The mortality rate of epilepsy surgery is usually quoted to lie between 0.5% and 1.0%. In the Kings/Maudsley series between 1976–2001, there were 451 temporal lobe resections with 2 perioperative deaths (0.44%), due cerebral edema of uncertain cause (1 case) and cerebral hemorrhage 2 weeks after the operation due to anticoagulation for a DVT (1 case).³ In the large series from Montreal, a recent review reported that there were no deaths in 526 operations.⁴ In the current authors’ analysis of 737 cases of temporal lobe surgery in the literature (see Chapter 19), there were two postoperative deaths (wound infection with osteomyelitis)⁵ and six late deaths, three seizure related^6,⁷ and in three cases no details were given.^5,⁸ It should not be forgotten, when considering the mortality of surgery, that the death rate in epilepsy (with active seizures) is two to three times increased in people with epilepsy, and the rate of death of patients on surgical waiting lists for epilepsy surgery is about 1 case per 100 per year.⁹ The death rate of epilepsy after surgery has been found to be reduced, especially if seizures are controlled, in some studies¹⁰ but not in others.¹¹ The risk of death due to surgery is clearly dependent on the underlying etiology. Operations for mesial temporal sclerosis or benign tumors, for instance, have very low mortality, whereas there is a greater risk for operations on vascular lesions and particularly arteriovenous malformations (AVMs). In two large series of surgically treated AVM, although at all cerebral sites and with any presentation, the operative mortality was 11%.^12,¹³

OTHER OPERATIVE MORBIDITY

Infection following epilepsy surgery occurs in most series at a rate below 1%, although is higher where intracranial recordings are carried out. In one series of 122 implantation procedures, operative complications included the need for repeated surgery for additional electrode placement (5.7%); wound infection (2.4%); cerebrospinal fluid leak (1.6%); and subdural hematoma, symptomatic pneumocephalus, bone flap osteomyelitis, and strip electrode fracture requiring operative retrieval (one patient [0.8%] each). There were four cases of transient neurological deficit (3.3%) and no permanent deficit or death associated with invasive monitoring.¹⁴ In Olivier’s series of 560 patients, meningitis was reported once, subdural abscess twice, and scalp infections five times. Hemorrhage along the track of the recording electrodes is another hazard, with a risk usually quoted at 1%.¹⁴

Neurological Complications of Temporal Lobe Surgery

Neurological complications following temporal lobe surgery are well summarized by Polkey (2004) (Table 20-1).¹⁵ A visual field loss—usually an homonymous superior quadrantanopia—is a common sequel of an extensive temporal lobectomy, due to damage to Meyers loop of the geniculocalcarine tract (see Figure 20-1). How common this complication is following more restricted operations is unclear but is certainly underestimated. Manji and Plant assessed 24 patients following temporal lobe surgery for epilepsy using sensitive methods and found a field deficit in 13 (54%).¹⁶ A proportion of these deficits might have been missed on more routine evaluation. In this series, details of the extent of resection were not given, but the surgery was carried out by experienced epilepsy surgeons using orthodox means.

TABLE 20–1 Complication Rates of Epilepsy Surgery Reported at the Palm Desert Conferences of 1987 and 1993

Complication	1987	1993
Transient hemiparesis	0.7%	4%
Permanent hemiparesis	0.7%	2%
III nerve palsy		<1%
Complete hemianopia	0.6%	3%
Speech disturbance	1.4%
Nonneurological mortality	0.47%	<1%

1987 figures from Van Buren 1993; 1993 figures from Pilcher 1993 (cited by Polkey 2004).

Figure 20–1 Tractography of the visual tracts showing the visual pathways visualized in green. Note: This extraordinary image in a patient with full-field stimulation, was produced by Professor Mark Cook in Melbourne using what is probably the most advanced MRI software globally for tractography. I am indebted to Professor Cook for permission to use this image.

Penfield was the first to report a transient hemipareisis in 5% of patients undergoing temporal lobectomy. The rate of hemipareisis depends on etiology and is now lower, between 1 and 3%, in patients following temporal lobe surgery for mesial temporal sclerosis. Hemipareisis is usually due to interference with perforating vessels supplying the internal capsule or the anterior choroidal or even posterior choroidal artery, or in the case of arteriovenous malformations, to hemorrhage or infarction. A 2% rate of mild but permanent hemipareisis due to anterior choroidal artery damage was reported in the recent series by Sindou et al.¹⁷ Diencephalic infarction occasionally occurs with hemianopia, ophthalmoplegia, and speech disturbance in addition to hemipareisis or hemiplegia. A higher risk of hemiparesis is conferred in surgery on tumors or vascular malformations, and also in operations on the insula region, due to the vacular palisade overlying the insula. Clearly, the surgical technique and the experience of the surgical team are also important factors, and the rate of vascular complications is generally higher the more inexperienced the surgeon—a major reason for the current recommendation that epilepsy surgery be carried out in appropriate centers.

Cranial nerve palsy has been reported (in less than 1% of cases). The third nerve is the most vulnerable but fourth and sixth nerve palsies also occur. A facial palsy can occur and is usually transient. Facial pain is another complication, possibly more common, which can be persistent and may be due to the section of superficial nerves during the craniotomy.

Transient dysphasia in a dominant hemisphere temporal resection is not uncommon, especially if grids or strips are inserted for electrocorticography (ECoG) or mapping purposes. The dysphasia is usually maximal 1 to 3 days postoperatively and usually largely resolves within a week, although minor speech disturbance may persist for months. Permanent dysphasia may occur when a dominant temporal lobectomy is extended too far posteriorly or when a lesionectomy is carried out in or under the posterior cortex. The anatomical location of language function in patients with temporal lobe epilepsy can be extremely variable, and dysphasia may occur even if the resection avoids the conventional anatomical locations of language. The risk can be reduced but not eliminated by preoperative language mapping. It is now usual not to extend the resection of a dominant middle or superior temporal gyrus back beyond 3.5 cm (in contrast to Penfield’s original 4.5 cm resection), and this precaution significantly reduces the risk of dysphasia. More posterior or basal resections should be conducted only in units with considerable experience and usually only with preoperative language mapping. Unfortunately, at present, preoperative fMRI is not sufficiently accurate to delineate language areas safely enough, and surgeons should not take the lack of language activation in any particular location on fMRI as a sign that there is no risk of postoperative dysphasia as this location is resected.

A range of other complications has been recorded. Complications include distant hemorrhage in the cerebellum, pneumocephalus, hematoma, meningitis (2% in the series of Sindou et al.¹⁷), acute hydrocephalus requiring shunt insertion (2% in the series of Sindou et al.¹⁷), scalp infections, and wound pain. It has also to be noted that most of the large series report complication rates retrospectively, and most also depend on surgical notes. Almost certainly this will lead to an underestimation of the true complication rate, but the extent of this underestimate is unclear. The published figures, however, should certainly be considered minimum estimates.

Effects of Temporal Lobe Surgery on Memory Function

It is probably not too much of an exaggeration to say that the neural substrates and mechanisms of human memory remain almost totally obscure, despite intense basic and clinical research in the past few decades. Indeed, much of what little has been discovered about the anatomical basis of memory is from operations on the temporal lobes in epilepsy. It has been recognized, at least from the time of Penfield, that the hippocampus is intimately involved in memory functions, but that a unilateral resection is often possible without much disturbance of memory, whereas a bilateral resection carries a significant risk for severe amnesia. Why this is and how the brain encodes or stores memories is, however, otherwise largely speculative—and although Nobel prizes have been won in this area, it has to be admitted that what knowledge there is has little utility from the clinical or practical points of view.

The effects of bilateral hippocampal and temporal resection are most famously recorded in great detail in the case of HM, whose severe permanent antegrade amnesia has been the subject of intensive study.¹⁸ Bilateral resections were then largely abandoned, but in the 1960s to 1980s, most of the work on postoperative patients (after unilateral or bilateral resections) was concerned with psychological theories of memory, which has been of little clinical or practical value. More recent work has taken a more prosaic and pragmatic emphasis on psychometric testing to identify tests that will predict the surgical outcome for memory after unilateral resection.

About one-third of patients currently suffer memory decline following temporal lobe resection. About 15% experience a slight improvement in memory; these patients are more likely to be seizure free, and the improvement is probably due to the removal of the adverse effects of seizures on memory.¹⁹ A number of factors are now generally accepted to be predictive of memory outcome and are useful considerations when counseling patients about the risk of temporal lobe surgery to memory (Table 20-2):

a. Dominant temporal resections are associated with more memory decline than nondominant resections.²⁰^–²² Furthermore, the memory decline is in the area of verbal memory—and this is more noticeable in the everyday life of patients than decline in visual memory.

b. There is some evidence also that patients with a high preoperative IQ do somewhat worse after surgery.²²

c. The effects of a unilateral resection on memory depend a great deal on preoperative memory functions (the better the memory preoperatively, the greater the potential decline after surgery). It is here that lateralized preoperative psychometric assessment is extremely helpful.²³^–²⁷ The less good the verbal memory is preoperatively in a patient with hippocampal epilepsy, the less adverse effect a dominant resection will have. Similarly, the worse the preoperative nonverbal memory, the less adverse effects a nondominant resection will have. Conversely, the better the verbal memory preoperatively, the more likely is noticeable memory decline after dominant surgery and the better the nonverbal memory the more likely is a deficit after nondominant resection. Notwithstanding the preceding, about one in five patients with a nondominant resection also experiences some loss of verbal learning and memory, so the laterality effects are not absolute or entirely reliable.

TABLE 20–2 Factors Influencing Memory Outcome After Unilateral Temporal Lobe Surgery

Age	Poorer outcome after age of 50 years
Gender	Slightly more risk to memory for males
Preoperative IQ	Patients with a higher IQ do less well.
Preoperative memory deficits	The better the preoperative memory, the more likely are postoperative deficits, and the likely there is to be a negative effect on functional memory outcomes.
Dominant vs. nondominant resections	Dominant resections have a generally greater negative effect on functional memory outcome than nondominant resections.
Unilateral hippocampal atrophy	The more atrophic the resected hippocampus is, the less effect will the surgery have on memory.
Bilateral hippocampal damage	Unilateral resections in the presence of bilateral hippocampal damage carry a significant risk of severe amnesia.
Cortical dysplasia	Resections for cortical dysplasia have a poorer memory outcome than resections for MTS.
Extent of surgical resection	Large dominant resections probably have a poorer memory outcome than smaller dominant resections, but the evidence is conflicting.
Seizure outcome after surgery	There is a better outcome in those patients who attain seizure freedom after surgery.
Affect, attention, psychiatric status	These can greatly affect memory function and are often more important functionally than even moderate changes in psychometric test results.

Furthermore, if there is evidence of contralateral preoperative memory dysfunction prior to a proposed resection (i.e., nondominant dysfunction in a proposed dominant resection or dominant dysfunction in a proposed nondominant resection), the greater is the risk of severe amnesia postoperatively. Indeed, so great is this risk, that in our own unit, the finding of significant contralateral memory dysfunction is taken as a major contraindication to surgery.

e. In temporal lobe epilepsy due to other pathologies, similar considerations apply, but the effects are less predictable. Nevertheless, the more normal the hippocampus appears on MRI scanning, the greater is the likelihood that resection will result in memory decline.

f. If there is evidence of bilateral hippocampal damage on MRI (in mesial temporal sclerosis or any other pathology), memory functions after unilateral surgery may be seriously affected (and occasionally an amnesic syndrome of that of HM will be produced). Thus, the volumetric assessment of both hippocampi is a useful part of presurgical assessment. However, some workers have reported good outcome following unilateral temporal lobe resection in the presence of bilateral hippocampal atrophy.²⁹

g. Age also seems important, and dominant resections in patients over 50 carry a significant risk to memory functioning.³⁰ The age of onset of the epilepsy is also important—and the earlier the epilepsy develops, the less is the risk to memory, perhaps due to greater plasticity of normal functioning in patients with early onset epilepsy.³¹

h. There is a suggestion that male patients suffer more memory decline than females after epilepsy surgery, but the differences are not great.³² Factors such as the duration of the epilepsy seem to have no predictive value for memory outcome.

i. The extent of surgical resection is also a factor. The degree of resection of lateral temporal cortex may be important, but the effects of memory are difficult to predict. The operation of selective amygdalohippocampectomy was conceived to try to limit memory loss by restricting the lateral resection. However, there is no clear consensus as to whether this operation really improves outcome, and improvements in memory outcome need to be balanced against the greater operative morbidity of the selective operation.^23,^33,³⁴

j. Patients with cortical dysplasia seem to be at greater risk of memory decline following unilateral temporal resections for epilepsy than patients with MTS.²³

Predictive models would be helpful in this regard, but although obviously possible to construct, would require standardized methods of assessment and multicenter international collaboration, and none are currently available. Furthermore, there remains in all cases a significant degree of uncertainty about memory outcome. On occasions, patients who have favorable preoperative factors suffer significant memory deficits postsurgery and vice versa.³⁵ Preoperative prediction of memory outcome is not yet totally reliable science, and for this reason patients should be warned that memory is at some risk, even if the preoperative assessment is favorable.

A final complicating factor is the poor correlation between memory changes on psychometric testing and the change in memory function noticed by the patient.³⁶ The reasons for this disjunction are not clear, but memory is greatly influenced by affective factors and the attentional states of individuals, and this may have more functional impact than, say, relatively minor changes in psychometric parameters. The lack of sensitivity of the current psychometric testing instruments to day-to-day experience of memory is, however, a factor that greatly limits their usefulness.

Psychiatric and Other Cognitive Adverse Effects of Temporal Lobe Surgery

The psychiatric adverse outcomes after temporal lobe surgery are common and among the most important but least studied. These psychiatric disturbances can be occasionally florid and severe, although commonly they are more subtle. They usually develop in the first year after surgery, although it has to be said that the very long–term psychiatric outcome has not been the subject of any systematic study, and instances of late psychiatric breakdown are not uncommonly encountered in clinical practice. The disturbances can take different forms and may have a variety of pathogenic mechanisms.

Psychosis after Epilepsy Surgery

When anterior temporal lobectomy became accepted as a therapy for temporal lobe epilepsy in the 1950s, there was also the hope that the operation would control the psychosis of epilepsy. Indeed, the psychosis of epilepsy was considered a positive indication for surgery. This situation rapidly changed when it became apparent that preoperative psychosis could be dramatically worsened by surgery, and the presence of a fixed preoperative psychosis is now considered a strong contraindication to surgical therapy.

In a summary of previous literature in 1992, a mean of 7.6% of patients (range 3.8% to 35.7% in different series) developed de novo psychosis after temporal lobe surgery.³⁷ More recent surveys have in fact shown lower rates, perhaps due to more cautious selection, with de novo psychosis occurring in about 1 to 4% of patients.³⁸^–⁴¹ At the Maudsley hospital, 11 patients from a recent series of 320 cases of temporal lobectomy for medically intractable epilepsy were found to develop a de novo schizophrenia-like psychosis postoperatively. This developed typically in the first year following the operation and was not related to postoperative-seizure outcome. Patients who became psychotic were more likely to have preoperative bilateral electroencephalogram (EEG) abnormalities, pathologies other than mesial temporal sclerosis, and a smaller amygdala on the unoperated side.⁴¹

If psychosis is present preoperatively, there is a risk of an exacerbation of the psychosis. In an early series from the Maudsley Hospital, Taylor reported that 16 of 100 patients were psychotic preoperatively, of whom only four were improved after the operation.⁴² Leinonen et al. (1994) reported that five of 57 (8.8%) adult patients developed postoperative psychosis after temporal lobectomy.⁴³ Two (3.5%) had preoperative psychosis, and three (5.3%) de novo psychoses. However, in more recent experience, less postoperative psychiatric morbidity has been reported. In a series of 74 patients from Denmark, 11 were psychotic preoperative, and after the operation one patient became nonpsychotic, five were unchanged, and five were improved.⁴⁴ A series of five patients with psychosis from the Montreal Neurological Institute were reported who underwent temporal lobectomy with seizure remission and without any change in their psychosis.⁴⁵ These patients, though, were highly selected and able fully to consent to the surgery, which suggests that the psychosis was mild. Certainly, the authors have observed, at an anecdotal level, florid postoperative psychoses developing in patients with relatively mild preoperative psychotic states. It is also commonly stated that both the preoperative and postoperative psychoses of epilepsy respond relatively well to psychotropic therapy, and this is also our anecdotal experience.

Affective Disorders after Temporal Lobe Epilepsy Surgery

Depression is common in temporal lobe epilepsy—occurring at lifetime frequencies of up to 30% in population-based studies. After temporal lobe surgery, there is a significant risk of depression, and occasionally this is severe and life threatening. Wrench et al. found 26% of 44 patients to be depressed 1 month after surgery and 30% at 3 months.⁴⁶ The disturbances of mood were significantly related to adjustment difficulties. Devinski et al.⁴⁷ and Devinsky/Altshuler et al.⁴⁸ report figures around 10% at 3 months. At 2 years, moderate or severe depression was reported by 17.6% of patients who were not seizure free. It is general experience that rates are higher in patients who continue to have seizures. It has been reported that depression is more common after right temporal resections,⁴⁹ but the current consensus now is that there are no striking differences between right and left temporal resections. The most important factor predicting the occurrence of depression is a history of preoperative depression. The suicide rate is elevated after temporal lobectomy, at least partly due to postoperative depression.⁵⁰

The precipitation of anxiety after temporal lobe surgery is a real and often unrecognized problem. Wrench et al. found that 42% of patients have an anxiety state 1 month postoperatively and 24% 3 months after surgery.⁴⁶ Patients who are seizure-free have somewhat lower levels of anxiety, but the relationship to seizure outcome does not appear to be a strong one.⁴⁷ The affective outcome of surgery can be a significant determinant of the patient’s satisfaction with surgery.⁵¹

It is likely that the affective responses of patients following surgery have many causes, including, for instance, postoperative adjustment, coping skills, loss of illness role, relationship changes, and fear of seizures. However, one has the strong impression also that there is a significant biological component, and until more is understood of the affective changes following temporal lobe surgery, these important complications will remain uncertain and unpredictable.

Other Psychiatric and Psychological Adverse Outcomes of Temporal Lobe Epilepsy Surgery

Large resections of brain tissue will inevitably have psychological consequences. The effects of temporal lobe resections on memory have been the focus of attention, and to a lesser extent studies of depression and psychoses, but it is remarkable how little systematic study there has been of other effects. A range of potential organic psychosyndromes has been almost completely ignored, and where information exists it is often anecdotal at best. There are probably several reasons for this highly unsatisfactory situation (Table 20-3), and we consider that this is an area that should be given a high priority for future studies.

TABLE 20–3 The Reasons for the Lack of Study of Psychiatric or Cognitive Consequences of Epilepsy Surgery

Deficiencies in psychological measurement tools—this is a major reason

An extraordinary lack of preoperative vs. postoperative comparisons in individual patients and the small numbers of patients possible in prospective studies

The lack of comparative control data in nonoperated patients with epilepsy

The focus on the physiological identification of the epileptogenic zone without the ability to assess cognitive function with the same accuracy

The inherent bias introduced by the natural optimism of teams carrying out epilepsy surgery

The difficulty of long-term study

The concept that certain areas of the brain are inneloquent (“silent cortex”) is no longer widely held among psychologists and cognitive neurologists, but still dominates thinking in the field of epilepsy surgery and particularly in relation to neocortical resections. Frontal lobe resections for frontal lobe epilepsy is a good example, where the aphorism of Rasmussen that the larger the resection the better the seizure outcome still results in enormous resections, with scant regard of psychological consequences. The most common resection in epilepsy is, of course, of temporal lobe cortex, and this is an area which by no measure can be considered ineloquent; yet there is little psychological study of the consequences on cognitive function other than crude measures of memory. There are complicating factors, not least the effects of epilepsy on normal functioning, but the absence of study is a shocking deficiency. Epilepsy surgery is not usually considered to be psychosurgery, and yet the psychological consequences in some cases result in marked impairment.

All surgery (and indeed all medical treatment) is a trade-off between risk and benefit. The benefits of seizure control are undoubtedly, for most patients, worth the risk of some adverse cognitive effects. This is a decision for individual patients, but it is a decision that must be informed—and the lack of information makes counseling difficult. The following are examples, by no means comprehensive, of areas in which knowledge is seriously deficient and yet that at an anecdotal level are common clinical observations.

EFFECTS OF TEMPORAL LOBECTOMY ON PSYCHOSEXUAL FUNCTION

Sexual function was first noted to be affected in temporal lobe epilepsy by Gastaut and Collomb, and hyposexuality is a common complaint among patients.⁵² Following temporal lobe surgery, however, there can be marked changes in sexual function. At an anecdotal level, a common change is diminution of libido and sometimes also male erectile dysfunction. More commonly described in the literature, but less common in practice, is the occurrence of sudden hypersexuality following temporal lobe surgery, which can on occasions result in serious social disturbance.⁵³ An extreme example is the Klüver-Bucy syndrome, which is a consequence of bilateral temporal lobectomy in animals and in humans, comprising of hypersexuality, visual agnosia, strong oral tendencies, overeating, and hypermetamorphosis (defined by Klüver and Bucy as the “excessive tendency to take notice of and to attend and react to every visual stimulus,” probably equating to environmental dependency syndrome in current terminology).⁵⁴ Cases are also occasionally recorded following unilateral temporal lobectomy.