31%Not stated/not disaggregatedin four studiesAseptic meningitis 1/9Superficial infection 1/9Electrode dislodgement 1/9Not stated/not disaggregated DD Phantom and stump painsnot always distinguishedD Completeness of lesion notalways statedMuch greater success whereclinically incomplete lesionSuccess correlates withsensory status: the lesssensory deficit the better theresultsCMM, conventional medical management; ODI, Oswestry Disability Index; SIP, sickness impact profile; VAS, visual analogue scale; HRQoL, health-related quality of life; NRS, numerical rating scale;MS, multiple sclerosis.Ó 2007 <strong>EFNS</strong> European Journal of Neurology 14, 952–970

Neurostimulation therapy for neuropathic pain 961appropriate stimulation cannot be achieved. However,this testing is not a guarantee of long-term success inneuropathic pain.Evidence identifiedWe identified a number of systematic reviews andmeta-analyses [20–22] and a few narrative but detailedreviews [23–25]. The majority of systematic reviews, aswell as primary studies, to date have focused onpatients with failed back surgery syndrome (FBSS) orcomplex regional pain syndrome (CRPS). ConcerningFBSS there are two class-II RCTs, the first showingthat SCS is more effective than reoperation [26] and thesecond that its addition is more effective than conventionalmedical care alone [27,28]. In these trials theresponders (pain relief >50%) to SCS were 47–48% vs.9–12% with comparator, at 6–24 months. In the pooleddata from case series in 3307 FBSS patients, the proportionof responders was 62%. In CRPS type I, resultsand evidence level are also good, with a single class-IIRCT of SCS compared with conventional care alone[29,30]. In this RCT, SCS reduced the visual analoguescale score by a mean 2.6 cm more than comparator at6 months and by 1.7 cm at 5 years. In the pooled datafrom case series (n ¼ 561) in CRPS I and II, the proportionof responders was 67%. Both RCTs and caseseries have also found significant improvement infunctional capacity and quality of life. In a pooledsafety analysis of SCS across all indications, the undesiredevents were mostly dysfunction in the stimulatingapparatus: lead migration (13.2%), lead breakage(9.1%), and other minor hardware problems [20]. Alsothe medical complications were minor and never lifethreatening and were usually solved, like the hardwareproblems, by removing the device. The overall infectionrate was 3.4%.The effect of SCS has also been studied in many otherconditions. We found positive case series evidence forCRPS II, peripheral nerve injury, diabetic neuropathy,PHN, brachial plexus damage, amputation (stump andphantom pains) and partial spinal cord injury, andnegative evidence for central pain of brain origin, nerveroot avulsion and complete spinal cord transection.However, all reports are class IV, thus preventing anyfirm conclusion. The efficacy and safety outcomes ofSCS are detailed by indication in Table 2.RecommendationsWe found level B evidence for the effectiveness of SCSin FBSS and CRPS I. The available evidence is alsopositive for CRPS II, peripheral nerve injury, diabeticneuropathy, PHN, brachial plexus lesion, amputation(stump and phantom pains) and partial spinal cordinjury, but still requires confirmatory comparative trialsbefore the use of SCS can be unreservedlyrecommended in these conditions.Deep brain stimulationDeep brain stimulation for the treatment of medicallyrefractory chronic pain preceded the gate theory [31].Deep brain targets in current use include the sensory(ventral posterior) thalamus and periventricular graymatter (PVG) contralateral to the pain if unilateral, orbilaterally if indicated. Both sites have been targets ofanalgesic DBS for three decades [32,33]. After accuratetarget localization using MRI, stereotactic computerizedtomography and brain atlas co-registration asappropriate, an electrode is stereotactically inserted intosubcortical cerebrum under local anesthesia. The electrodesare connected to a subcutaneous IPG, placed inthe chest or abdomen.The mechanisms by which DBS relieves pain remainunclear. Animal experiments have shown that thalamicstimulation suppressed deafferentation pain, mostprobably via thalamo-corticofugal descending pathways.Autonomic effects of PVG stimulation are underinvestigation, a positive correlation between analgesicefficacy and magnitude of blood pressure reductionhave been demonstrated in humans [34]. It is currentlybelieved that stimulation of ventral PVG engages nonopioiddependent analgesia commensurate with passivecoping behaviour whereas stimulation of dorsal PVGinvolves opioid-related Ôfight or flightÕ analgesia withassociated autonomic effects [34]. The effect of frequency,lower frequencies (5–50 Hz) being analgesicand higher frequencies (>70 Hz) pain-provoking,suggests a dynamic model whereby synchronousoscillations modulate pain perception.As with any implanted technique of neurostimulationfor treating pain, patient selection is a major challenge.Trial stimulation via externalized leads can identifythose in whom DBS is not efficacious or poorly tolerated[35,36]. However, successful trial stimulation hasnot resulted in long-term success for up to half of cases.Contraindications include psychiatric illness, uncorrectablecoagulopathy, and ventriculomegaly precludingdirect electrode passage to the surgical target [37].Evidence identifiedWe identified several reviews and one meta-analysis[37], which conclude that DBS is more effective fornociceptive pain than for neuropathic pain (63% vs.47% long-term success). In patients with neuropathicpain, moderately higher rates of success were seen inpatients with peripheral lesions (phantom limb pain,radiculopathies, plexopathies and neuropathies) [37].We identified a number of primary studies, for 623Ó 2007 <strong>EFNS</strong> European Journal of Neurology 14, 952–970

Neurostimulation therapy for neuropathic pain 963Table 5 Summary of efficacy and safety of deep brain stimulation by indication from other, older studies (after Bittar et al. 2005) [37]IndicationVolume ofevidence(no. patients)Success on initialstimulationSuccess on chronicstimulationLong-termpercentage successAmputation pain9 7 4 44(phantom and stump)Post-stroke pain 45 24 14 31FBSS 59 54 46 78Peripheral nerve injury 44 36 31 70Post-herpetic neuralgia 11 6 4 36Intercostal neuralgia 4 3 1 25Brachial plexus damage/avulsion 12 9 6 50Malignancy pain 23 19 15 65Facial pain (trigeminopathic) 32 21 12 38Central pain of spinal cord origin 47 28 20 43Other 35 28 22 63Table 6 Summary of efficacy and anatomical targets from other, olderstudies (after Bittar et al. 2005 [5])Anatomical siteof DBSVolume ofevidence no.patientsNumbersuccessfullong-termPVG 148 117 79PVG and55 48 87ST or ICST 100 58 58ST or IC 16 6 38PercentagesuccessPVG, periventricular gray matter; ST, sensory thalamus; IC, internalcapsule.concomitant to MRI-guided ÔneuronavigationÕ. Intraoperativecortical stimulation with clinical assessmentor EMG recordings can help to determine the positionof the electrodes. One or two quadripolar electrodes areimplanted over the motor representation of the painfularea, either parallel or orthogonal to the central sulcus.The electrode is connected to a subcutaneous IPG. Thestimulation parameters are optimized post-operatively,keeping the intensity below motor threshold, and thestimulation is usually set on cyclic mode (alternatingÔonÕ and ÔoffÕ periods).The mechanism of action of MCS remains hypothetical.Tsubokawa et al. [40] showed that MCSattenuated abnormal thalamic hyperactivity afterspinothalamic transection in cats, and considered thatsuch effect involved retrograde activation of somatosensorycortex by cortico-cortical axons [41]. However,positron-emission tomography and SEPs failed to showany significant activation of sensory-motor cortex duringMCS, whilst a strong focal activation was observedin thalamus, insula, cingulate-orbitofrontal junctionand brainstem [42,43], suggesting that MCS-inducedpain relief may relate to (i) top-down activation ofdescending pain control systems going from motorcortex to thalamus, and perhaps to motor brainstemnuclei and (ii) blunting of affective reactions to pain viaactivation of orbitofrontal-perigenual cingulate cortex[43]. Both hypotheses have received recent support fromstudies in animals and in humans [44–46]. The fact thatmany of the regions activated by MCS contain highlevels of opioid receptors suggests that long-lastingMCS effects may also involve secretion of endogenousopioids.Eligible patients should be resistant or intolerant tomain drugs used for neuropathic pain [1,2]. Somestudies include pre-operative sessions of transcranialmagnetic stimulation, which is regarded predictive ofthe MCS outcome (see Repetitive transcranial magneticstimulation). Candidates to MCS have sometimesexperienced failure of other neurosurgical procedures,such as radicellectomy (DREZ-lesion), anterolateralcordotomy, trigeminal nerve surgery or SCS.Evidence identifiedOur search disclosed no systematic review or metaanalysis,but found a relatively large number of studies(mostly case series) on CPSP and facial neuropathicpain. In CPSP, we extracted 143 non-overlapping patientsfrom 20 case series: the average proportion ofsuccess was about 50%. Slightly better results (60% ofresponders, based on 60 patients from eight series)were obtained in facial neuropathic pain, central orperipheral. Most of these case series were class IV.Two studies can be classified as class III, because theyhad a comparator (results of other treatments, surgicalor pharmacological), and outcome assessment andtreatment were dissociated: Katayama et al. [39]. hada 48% success rate in patients with CPSP and Nutiet al. [4]. A 52% success rate in 31 patients with variousneuropathic pain conditions, mostly CPSP. Oneof these papers provided follow-up results up toÓ 2007 <strong>EFNS</strong> European Journal of Neurology 14, 952–970

964 G. Cruccu et al.4 years [4]. In phantom pain, brachial plexus or nervetrunk lesion, spinal cord lesions or CRPS, we onlyfound case reports. Most common undesired eventswere related to some malfunction of the stimulatingapparatus (e.g. unexpected battery depletion). Seizures,wound infection, sepsis, extradural haematoma,and pain induced by MCS have also been reported.Overall 20% of patients experience one or morecomplications, in general of benign nature. Details ofthe search with summary of benefits/harms can befound in Table 7 and 8.RecommendationsThere is level C evidence (two convincing class IIIstudies, 15–20 convergent class IV series) that MCS isuseful in 50–60% of patients with CPSP and central orperipheral facial neuropathic pain, with small risk ofmedical complications. The evidence about any othercondition remains insufficient.Repetitive transcranial magnetic stimulationThe use of rTMS in patients with chronic pain aims atproducing analgesic effects by means of a non-invasivecortical stimulation [47]. The stimulation is performedby applying on the scalp, above a targeted cortical region,the coil of a magnetic stimulator. A focal stimulationusing a figure-of-eight coil is mandatory. Theintensity of stimulation is expressed as a percentage ofthe motor threshold of a muscle at rest in the painfulterritory. The stimulation is performed just below motorthreshold. The frequency and the total number ofdelivered pulses depend on the study. One single sessionshould last at least 20 min and should include at least1000 pulses. Daily sessions can be repeated for one orseveral weeks. There is no induced pain and no need foranaesthesia or for hospital stay during the treatment.The rationale is the same as for implanted MCS. Thestimulation is thought to activate some fibres that runTable 7 Summary of efficacy and safety of MCS in CPSPIndicationVolume ofevidence [no. trials(no. patients)] <strong>EFNS</strong> class Summary of efficacy Summary of harms CommentsCPSPSystematic review andmeta-analysisNonePrimary studies(1991–2006)No RCT[20 cases series, withmuch overlap (143non-overlappingpatients)]Rasche et al. 2006,Nuti et al. 2005 [4](+Mertens et al. 1999+G-Larrea et al. 1999)[43]Saitoh et al. 2003(+Saitoh et al. 2001)Fukaya et al. 2003+Katayama et al. 2001 [39]+Katayama et al. 1998+Yamamoto et al. 1997Nguyen et al. 2000+Nguyen et al. 2000+Nguyen et al. 1999+Nguyen et al. 1997Nandi et al. 2002Carroll et al. 2000Fujii et al. 1997Katayama et al. 1994Tsubokawa et al. 1993+Tsubokaw et al. 1991Drouot et al. 2002All class IV(unless indicatedotherwise)IIIIIICase series(8–45 cases)Satisfactory painrelief (‡50%)reportedin 0–100% of cases(all series)In series with n >20 cases satisfactorypain relief in48–52% of patientsMost commoncomplications:26% (battery failure,seizures, woundinfection and sepsis)Pain induced by MCSPhantom painExtradural haematomaSeizuresHardware malfunctionOverall 20% of patientsexperience one or morecomplications and ingeneral of benignnatureMany patient duplicationsor reinterventions makingtotal nb of cases difficultto calculate. Reports withduplicated data werepooledEfficacy related topre-operative responseto drugs? (Yamamoto 1997,n ¼ 28)Efficacy related to sensorysymptoms? (Druot 2002,n ¼ 11)Efficacy related to motorsymptoms? (Katayama1998, n ¼ 31)CPSP, central post-stroke pain; MCS, motor cortex stimulation.Ó 2007 <strong>EFNS</strong> European Journal of Neurology 14, 952–970

Neurostimulation therapy for neuropathic pain 965Table 8 Summary of efficacy and safety of motor cortex stimulation in facial painIndicationVolume of evidence[no. trials (no. patients)]<strong>EFNS</strong>classSummaryof efficacy Summary of harms CommentsFacial painSystematic review andmeta-analysis:NonePrimary studies (1991–2006)No RCTCase series (60 patients)Rasche et al. 2006 (3/50)Brown Ptiliss 2005 (10/60)Nuti et al. 2005 [4] (5/60)Drouot et al. 2002 (15)Nguyen et al. 2000a (12/83)+Nguyen et al. 2000b same+Nguyen et al. 1999 same+Nguyen et al. 1997 (7/100)Ebel et al. 1996 (7/43)Katayama et al. 1994 (3/66)Meyersonl 1993 (5/100)All class IVCase seriesSatisfactory painrelief (‡50%)reported in 43–100%of cases(all series)No series with n >20casesMean percent ofpatients withsatisfactory painrelief: 66%Most commoncomplications:26% (battery failure, seizures,wound infection and sepsis)Pain induced by MCSExtradural haematomaSeizuresHardware malfunctionOverall 20% of patientsexperience one or morecomplications, in generalof benign natureMany patient duplicationsor reinterventions makingtotal nb of cases difficultto calculate. Reports withduplicated data were pooled.Small series but sometimeslong follow-up: 72 mMCS, motor cortex stimulation.through the motor cortex and project to remote structuresinvolved in some aspects of neuropathic painprocessing (emotional or sensori-discriminativecomponents). The method is non-invasive and can beapplied to any patient with drug-resistant, chronicneuropathic pain, who could be candidate for theimplantation of a cortical stimulator. As the clinicaleffects are rather modest and short-lasting beyond thetime of a single session of stimulation, this methodcannot be considered a therapy, except if the sessions ofstimulation are repeated for several days or weeks.Evidence identifiedWe identified some reviews, none systematic, and 14controlled studies that used sham stimulations incrossover or parallel groups, 280 patients with definiteneuropathic pain (CPSP, spinal cord lesions, trigeminalnerve, brachial plexus, or limb nerve lesions, phantompain and CRPS II). Efficacy, rather than varying betweenpain conditions, mostly depends on stimulationparameters. There is consensus from two RCTs in patientswith CPSP or various peripheral nerve lesionsthat rTMS of the primary motor cortex, when appliedat low-frequency (i.e. 1 Hz or less), is ineffective (classII) [48,49]. Focal-coil stimulations at high-rate (5–20 Hz), of long-duration (at least 1000 pulses) andpossibly repeated sessions, induce pain relief (>30%) inabout 50% of patients (class II/III) [50–52]. The effectbegins a few days later and its duration is short,

966 G. Cruccu et al.Table 9 Summary of efficacy and safety of rTMS, primary motor cortex stimulation, 1 Hz or lessIndicationVolume of evidence[no. trials(no. patients)]<strong>EFNS</strong>classSummaryof efficacySummaryof harms<strong>EFNS</strong>gradeCommentsStroke (n ¼ 32),Spinal cord lesion(n ¼ 4), Trigeminalnerve lesion (n ¼ 1),Brachial plexus orlimb nerve lesion(n ¼ 8),Phantom limb pain(n ¼ 14)New primary studiesThree sham-controlledtrialsAll negative (59 p.)Lefaucheur et al. 2001aAndre´ -Obadia et al. 2006Irlbacher et al. 2006IIIIIIINo efficacyPain relief‡30%: 5%(mean painrelief: 4%)No reportedcomplicationsBNo significant effectcompared with shamstimulationTable 10 Summary of efficacy and safety of rTMS, primary motor cortex stimulation, 5 Hz or moreIndicationVolume ofevidence [no. trials(no. patients)<strong>EFNS</strong>classSummary of efficacySummary ofharms<strong>EFNS</strong>gradeCommentsStroke (n ¼ 98),Spinal cord lesion(n ¼ 24), Trigeminalnerve lesion (n ¼ 60),Brachial plexus orlimb nerve lesion(n ¼ 36) CRPS(n ¼ 10)Stroke (n ¼ 20),Spinal cord lesion(n ¼ 6), Trigeminalnerve lesion (n ¼ 1),Brachial plexusor limb nerve lesion(n ¼ 8) Phantomlimb pain (n ¼ 15),CRPS (n ¼ 2),non-neuropathic(n ¼ 1)New primary studies[11 sham-controlledtrial (281 p.)]Positive studies(228 p.)Lefaucheur et al.2001aLefaucheur et al.2001bLefaucheur et al.2004aPleger et al. 2004Khedr et al. 2005Hirayama et al.2006Lefaucheur et al.2006aLefaucheur et al.2006bNegative studies(53 p.):Rollnik et al. 2002Andre´ -Obadiaet al. 2006Irlbacher et al. 2006IIIIIIIIIIIIIIIIIIIIIIIIIPain relief ‡30%:46%(mean pain relief:26%)104 responders/206patients. Efficacyregarding indication:idem for stroke vs.trigeminal nerve lesion(Lefaucheur et al. 2004;Khedr et al. 2005) orbrachial plexus lesion(Lefaucheur et al. 2004);better for thalamic vs.brainstem stroke(Lefaucheur et al. 2004)poorer results forspinal cord lesion(Lefaucheur et al. 2004)CRPS: no significantdifferencewith the other causes(Pleger et al. 2004)No reportedcomplicationsBNo significant effectcompared to shamstimulation in caseof circular coil(Rollnik et al. 2002)or 5 Hz-rTMS(Irlbacher et al. 2006)and

Neurostimulation therapy for neuropathic pain 967MCSEpidural stim.Motor cortexrTMSTranscranialMagnetic stim.Motor cortexDBSStim. Thalamus –Periventricular GreySCS(SCS Cervical)Epidural stim.Dorsal columnsTENSHigh frequencyLow intensityLow frequencyHigh intensityAcupuncture–likeFigure 1 Schematic representation of different neurostimulation procedures, e.g. for a patient with pain in the left hand because ofperipheral nerve injury.a placebo effect or not. This may also hold true forneuropathic pain patients. SCS has class II RCT evidence.Its efficacy has been so far demonstrated in twoconditions, which are not Ôdefinitely neuropathicÕ: FBSSand CRPS type I. Pain in FBSS is usually mixed and itis difficult to extract the neuropathic component, andCRPS I is still a Ôputative neuropathicÕ pain.Spinal cord stimulation, DBS and MCS are typicallyused when all other treatments have failed. This contextshould be taken into account when making recommendations.We analysed only published evidence.Thousands of stimulators are implanted every year andonly a tiny minority appears in published studies. Absenceof evidence is not evidence of absence of effect,and low-level evidence (i.e. case series) should be givensome credence. For some indications, there was a considerablebody of ÔpositiveÕ case series findings, sometimesover long periods of time. Furthermore, the wholefield has been largely characterized by a heavy dependenceon the outcome measure Ô50% pain reliefÕ as athreshold indicator of success, after both trial anddefinitive stimulation, which may distort the true picture.Others have found a 30% pain relief to correspondto a clinically meaningful success [5], and factorsbeyond changes in pain intensity are also relevant.Although not a new therapy, DBS has metamorphosedconsiderably over the last decade, concomitantlywith advances in both stimulator technology andneuroimaging techniques, leading to improved efficacyand reduced complications. DBS should be performedin experienced, specialist centres, using establishedoutcome measures and willing to publish their results.Whereas its efficacy in CPSP is controversial, DBSappears more promising for phantom limb pain andtrigeminal neuropathic pains.Motor cortex stimulation is useful in CPSP and incentral or peripheral facial pain. Interestingly, theproportion of good and excellent results increasesconsistently in patients with facial pain relative to allother classes. The reason is not yet established.Candidates for MCS have neuropathic pain that hasbeen resistant to drugs and often to other interventions.In view of the potential development of thismethod, it is of the utmost importance that placebo-Ó 2007 <strong>EFNS</strong> European Journal of Neurology 14, 952–970

968 G. Cruccu et al.controlled, double-blind studies are produced to increasethe level of evidence, particularly because ofMCS, not being perceived by the patient, allows aperfect placebo.As with TENS, the efficacy of rTMS seems to increasewith dose: higher frequency, longer duration ofthe session, and more sessions tend to yield better results.Because the clinical effects are rather modest andshort-lasting, rTMS cannot be considered as a therapeuticmethod for the long term, except if the sessionsof stimulation are repeated for several days or weeks.Currently, rTMS can be proposed as a non-invasivepre-operative therapeutic test for patients with drugresistantchronic pain who are candidates for surgicallyimplantedchronic MCS.Concerning harms (detailed in the Tables), TENSand rTMS are virtually harmless. SCS, DBS and MCSdo entail adverse events in a large proportion of patients(up to 20% with MCS and 40% with SCSexperience one or more complications). However, mostof these are simple lead migration or battery depletionthat do not produce physical harm and can usually besolved. Real harms are few, usually wound infection(3.4% with SCS, 7.3% with DBS and 2.2% withMCS) and very rare cases—often single cases—ofaseptic meningitis, transient paraparesis, epiduralhaematoma, epileptic seizures and skin reactions, nonebeing life-threatening. Our search disclosed one caseonly of pre-operative death 20 years ago [53]. Indeed,one of the reasons for the use of neurostimulationtherapy is that the application of low-intensity electricalcurrents is not associated with any of the sideeffects entailed by drugs.Finally, we feel that neurostimulation therapy willprove to be useful for a broader indication than issuggested by our search. We hope that future trials aredesigned bearing in mind the EBM requirements. Althoughit is admittedly difficult to find a credible placebofor neurostimulation therapy, the investigatorsmay compare their procedure to other treatments.Furthermore, we recommend that investigators payattention to definition of diagnosis, inclusion criteria,blind assessment of the outcomes, and impact on patient-relatedvariables such as quality of life and dailyliving activities.Declaration conflict of interestRST has a consultant contract with Medtronic, as anexpert in Health Care Policy and Clinical Trial Design.PH, LGL, JPL and BS received honorarium fromMedtronic for lectures or advisory boards. The otherauthors have nothing to declare.Supplementary MaterialsThe following supplementary material is available forthis article:Appendix 1. SCS search strategiesAppendix 2. Full list of studiesThis material is available as part of the online articlefrom:http://www.blackwell-synergy.com/doi/abs/10.1111/j.1468-1331.2007.01916.x(This link will take you to the article abstract).Please note: Blackwell Publishing are not responsiblefor the content or functionality of any supplementarymaterials supplied by the authors. Any queries (otherthan missing material) should be directed to the correspondingauthor for the article.References1. Finnerup NB, Otto M, McQuay HJ, et al. Algorithm forneuropathic pain treatment: an evidence based proposal.Pain 2005; 118: 289–305.2. Attal N, Cruccu G, Haanpaa M, et al. <strong>EFNS</strong> <strong>guidelines</strong>on pharmacological treatment of neuropathic pain.European Journal of Neurology 2006; 13: 1153–1169.3. Cruccu G, Anand P, Attal N, et al. <strong>EFNS</strong> <strong>guidelines</strong> onneuropathic pain assessment. European Journal ofNeurology 2004; 11: 153–162.4. Nuti C, Peyron R, Garcia-Larrea L, et al. Motor cortexstimulation for refractory neuropathic pain: four yearoutcome andpredictors of efficacy. Pain 2005; 118: 43–52.5. Farrar JT, Young JP Jr, LaMoreaux L, et al. Clinicalimportance of change in chronic pain intensity measuredon an 11-point numerical pain rating scale. Pain 2001; 94:149–158.6. Brainin M, Barnes M, Baron JC, et al. Guideline StandardsSubcommittee of the <strong>EFNS</strong> Scientific Committee.Guidance for the preparation of neurological management<strong>guidelines</strong> by <strong>EFNS</strong> scientific task forces – revisedrecommendations. European Journal of Neurology 2004;11: 577–581.7. Melzack R, Wall PD. Pain mechanisms: a new theory.Science (NY) 1965; 150: 971–979.8. Fukazawa Y, Maeda T, Hamabe W, et al. Activation ofspinal anti-analgesic system following electroacupuncturestimulation in rats. Journal of Pharmacological Science2005; 99: 408–414.9. Zhang GG, Yu C, Lee W, et al. Involvement of peripheralopioid mechanisms in electro-acupuncture analgesia.Explore (NY) 2005; 1: 365–371.10. McQuay HJ, Moore RA, Eccleston C, et al. Systematicreview of outpatient services for chronic pain control.Health Technology Assessment 1997; 1: 1–135.11. Reichstein L, Labrenz S, Ziegler D, et al. Effective treatmentof symptomatic diabetic polyneuropathy by highfrequencyexternal muscle stimulation. Diabetologia 2005;48: 824–828.12. Kumar D, Marshall HJ. Diabetic peripheral neuropathy:amelioration of pain with transcutaneous electrostimulation.Diabetes Care 1997; 20: 1702–1705.Ó 2007 <strong>EFNS</strong> European Journal of Neurology 14, 952–970

Neurostimulation therapy for neuropathic pain 96913. Hamza MA, White PF, Craig WF, et al. Percutaneouselectrical nerve stimulation: a novel analgesic therapy fordiabetic neuropathic pain. Diabetes Care 2000; 23: 365–370.14. Forst T, Nguyen M, Forst S, et al. Impact of low frequencytranscutaneous electrical nerve stimulation onsymptomatic diabetic neuropathy using the new salutarisdevice. Diabetes, Nutrition and Metabolism 2004; 17: 163–168.15. Cheing GL, Luk ML. Transcutaneous electrical nervestimulation for neuropathic pain. The Journal of HandSurgery 2005; 30: 50–55.16. Thorsteinsson G, Stonnington HH, Stillwell GK, et al.Transcutaneous electrical stimulation: a double-blind trialof its efficacy for pain. Archives of Physical Medcine andRehabilitation 1977; 58: 8–13.17. Rutgers MJ, Van Romunde LKJ, Osman PO. A smallrandomized comparative trial of acupuncture versustranscutaneous electrical neurostimulation in postherpeticneuralgia. Pain Clinic 1988; 2: 87–89.18. Bloodworth DM, Nguyen BN, Garver W, et al. Comparisonof stochastic vs. conventional transcutaneouselectrical stimulation for pain modulation in patients withelectromyographically documented radiculopathy. AmericanJournal of Physical Medicine and Rehabilitation 2004;83: 584–591.19. Sindou MP, Mertens P, Bendavid U, et al. Predictivevalue of somatosensory evoked potentials for long-lastingpain relief after spinal cord stimulation: practical use forpatient selection. Neurosurgery 2003; 52: 1374–1383.20. Cameron T. Safety and efficacy of spinal cord stimulationfor the treatment of chronic pain: a 20-year literature review.Journal of Neurosurgery 2004; 100(3 Suppl. Spine):254–267.21. Taylor RS, Van Buyten JP, Buchser E. Systematic reviewand meta-analysis of the effectiveness of spinal cordstimulation in the management of failed back surgerysyndrome. Spine 2005; 30: 152–160.22. Taylor RS, Van Buyten JP, Buchser E. Spinal cord stimulationfor complex regional pain syndrome: a systematicreview of the clinical and cost effectiveness literature andassessment of prognostic factors. European Journal ofPain 2006; 10: 91–101.23. Simpson BA. Spinal cord stimulation. Pain Reviews 1994;1: 199–230.24. Simpson BA. Spinal cord and brain stimulation. In: WallPD, Melzack R, eds. Textbook of Pain, 4th edn. London:Churchill Livingstone, 1999: 1253–1381.25. Simpson BA, Meyerson BA, Linderoth B. Spinal cord andbrain stimulation. In: McMahon SB, Koltzenburg M, eds.Wall and Melzack’s Textbook of Pain, 5th edn. ElsevierChurchill Livingstone, 2006: 563–582.26. North RB, Kidd DH, Farrokhi F, et al. Spinal cordstimulation versus repeated lumbosacral spine surgery forchronic pain: a randomized, controlled trial. Neurosurgery2005; 56: 98–106.27. Kumar K, North R, Taylor R, et al. Spinal cord stimulationversus conventional medical management: a prospective,randomised, controlled, multicentre study ofpatients with failed back surgery syndrome (PROCESSstudy). Neuromodulation 2005; 8: 213–218.28. Kumar K, Taylor RS, Jacques L, et al. Spinal cordstimulation versus conventional medical management forneuropathic pain: a multicentre randomised controlledtrial in patients with failed back surgery syndrome. Pain2007; accepted for publication.29. Kemler MA, Barendse GAM, Van Kleef M, et al. Spinalcord stimulation in patients with chronic reflex sympatheticdystrophy. The New England Journal of Medicine2000; 343: 618–624.30. Kemler MA, de Vet HC, Barendse GA, et al. Spinal cordstimulation for chronic reflex sympathetic dystrophy –five-year follow-up. The New England Journal of Medicine2006; 354: 2394–2396.31. Heath RG, Mickle WA. Evaluation of seven yearsÕexperience with depth electrode studies in human patients.In: Ramey ER, O’Doherty DS, eds. Electrical Studies onthe Unanesthetized Brain. New York: Paul B. Hoeber,1960: 214–247.32. Hosobuchi Y, Adams JE, Rutkin B. Chronic thalamicstimulation for the control of facial anesthesia dolorosa.Archives of Neurology 1973; 29: 158–161.33. Richardson DE, Akil H. Long term results of periventriculargray self-stimulation. Neurosurgery 1977; 1: 199–202.34. Green AL, Wang S, Owen SL, et al. Stimulating the humanmidbrain to reveal the link between pain and bloodpressure. Pain 2006; 124: 349–359.35. Owen SL, Green AL, Nandi D, et al. Deep brain stimulationfor neuropathic pain. Neuromodulation 2006; 9:100–106.36. Owen SL, Green AL, Stein JF, et al. Deep brain stimulationfor the alleviation of post-stroke neuropathic pain.Pain 2006; 120: 202–206.37. Bittar RG, Kar-Purkayastha I, Owen SL, et al. Deepbrain stimulation for pain relief: a meta-analysis. Journalof Clinical Neuroscience 2005; 12: 515–519.38. Hamani C, Schwalb JM, Rezai AR. Deep brain stimulationfor chronic neuropathic pain: long-term outcome andthe incidence of insertional effect. Pain 2006; 125: 188–196.39. Katayama Y, Yamamoto T, Kobayashi K, et al. Motorcortex stimulation for post-stroke pain: comparison ofspinal cord and thalamic stimulation. Stereotactic FunctionalNeurosurgery 2001; 77: 183–186.40. Tsubokawa T, Katayama Y, Yamamoto T, et al. Treatmentof thalamic pain by chronic motor cortex stimulation.Pacing and Clinical Electrophysiology 1991; 14: 131–134.41. Tsubokawa T, Katayama Y, Yamamoto T, et al. Chronicmotor cortex stimulation in patients with thalamic pain.Journal of Neurosurgery 1993; 78: 393–401.42. Peyron R, Garcia-Larrea L, Deiber MP, et al. Electricalstimulation of precentral cortical area in the treatment ofcentral pain: electrophysiological and PET study. Pain1995; 62: 275–286.43. Garcia-Larrea L, Peyron R, Mertens P, et al. Electricalstimulation of motor cortex for pain control: a combinedPET-scan and electrophysiological study. Pain 1999; 83:259–273.44. Rusina R, Vaculin S, Yamamotova A, et al. The effect ofmotor cortex stimulation in deafferentated rats. NeuroEndocrinology Letters 2005; 26: 283–288.45. Senapati AK, Huntington PJ, Peng YB. Spinal dorsalhorn neuron response to mechanical stimuli is decreasedby electrical stimulation of the primary motor cortex.Brain Research 2005; 1036: 173–179.Ó 2007 <strong>EFNS</strong> European Journal of Neurology 14, 952–970

970 G. Cruccu et al.46. Peyron R, Faillenot I, Mertens P, et al. Motor cortexstimulation in neuropathic pain. Correlations betweenanalgesiceffect and hemodynamic changes in the brain. APET study. Neuroimage 2007; 34: 310–321.47. Lefaucheur JP. The use of repetitive transcranial magneticstimulation (rTMS) in chronic neuropathic pain. NeurophysiologieClinique 2006; 36: 117–124.48. Lefaucheur JP, Drouot X, Keravel Y, et al. Pain reliefinduced by repetitive transcranial magnetic stimulation ofprecentral cortex. Neuroreport 2001; 12: 2963–2965.49. Andre-Obadia N, Peyron R, Mertens P, et al. Transcranialmagnetic stimulation for pain control. Doubleblindstudy of different frequencies against placebo, andcorrelation with motor cortex stimulation efficacy.Clinical Neurophysiology 2006; 117: 1536–1544.50. Khedr EM, Kotb H, Kamel NF, et al. Longlasting antalgiceffects of daily sessions of repetitive transcranialmagnetic stimulation in central and peripheral neuropathicpain. Journal of Neurology, Neurosurgery andPsychiatry 2005; 76: 833–838.51. Lefaucheur JP, Drouot X, Nguyen JP. Interventionalneurophysiology for pain control: duration of pain relieffollowing repetitive transcranial magnetic stimulation ofthe motor cortex. Neurophysiologie Clinique 2001; 31:247–252.52. Lefaucheur JP, Drouot X, Menard-Lefaucheur I, et al.Neurogenic pain relief by repetitive transcranial magneticcortical stimulation depends on the origin and the site ofpain. Journal of Neurology, Neurosurgery and Psychiatry2004; 75: 612–616.53. Levy RM, Lamb S, Adams JE. Treatment of chronic painby deep brain stimulation: long term follow-up and reviewof the literature. Neurosurgery 1987; 21: 885–893.Ó 2007 <strong>EFNS</strong> European Journal of Neurology 14, 952–970