The best way of surgically treating severe pain is to identify the cause of the pain and to remove it. This objective is not always possible; the cause of the pain may not be identifiable or it may not be amenable to treatment by surgery or by other means. So-called intractable pain therefore has to be treated symptomatically.

The optimal surgical treatment for any intractable pain would have its effect confined to the painful area, be simple and inexpensive to perform, and be associated with a low mortality and morbidity. In particular, it should be associated with a low risk of neurological deficit. It should also be effective and long lasting. Unfortunately, most forms of neurosurgical intervention for pain relief involve creating a lesion within the nervous system, with inevitable risk to neurological function. For this reason neurosurgeons are often called in much too late in the course of treatment of patients with intractable pain.

There are three principal methods of neurosurgical treatment for pain. These comprise:

1.Operations that interrupt nociceptive pathways by creating lesions in peripheral nerves, nerve roots or ganglia, the spinal cord, various parts of the brain and brain-stem, and the sympathetic nervous system.

2.Electrical stimulation of pain suppressive systems or blocking pain pathways (peripheral nerve, spinal cord, or brain).

3.Administration of various drugs to the intraspinal or intraventricular compartments of the cerebrospinal fluid pathways.

The choice of treatment will be determined not only by the most appropriate therapy for the particular pain syndrome but will also depend upon whether the pain is caused by a benign or malignant condition, the age and general medical condition of the patient, life expectancy, the mental state and, to some extent, the intelligence of the patient if some form of patient-operated implant is being considered. The balance between the cost of surgical and conservative forms of treatment may also be a deciding factor.

Only a few of the more common forms of intractable pain (Table 1) 603 are discussed here. The treatment of trigeminal neuralgia is reviewed elsewhere.

One should generally start with the simplest and most peripheral procedure and progress centrally, since more central operations may be associated with serious complications or side-effects. Alternatives such as spinal cord stimulation may be innocuous but the cost of such procedures may make them inappropriate in patients with pain due to malignancy and who have limited life expectancy.

A careful history and examination must first be undertaken to establish whether there is an identifiable and treatable cause for the pain. The exact area affected must be determined, along with the type of pain (burning, stabbing, nagging, tingling) and any altered sensation in the affected area. This last point will help to establish whether the pain is somatic or neurogenic. The former is probably due to chronic activation of nociceptors responsible for acute pain, while the latter (also known as central, deafferentation or dysaesthetic pain) is probably a result of changes within some part of the transmission system rather than of activation of peripheral nociceptors. Deafferentation pain is more difficult to treat and includes all forms of pain associated with neurological injury excepting that caused by neurological compression. Most operations undertaken to relieve pain tend to cause a degree of numbness in the affected area and there is often little point in embarking on such procedures if the affected area is already numb.

The surgeon may interrupt or alter the transmission of painful stimuli anywhere along the pathways shown in classical form in Fig. 1 2330.

The most successful surgical procedures performed on peripheral nerves are generally those used in treating pain caused by compression, such as carpal tunnel syndrome and ulnar nerve entrapment at the elbow (except, perhaps, in its chronic form). The dorsal root or dorsal root ganglia may be approached by an open operation (usually intradural) or percutaneously using a radiofrequency generator and suitable electrode. These procedures have been most commonly used in the treatment of pain in the trunk (e.g. post-thoracotomy syndrome). The results are variable, and although pain relief is often achieved there may be a high recurrence rate and unpleasant side-effects such as post-rhizotomy dysaesthesia. To obtain effective analgesia in any given dermatome the affected root needs to be cut together with two roots above and two below. This produces a widespread total deafferentation, and the method is therefore unacceptable for use in the limbs.

Selective posterior rhizotomy (rhizidiotomy) makes use of the fact that at the spinal cord/rootlet junction (around 1 mm from the cord) large and small fibres are dissociated anatomically. The small fibres are found on the undersurface of the rootlet and then regroup in the ventrolateral part of the cord to enter the tract of Lissauer. It is possible to cut just the small diameter nociceptive fibres at their entrance to the spinal cord, activating pain inhibitory mechanisms while preserving propioception. This technique has been used successfully for treating pain syndromes affecting the limbs.

Following laminectomy or hemilaminectomy with exposure of the spinal cord, small lesions are produced with radiofrequency generators or lasers at the point of entry of the dorsal roots in to the spinal cord within the intermediolateral sulcus. The operation was designed to treat central pain thought to be due to hyperactivity of cells within the dorsal horn which relay via the spinoreticular and spinothalamic tracts to supraspinal structures. Dorsal root entry zone lesions destroy either the hyperactive neurones or the cells of origin of the spinothalamic and spinoreticular tracts.

Dorsal root entry zone operations were first used for the treatment of brachial plexus avulsion injuries, and produced long-term pain relief in around 70 per cent of patients treated. They have also been used in the treatment of the central pain of paraplegia, post-herpetic neuralgia, atypical facial pain, and pain associated with failed back syndrome and arachnoiditis. They are partially successful in treating paraplegic pain but otherwise of little value in these other conditions. Although few serious post-operative side-effects are reported, there is probably a morbidity rate of at least 10 per cent.

The main pain-carrying tract in man is traditionally considered to be the anterolateral spinothalamic pathway; however, only a small number of these fibres truly reach the thalamus without synapse. Nevertheless, spinothalamic tractotomy (anterolateral cordotomy) has been practised to good effect since the beginning of this century. It may be performed as an open or percutaneous procedure on the opposite side to the painful area and at least four vertebral segments of the cord above the affected dermatomes. The axons of the spinothalamic neurones cross the midline at the levels of the anterior commissures over several segments. Many surgeons prefer to perform all such operations in the high cervical region, where the crossover is certain to have been completed.

The operation should produce relief of pain and loss of temperature sensation in the affected area, with good analgesia. It is most appropriate for unilateral malignant pain and for treating superficial or deep somatic pains rather than visceral pain. Bilateral operations substantially increase the risks of neurological deficits. Provided the patient is fit for surgery and has a life expectancy of more than a few months, this is a good method of relieving pain; it is, however, often overlooked in treatment of the terminally ill. Operative morality in the last group of patients is high (about 10 per cent) and the principal risks are those of respiratory failure, motor weakness, dysaesthesia, and sphincter dysfunction. Most surgeons perform the open operation.

Commissural myelotomy is an alternative to bilateral anterolateral cordotomy and is appropriate for the treatment of malignant pain in both lower limbs or in the perineal region. Pain and temperature fibres decussate in the spinal cord whereas light touch and motor fibres decussate in the lower brain-stem: a midline myelotomy which divides the commissures at an appropriate level can be used to treat bilateral pains of various types. An extensive laminectomy is necessary.

Central myelotomy by stereotaxic means and cordectomy are rarely performed pain procedures. The former requires knowledge of stereotaxic techniques and the latter is rarely appropriate, even in patients with pre-existing complete spinal cord damage.

The most commonly performed procedures are those used in the treatment of trigeminal neuralgia as discussed elsewhere. Attention has been recently to focus on the use of dorsal root entry zone lesions made in the spinal nucleus of the fifth nerve and Lissauer's tract extending from the C2 dorsal root to the level of the obex. This promises to be one of the few successful treatments for chronic facial post-herpetic neuralgia.

The most common pain relieving procedure performed within the brain has been that of stereotaxic thalamotomy. Using a stereotaxic frame applied to the head, and with the guidance of MRI or CT scanning, suitable targets for a radiofrequency generated burn or for the tip of a stimulating electrode can be readily identified. Those targets have traditionally been the centromedian nucleus of the thalamus and the parafascicular nuclei.

Such operations have the disadvantages of being expensive (in the case of neurostimulatory techniques), having potentially serious side-effects such as hemiplegia or speech deficits, and being somewhat unpredictable in producing pain relief. Thalamotomy is considered the most appropriate operation for the relief of pain due to malignancy in the head and neck. The most recent pathophysiological and neurochemical data suggest that treatment of central pains such as thalamic syndrome should be directed towards the more rostral intralaminar nuclei of the thalamus.

Injection of alcohol into the intrasellar region has been used to destroy the pituitary gland in patients with widespread pain from advanced carcinoma of the breast and prostate. Between 50 and 70 per cent of patients may gain complete pain relief following this procedure but endocrine disorders inevitably follow this operation.

Focal stereotaxic lesions within the frontal lobes and their projections which have found more general acceptance include cingulotomy, subcaudate tractotomy, and inferior frontomedial leucotomy. Their use is controversial owing to possible alterations of the psyche of treated patients.

Sympathetic denervation has been used in the treatment of limb, cardiac, and abdominal visceral pain. Open sympathectomy has largely been replaced by the use of oral or intravenous sympatholytic drugs, percutaneous radiofrequency lesions, and chemical procedures. The most common indications are those of causalgia, Sudeck's atrophy, and ischaemic pain in the limbs due to peripheral vascular disease. Cardiac sympathectomy in the treatment of angina has largely been superseded by improved medication and coronary artery bypass grafts, although percutaneous radiofrequency cardiac sympathectomy may still have a place in the treatment of medically intractable pain. Malignant pain affecting the pancreas, liver, gallbladder, and stomach together with painful chronic relapsing pancreatitis has been treated by chemical sympathectomy of the caeliac plexus and splanchnic nerves using an injection of 50 per cent alcohol or phenol.

Modern devices for stimulating any part of the central nervous system to produce pain relief comprise an electrode or set of electrodes applied to either a peripheral nerve, the epidural space overlying the back of the spinal cord, or a chosen target within the brain. The electrodes are linked to a receiver system implanted under the skin and activated by an external radiotransmitter, or are connected directly to a totally programmable implanted activating device. Peripheral nerve and spinal cord stimulators are inserted by a small operation, and are associated with few serious complications. The stereotaxic implantation of a deep brain stimulating system is complex and more prone to unwanted sequelae.

The likely benefits of a peripheral nerve stimulator are assessed by percutaneous stimulation of the nerve prior to implantation. Temporary electrodes are used before a complete spinal cord or deep brain stimulating system is implanted. One of the chief problems with this form of treatment is the expense involved; in addition the life expectancy of some systems may not exceed 5 years. Peripheral nerve stimulators have been used for pain following peripheral nerve damage in the limbs, and they probably act either by blocking nociceptive afferents or by more central inhibition of spinothalamic tract neurones.

Spinal cord stimulation was based on the gate control theory such that stimulation in the region of the dorsal columns would mainly inhibit C fibres—hence the original term dorsal column stimulation. However, there are many ways, both electrical and neurochemical, by which spinal cord stimulation might produce pain relief. At operation an electrode is inserted into the epidural space, either percutaneously or through a small laminectomy. The latter allows more accurate placement of the electrode and it can be secured to the dura. The leads are exteriorized for a period of temporary stimulation. If a satisfactory result is obtained then the device may be fully implanted. Spinal cord stimulation is mainly used in the treatment of syndromes such as the ‘failed back’. Some believe that this is the only form of worthwhile treatment in such cases. It may prove beneficial in patients with peripheral vascular disease, by easing the pain associated with ischaemic limbs.

The anatomical and physiological postulates underlying the choice of targets within the brain to effect pain control are complex. Two main target areas have been chosen. Stimulation of lateral margin of the periaqueductal and periventricular grey matter is thought to affect a pathway running from the mid-brain to the dorsal horn, inhibiting nociceptive neurones. The second target area is the VPM/VPL nuclei of the thalamus, which ultimately inhibit spinothalamic tract neurones. Electrodes, implanted into the brain at these sites using stereotaxic guidance systems, are connected subcutaneously to a suitable stimulating device. The parameters of stimulation may be altered by radiotelemetry. These specialized techniques are probably best carried out in just a few centres in each country, and are most useful in the treatment of pain of central origin.

Opiate drugs injected into the spinal fluid produce analgesia by their direct action on the spinal cord, and opioid receptors have been identified in both the spinal cord and the brain. Chronic administration of opiates and other analgesic substances into the cerebrospinal fluid, by the lumbar route or the intraventricular route, has been used to produce long-term pain relief. The doses employed are a fraction of those required when the same drug is administered orally or intravenously. Powerful opiates such as morphine may cause respiratory depression and urinary retention, and tolerance may develop. These disadvantages may be offset by the reduction in other adverse effects, including sedation, which are associated with administration of opiates by other routes.

Many substances other than opiates have been administered intrathecally to relieve pain: clonidine has been one of the more successful. Most are thought to moderate nociceptive information to neurones in the dorsal horn. The drugs may be delivered by continuous infusion or in carefully controlled bolus doses through a catheter implanted into the ventricle or into the lumbar sac and connected to a reservoir and pump system implanted under the skin of the abdominal wall. This system has been most widely used to alleviate pain in cancer patients who have a reasonable life expectancy. Pain associated with spasticity following spinal injury can effectively be treated with intrathecal baclofen. Disadvantages other than side-effects of the drugs are the cost, risks of infection, and the frequency of pump and catheter failures. These recent advances in neurostimulation and in delivery systems which allow small aliquots of drug to be administered to exact loci within the central nervous system point the way toward the development of more precise and effective methods of pain relief while reducing the incidence of neurological deficit.