8.4.1. Introduction and current practice
Neurosurgery for mental disorders has changed significantly since its initial introduction in the 1930s (see Freeman et al., 2000, for a historical overview) and although there remains interest worldwide, the number of centres in the UK offering neurosurgery has decreased over the last 10 years. Freeman and colleagues (2000) suggest there is greater specificity as procedures have evolved, but it is less clear whether surgical innovation is due to theoretical advances or to pragmatic considerations. Neurosurgery has been recommended by the Expert Consensus Guideline for treatment refractory OCD in the case of non-response to three or more trials of SRIs (including clomipramine) and to CBT as an ‘infrequently needed, but sometimes life saving intervention’ (March et al., 1997). Although the number of neurosurgical interventions conducted in the UK for OCD has decreased in recent times (see Freeman et al., 2000), there are currently two centres (Cardiff and Dundee) where patients with OCD from England and Wales may be referred for assessment for possible neurosurgical intervention (Matthews & Eljamel, 2003).
It is important to review the evidence for ablative neurosurgery for OCD given that earlier reviews have reported promising results. For example, Kiloh and colleagues (1988) reported that among 478 patients from 24 studies between 1961 and 1980, 58% showed marked improvement. Over half the operations in this review were non-stereotactically guided. A review of 12 studies from 1961 to 1988 by Waziri (1990) reported that 67% of the 300 patients fell into the ‘symptom free’ or ‘minor symptoms’ categories; all but three studies used stereotactically guided procedures. Finally, Freeman and colleagues (2000) reviewed five studies involving 198 patients, and found an identical result of 67% fell into these categories.
Although a variety of different procedures are used, all involve the ablation or disconnection, of ventral and medial prefrontal cortical areas. Four main ablative procedures have been used in OCD:
Anterior capsulotomy
Anterior cingulotomy
Subcaudate tractotomy
Limbic leucotomy.
In addition, there is one non-ablative procedure, deep brain stimulation, which has been investigated with OCD. A second procedure, vagal nerve stimulation, has been discussed as a potential treatment for OCD. Although it is used for other conditions including depression (George, 2000), it has not, to our knowledge, been investigated in the treatment of OCD and will not be reviewed.
8.4.3. Ablative procedures
8.4.3.1. Subcaudate tractotomy
This procedure was developed by Geoffrey Knight in the UK (1969). Although it was used extensively for affective disorders for several years, the operation is no longer performed. Radioactive 90-Yttrium rods were inserted into the target area, a region called the substantia innominata, found below the head of the caudate nucleus. Much of the literature on this technique describes the treatment of depression and there are limited reports on its use for OCD (Bartlett & Bridges, 1977; Cosyns et al., 1994; Goktepe et al., 1975; Hodgkiss et al., 1995; Strom-Olsen & Carlisle, 1971). In an early study (Strom-Olsen & Carlisle, 1971), 20 patients with OCD received this procedure; ten improved, but four relapsed during the follow-up period. In a further study (Goktepe et al., 1975), 50% of 18 patients were reported to have shown significant improvement or better. These early studies reported global ratings only rather than specific measures of OCD symptoms. Although this intervention as described above is no longer practiced as such, lesions in the same area of the brain are part of the limbic leucotomy described below.
8.4.3.2. Anterior capsulotomy
Two main procedures for making lesions in the anterior capsule have been described, namely, radiofrequency thermocapsulotomy and radiosurgical gamma knife capsulotomy (Rasmussen et al., 2000). There are a number of earlier reports between 1961 and 1982 but these used global rather than OCD-specific scales to evaluate thermo-capsulotomy (see Freeman et al., 2000 for a review).
One prospective series from the Karolinska Hospital in Stockholm, Sweden, has been reported on extensively (Lippitz et al., 1997; Lippitz et al., 1999; Mindus et al., 1994; Mindus et al., 1999; Nyman et al., 2001). Twenty-four patients from 1979–1990 underwent thermocapsulotomy and 19 were included in a study on personality characteristics (Mindus et al., 1999). Patients were assessed pre-intervention and a mean of 8 years later. Five of the patients received a second intervention during the follow-up period. The authors reported that five patients were unchanged or deteriorated on the CPRS-OC, six were less than 50% improved and eight were more than 50% improved. On 15 self-report scales designed to measure personality (Karolinska Scales of Personality), there was a decrease at 8-year follow-up (average) in anxiety proneness dimensions, but no evidence of changes for other aspects of personality, except one person who had haemorrhaged showed increases in psychopathic traits.
The patients who had received thermocapsulotomy between 1978 and 1999 (n = 21) were compared at follow-up with a group who had been assessed but had not yet received neurosurgery (n = 8) (Nyman et al., 2001). No significant differences were found on a battery of neuropsychological tests but the study would be, in all likelihood, insufficiently powered to detect changes on these tests given the small numbers of patients involved. The authors conclude that ‘patients with OCD by capsulotomy generally perform in the lower region of the normal range or show mild impairment on standardized tests 2 to 15 years after the operation’.
Lippitz and colleagues (1999) reported again on this series but also included those who had received the gamma knife intervention. They conducted a retrospective study to define potential common topographic denominators among the lesions from the patients reported earlier (Mindus et al., 1999) who had been successfully treated with gamma knife capsulotomy (n = 10). The 19 patients receiving thermocapsulotomy reported in this study are the same as those reported in the Mindus and colleagues (1999) study. Magnetic Resonance Imaging indicated the location and size of the lesion. Clinical outcome was reported, mostly using the CPRS-OC although later patients may have been rated on the Y-BOCS. Results are presented as percentage improvement on psychiatric rating scales and showed that 9/19 of the patients who received thermocapsulotomy and 7/10 who received gamma knife capsulotomy showed at least a 50% improvement on either CPRS or Y-BOCS.
Christensen and colleagues (2002) reported on two cases of severe OCD in ‘younger’ (18-yrs) and ‘older’ (64-yrs) patients and who received capsulotomy. They reported a successful outcome based on a reduction in OCD symptoms on the Y-BOCS.
Oliver and colleagues (2003) reported on a series of 15 patients who underwent thermo-capsulotomy. Three patients received a second capsulotomy. Forty-six per cent reported a 50% or greater reduction on the Y-BOCS. Two patients reported transient adverse effects (one case each of hallucinations and seizure) and one ‘postoperative bifrontal swelling with permanent behaviour impairment’. Finally, a review by Greenberg and colleagues (2003) reported preliminary results of an unpublished study of 15 patients who had undergone gamma knife capsulotomy. The authors reported that single bilateral lesions in the anterior capsule were ineffective. However, following placement of a second set of bilateral lesions, four of the 15 were judged as showing a 35% decrease on the Y-BOCS and at least a 15-point improvement on the Global Assessment Scale at 5-year follow-up. They also reported a second study of 16 patients who received two pairs of bilateral lesions. At 3-year follow-up, ten met the same improvement criteria. Although the second series would seem to suggest a more positive outcome than the first series, results from neither series have yet been subject to peer review and should be treated cautiously.
In summary, anterior capsulotomy has been reported to be effective based mainly on retrospective trials using global ratings and, more recently, on prospective case series or open trials using OCD specific measures. Although the data are reported in increasing detail on a broader range of measures, conclusions are ultimately limited by the design and the small number of patients involved in the more recent series. The serious persistent adverse effects (1/19 in the Swedish series and 1/15 in the Oliver study) would suggest caution.
8.4.3.3. Anterior cingulotomy
In an early study, Fodstad and colleagues (1982) randomised two patients to either stereotactic anterior capsulotomy or cingulotomy. Follow-up to 12 to 24 months indicated that all four remained improved or much improved at the last follow-up. However, those who had received cingulotomy were judged as having a lesser response compared with those who had received capsulotomy.
A case series of five patients underwent what is labelled a ‘modified leucotomy’ but describe lesions to the cingulated gyrus and so would correspond to cingulotomy (Tippin & Henn, 1982). The authors reported improvement on a five-point clinician-rated global outcome measure, with four out of five classed as marked improvement or symptom free at follow-ups at 1 to 6.6 years. No specific measures of OCD were used.
There are two series, one retrospective and one prospective, from Massachusetts General Hospital in Boston that have each been described in several reports. The retrospective series described in most detail by Jenike and colleagues (1991) followed-up 35 patients with OCD who had received cingulotomies from 1962 until the late eighties. It was determined that 33 would have met criteria for OCD according to DSM-III-R; 26 of these also met criteria for depression. Of the 33, 23 had received additional interventions, second cingulotomies in 16 cases, second and third cingulotomies in four cases, and other interventions in six cases. In this early series, only three patients had received behaviour therapy and six had received clomipramine pre-operatively. Adverse effects included seizures (3/33), decreased memory (1/33), suicide (4/33) and death from myocardial infarction 6 weeks post-operatively (1/33 with previous history of cardiac problems). Based on retrospective ratings of preoperative symptoms of 14 patients who were interviewed, 8 (58%) showed moderate to marked improvement. However, in at least two cases, improvement was attributed to additional treatment. Overall, the authors conclude that 9/29 surviving patients showed significant improvement that could be attributed to psychosurgery.
Dougherty (2002) reported on a prospective study of 44 patients with treatment refractory OCD who received cingulotomy from 1989 onwards. Data from some of these patients had previously been reported elsewhere (Baer et al., 1995; Spangler et al., 1996). All patients had symptoms that had previously not responded to treatment regimes of both medication and behaviour therapy. Of the 44 patients operated on during the study period, 17 patients had received two cingulotomies and one had undergone a third. Clinical outcome measures, including the Y-BOCS, were administered pre- and post-intervention (mean of 6.7 months), at the first post-surgical procedure (mean of 7 months) after the second cingulotomy, and at a mean of 32 months' follow-up from the first cingulotomy. Response rate was defined as 35% improvement on the Y-BOCS and a CGI Global Improvement score of less than or equal to 2. In addition, to be considered as a responder, patients needed to attribute their improvement to the cingulotomy. At first follow-up, 5/44 patients met the response criteria and 14/44 at final follow-up. Nine patients (9/44) reported adverse effects post-operatively: memory problems (2/44), apathy (1/44), urinary disturbance (3/44), seizure (1/44), post-operative oedema with resulting hydrocephalus (1/44). One patient, who showed improvement in OCD symptoms, committed suicide 6 years after the operation. In all but two cases (one each of urinary incontinence and seizure), the adverse effects resolved.
Finally, 14 patients with refractory OCD underwent cingulotomy in a prospective study in Korea (Kim et al., 2003). All patients referred for the trial had received behaviour therapy and medication. Patients were assessed using the Y-BOCS, CGI and the Hamilton Rating Scales for Depression and Anxiety. The response criteria were defined as at least 35% improvement on the Y-BOCS and a CGI score of 1 or 2 (improved or very much improved). Four patients met this criterion at 6-month follow-up and six at 12-month follow-up. Mean improvement rate on Y-BOCS was 36%. Two patients reported transient headache, three gained weight and one lost weight but normalised subsequently. There was no evidence of cognitive dysfunction on a range of measures.
In sum, these studies would suggest that cingulotomy has shown some effects with up to a third of patients meeting differing criteria for response. The two prospective studies are stronger designs than those found for some other interventions and used conventional definitions of response (35% reduction in Y-BOCS score). Nevertheless, they remain limited by the lack of control groups. There were relatively few persistent adverse effects but the suicide rate in the retrospective study (Jenike et al., 1991) would indicate caution.
8.4.3.4. Limbic leucotomy
Limbic leucotomy, a multi-target procedure, was developed in the UK by Desmond Kelly in the 1970s and consists of lesions corresponding to both cingulotomy and subcaudate tractotomy. An early study reported a high response rate with OCD (Mitchell-Heggs et al., 1976) but was methodologically flawed and the results have been disputed (Chiocca & Matuza, 1990).
A number of reports describe an Australian series of 26 patients operated on from 1972 until 1989, most of whom (n = 17) received both cingulate and orbitomedial lesions similar to limbic leucotomy while six received cingulate lesions and four received orbitomedial lesions only (Cumming et al., 1995; Hay et al., 1993; Sachdev & Hay, 1995; Sachdev, 1996). The first six patients received the orbitomedial lesions through open neurosurgery; the remaining all received stereotactic interventions. Although this series is more difficult to interpret because of the mixed interventions, it is worth considering because of the extensive reports on cognitive and personality function. At follow-up of 10 years (mean), five (19%) were considered much improved or recovered, five were moderately improved, six were mildly improved, six showed no improvement, four were worse (Hay et al., 1993). Of the four who were rated worse, three died by suicide and one had marked personality change. In addition to transient post-operative adverse effects in three patients, one suffered from post-operative haemorrhage, delirium, periods of psychosis and permanent personality change. An additional three were considered to have shown personality change and two suffered from recurrent seizures. In all, 6/26 suffered serious permanent adverse effects. In a report on self- and informant-rated personality change in a subgroup of 16, the majority reported little change in personality across 34 items (Sachdev & Hay, 1995).
Finally, 17 of this series were compared with a control group of patients with long-term OCD on a battery of neuropsychological tests. There was no evidence of impaired IQ or memory function compared with the OCD controls, but there was evidence of impaired performance in the operated group on executive function (Cumming et al., 1995). Interestingly, a magnetic resonance imaging (MRI) study on 14 of these patients revealed accurately placed orbitomedial lesions (10/10), accurately placed cingulated lesions in only 10 of 13, and inadvertent lesions to the anterior capsule in three cases (Sachdev, 1996).
A more recent case series has been reported with 21 patients with major depressive disorder (MDD) or OCD who underwent stereotactic limbic leucotomy between 1993 and 1999 (Montoya et al., 2002). All had symptoms that failed to respond to pharmacotherapy and over three quarters had received ECT.
In this study, there were 15 patients for whom the primary indication for surgery was OCD, although three of those whose primary indication was for MDD also had OCD. Eight of those in the OCD group also received a diagnosis of MDD. For a significant proportion of the participants (76%), limbic leucotomy was the second (following unsuccessful bilateral anterior cingulotomy, n = 5) or third neurosurgical procedure (following enlargement of earlier cingulotomy lesions). It is unclear what percentage of patients with OCD had undergone a second or third operation as the authors have reported for the whole sample only.
Two patients, one with MDD and one with OCD, both with previous history of suicidal attempts, died by suicide during the follow-up period. Transient somnolence (6/21), apathy (5/21), and fever (2/21) were reported. Post-operative and persistent headache were reported by one patient. Four patients reported post-operative seizures that were transient in all but one. Five patients (5/21) reported bladder incontinence that was persistent in three (3/21). Finally, five patients (24%) reported short-term memory problems that were persistent for two people (10%).
Outcome measures used were the Y-BOCS and CGI Global Improvement scale. Response was determined in a similar way to other neurosurgical studies of OCD, namely, an improvement score of 1 or 2 on the CGI (very much or much improved) and a 35% decrease in Y-BOCS score. Of the 15 patients with OCD, only nine had pre- and post-operative ratings on the Y-BOCS and only eight had both Y-BOCS and CGI ratings. Of these eight patients, only two (25%) met the double criteria for response. When based on one or other of the measures the authors reported a 36% (Y-BOCS) to 62% (self-rated CGI) response. The authors also reported that for the entire cohort (N = 21), the response rate was better for those undergoing limbic leucotomy where it was the second or third procedure. In sum, this case series describes a group of chronically ill patients who had received multiple previous non-neurosurgical treatments and indeed 76% had also received one or more previous cingulotomies. The findings are limited by the partial data collected and the difficulty in separating out some of the findings for the OCD patients alone.
Given the shortcomings in the designs, the lack of robust evidence of efficacy and the high rate of persistent adverse events reported in both series, the evidence for this type of multi-target procedure is not promising.
8.4.3.5. Neurosurgery in BDD
Phillips (2002) noted one published case report and two personal communications describing benefit in three individuals with BDD (modified leucotomy in one, capsulotomy in one, and bilateral anterior cingulotomy and subcaudate tractotomy in one) and no benefit in two individuals (who received anterior capsulotomy).
8.4.3.6. Clinical summary
Although there are reports of improvement for three of the procedures (capsulotomy, cingulotomy and limbic leucotomy), there are also sufficient reports of both transient and persistent adverse effects to cause concern. The more recent prospective series all report on criteria for entry to these studies that include previous trials of pharmacological treatments and ERP, supporting the contention that at least in the last 15 years, these are generally treatments of last resort. Given the relative rarity of these interventions, studies are generally small and conducted over long periods. Importantly, none have control conditions, although it is unlikely that credible sham procedures could be ethical for ablative procedures, nor are they likely to be possible as patients are unlikely to accept randomisation for treatments of last resort.
Some studies have used comparison groups to compare neuropsychological function and personality function in people with OCD who have received neurosurgery with those who have not. While these studies are attempting to answer important questions about effects of neurosurgery, the small sample size and the fact that they are essentially testing null hypotheses means that the most that can be concluded is the absence of large effects, but nothing can be said about whether smaller but potentially clinically significant differences may be present. Although there have been significant improvements in the measurement of obsessive-compulsive symptoms, there are still few prospective studies investigating neuropsychological changes and those measuring personality changes are using measures that are not widely known. In summary, the evidence on whether adverse changes occur in neuropsychological and personality function is inconclusive.
8.4.4. Non-ablative procedures
Concerns regarding the irreversibility and possible long-term adverse effects of ablative neurosurgical procedures have led to the investigation of a number of nondestructive neurophysiological interventions. Tissue damage could still occur during the intervention or through repeated stimulation, but the intervention does not seek to produce lesions.
8.4.4.1. Deep brain stimulation
Electrical deep brain stimulation is a relatively new technique and developed as a treatment for OCD through collaboration between Belgian and Swedish researchers. The effects of electrical stimulation of the brain have been previously investigated during stereotactic surgery for OCD before permanent lesions were made. For example, Laitinen and Singounas (1988) reported on the effects of stimulation in a series of 20 patients undergoing neurosurgery under local anaesthetic. However, the therapeutic use differs in that electrodes are implanted within brain structures, which are then stimulated via an external electrical source. Any lesions that result are not deliberate and are considered to be small relative to those made deliberately during ablative procedures. The efficacy and safety of DBS for the treatment of neurological conditions such as Parkinson's disease are well established. It has not yet been used for OCD in the UK although it is used for Parkinson's disease.
Essentially, DBS uses high frequency pulses that have complex effects including blocking of the targeted area and mimics the effect of tissue lesioning without destroying them (Tass et al., 2003). In an initial report, Nuttin and colleagues (1999) described four patients who were treated with DBS; three were reported to have improved although little detail was provided. Since then, a series of six patients has been reported in several publications that address a range of variables as well as issues about placement of electrodes and stimulation parameters (Gabriels et al., 2003; Nuttin et al., 2003a; Nuttin et al., 2003b). In this series, quadripolar electrodes stereotactically implanted in both anterior limbs of the internal capsules in six patients with OCD, all of whom had been deemed to have severe OCD by a selection committee. Four of the patients were randomly crossed over from continuous stimulation to stimulation off. Two were not crossed over (one received a capsulotomy and one was still in the post-operative screening phase).
The authors reported improvement in the ‘stimulator on’ condition compared with ‘stimulator off’, but these data should be interpreted with caution due to the small size of the study. Of the four patients, one did not respond and 3 responded (improvement of >35% on the Y-BOCS). A number of side effects were reported including transient hypomanic states, swelling of the face, awareness of the leads attached to the electrodes, fatigue, and weight loss/gain (Nuttin et al., 2003b). There were also technical difficulties due to broken electrical contacts and fracture of an electrode, but the main technical issue was short battery life necessitating replacement of the stimulators every 5 to 12 months. There is one additional independent report of DBS (Anderson & Ahmed, 2003) that reported a positive outcome in a case of severe OCD.
8.4.4.2. Clinical summary
DBS is a very recent procedure. The studies so far are too small to reach any conclusions about efficacy. The side effects reported so far suggest extreme vigilance especially as it is not yet established whether DBS is completely reversible. The technical issues around battery life would seem to present a significant limitation at the moment. The relatively non-destructive nature of the intervention means that well designed controlled trials would be possible.
8.4.5. Issues about neurosurgery for OCD
All of the recent series report selection criteria and despite agreement on general principles such as severity, disability and non-response to previous treatment, there is a degree of variability. There is also wide variability in terms of the range of assessments used although all recent studies have used the Y-BOCS. Consequently, combining or comparing data from different centres is almost impossible and centres with few operations do not contribute at all to these series. As several authors have pointed out, international agreement on these issues would allow the field to advance.
The main issues to be agreed, both in routine practice and research, are selection criteria that reflect severity and chronicity of OCD, the definition of adequate previous treatment, the need for independent oversight, a standardised assessment protocol pre-and post-operatively, and agreements and protocols about post-operative care.
Matthews and Eljamel (2003) have outlined a detailed series of criteria both for inclusion and for determining adequacy of previous treatment. These criteria are broadly in line with, and generally exceed, those reported by authors in Sweden, the US and Korea. These criteria, when interpreted in the spirit intended by the authors themselves, will result in decisions that properly balance the potential risks against the potential benefits and reserve consideration of such interventions for truly intractable cases of OCD. There is a potential problem with all guidelines, which results not from their intention, but from the possibility that they are applied in a superficial or unsophisticated way. In particular, the question of the adequacy of previous treatment is particularly difficult to determine. As Bejerot (2003) points out, ‘yesterday's intractable OCD patient may well be treatable today and intractability may depend on the ambition of the prescriber’.
Although this statement apparently refers to pharmacological treatments, similar issues apply to psychological treatments. For psychological treatments, an adequate trial cannot be defined by number, duration and frequency of sessions alone. Engagement in therapy is believed to be essential to outcome and therapeutic strategies must be implemented. Unwillingness or inability to engage actively in therapy is a feature of some people with severe OCD. Sometimes several unsuccessful attempts to engage in therapy precede a successful attempt, perhaps with a therapist of equal skill who has a different style, or who is using the same basic techniques in a different manner. Consequently, guidelines such as those by Matthews and Eljamel (2003) must remain guidelines to inform sound clinical judgment as the authors themselves intend through thorough consideration of the complex issues involved in determining adequacy of previous treatment.
It is essential, as is commonly the practice in the UK, that multidisciplinary teams with specific expertise in the management and treatment of OCD be involved in treatment and assessment before concluding that a patient has not responded to adequate treatment. In addition, given that non-ablative methods are now being developed, it may be that such methods should be considered first when evidence, or more conclusive evidence, for their efficacy in OCD becomes available.
Assessment should include standardised measures of obsessive-compulsive symptoms, depression and anxiety, psychosocial functioning including quality of life, personality, cognitive function, and possible adverse effects. Ratings should be sought from independent assessors, the patient, and family members or friends. If it is decided to proceed with an intervention, protocols for post-operative care should be agreed that allow provision for pharmacological and psychological care.
8.4.6. Clinical summary
The evidence is generally inconclusive for each of the medical interventions considered. In addition to a variety of design issues that prevent strong inferences from being made, evidence for efficacy is limited. For those with the strongest supportive clinical evidence, the proportion of those who respond must be considered within the context that all neurosurgical techniques have the potential for serious adverse effects, including persistent ones as reported in many of the studies reviewed. However, it is important to note that, especially in the more recent studies, many of the participating patients (although not all) had severe OCD with high Y-BOCS scores, poor functioning, and significant levels of psychiatric comorbidity, including, in many cases, severe depression. In addition, most participants had previously received evidence-based standard treatments and independent selection committees had approved surgery. For such patients, considered refractory to other treatment approaches, even the relatively modest rates and degrees of response reported in these studies may be clinically important.
At present, there is no compelling evidence comparing different neurosurgical procedures. Several studies have suggested multiple lesions may be needed for both capsulotomy and cingulotomy which may, as some data would suggest, increase the chance of response, but may also increase the risk of adverse effects. Small samples have prevented identification of predictors of response to any of the treatments. In conclusion, the quality of available evidence addressing the efficacy and safety of neurosurgical treatments is variable and does not support a clear recommendation. However, it is possible that a small number of patients with the most severe, chronic, disabling and treatment-refractory forms of OCD, for whom quality of life is very poor, will continue to be assessed as candidates for neurosurgery. This may, in the future, include non-ablative procedures. Such assessments should be conducted by established, expert, multidisciplinary teams who possess experience in the management of severe OCD, work from detailed pre- and post-operative protocols and structured long-term follow-up, are subject to appropriate independent oversight, and are committed to sharing and publication of audit information.
