It is inevitable that some patients will be lost to follow-up in longitudinal studies. However a number of these series are limited by systematic bias in the sample studied. Long-term neuropsychological follow-up was only a routine part of the postoperative follow-up in one surgical series (the Swedish studies). In the German study, patients were offered financial compensation to return for a follow-up neuropsychological assessment and comprised 70 % of the surgical series [10], while in the UK study, only 25 % of eligible postoperative candidates underwent a repeat neuropsychological assessment 5 years or more after their operation [8]. These participants were patients who remained under the care of the specialist hospital in London, a national center for epilepsy surgery, and therefore were more likely to have ongoing seizures than those who had been discharged back to their local neurology services. Only two of the studies used medically treated epilepsy controls [10, 11] and only the Swedish series employed a healthy control group [6, 7, 9].

Two long-term series, those from the Netherlands [5] and Sweden [6, 7, 9] report longitudinal data with neuropsychological assessments conducted at specified time intervals following surgery. In these series, the long-term outcome was assessed at 6 years and 10 years respectively. In the other series, the long-term assessments were conducted at different time points for the candidates with the greatest range being between 5 and 17 years following the surgery [8]. Combining data from a wide range of follow-up points since surgery into a single measure of “long-term follow-up” may obscure patterns in the trajectory of change over time.

All of the studies reported data from patients who had undergone both right and left temporal lobe resections. In the Swedish studies, the patients were dichotomized by language dominance rather than a right vs. left distinction, that is, dominant vs. non -dominant resection. In all but one of the studies, the surgical candidates had undergone a standard temporal lobe resection. The outcome data in the Dutch series is based upon a series of patients who underwent a selective amygdalo-hippocampectomy [5].

The majority of studies reported changes at a group level, that is, type of surgery by time of assessment. The German, UK, and Swedish studies also reported data using reliable change indices for individual patients.

The majority of patients in these series had stable memory function at the long-term follow-up assessment, although as Baxendale et al. [8] report, many were functioning below the 15th percentile on the test norms preoperatively with little capacity for further decline, particularly when defined by rigorous reliable change criteria.

There are conflicting findings regarding the relationship between progressive memory decline and continuing seizures in the long-term outcome literature. Generally, studies that have employed group level analyses have failed to find a significant relationship [5, 6] while those that have looked at individual trajectories suggest that ongoing seizures after surgery are associated with a progressive deterioration in memory skills [8, 10]. These findings suggest that there is a subset of surgical patients in whom ongoing seizures are associated with progressive memory impairments. These patterns are obscured in group analyses. Surgery on the left or language dominant side is consistently identified across the long-term outcome studies as a risk factor for progressive deterioration in verbal memory decline, particularly in those with ongoing seizures following surgery and good preoperative function. When compared to a medically treated control group, surgery appears to accelerate memory decline and exceeds it when performed on the left, or if seizures continue postoperatively [10].

Some improvements in verbal memory function have been reported at long-term follow-up in patients who have undergone right or non-dominant temporal lobe resections [7, 10] although Alpherts et al. [5] suggest that these improvements may not always be sustained in the long-term. Practice effects are rarely considered or adequately controlled for in longitudinal studies. Even when reliable change indices are employed to determine change over time, they are rarely based on the same time intervals as those used in the longitudinal study.

The long-term data suggests that the memory functions of most patients who undergo temporal lobe surgery for epilepsy, stabilize within the first few years after the operation and that they subsequently follow normal age-related decline gradients [13]. The extent of the stepwise postoperative decline, and its subsequent acceleration, will determine when the patient will eventually develop clinically significant memory deficits. (See Fig. 5.1.) The majority of patients who have been studied to date had surgery in their late twenties or early thirties. Even at their long-term follow-up they have a mean age (across the studies) in their forties. Normal age-related cognitive declines are noticed in healthy individuals toward the end of the fifth decade and accelerate thereon [14]. The effects of this normal age-related deterioration in function in the latter life of postoperative patients remain a critical unknown. Indeed, little is known about the trajectories of cognitive function in people with longstanding epilepsy in their seventh decade and beyond.

Many surgical candidates already function below the second percentile on standardized memory tests prior to surgery [8]. Postoperative deteriorations in memory function in either the short- or longer term will not be detected in group studies or in studies that utilize reliable change indices to measure individual change, as these patients are already functioning at, or very close to, the floor of standardized tests. It is therefore likely that outcome studies have underestimated the true extent of postoperative memory deterioration. There is a need to develop new measures that are sensitive to change in this group; these may well rely on more behaviorally based indices of function.

We know that postoperative memory capacity is a function of the integrity of the structures removed during surgery against the reserve of those left in situ [15]. A number of studies have utilized biomarkers to quantify this reserve and capacity in multivariate models to predict postoperative memory function [16–23]. These have yet to be employed in long-term outcome studies, which will also need to incorporate postoperative factors that indicate the likely focus of the postoperative seizures. The focus of postoperative seizures is likely to be a critical factor in determining postoperative memory patterns. Postoperative seizures with contralateral involvement may be associated with a progressive decline in memory function, whereas those with a residual focus in the operated hemisphere may experience less deterioration.

Despite over half a century of successful epilepsy surgery, knowledge of the long-term neuropsychological consequences of extra temporal resections remains limited to clinical experience. The long-term data that is available has focused primarily on memory and intellectual functions following temporal lobe surgery, very little is known about the long-term impact of the surgery on language and executive functions. The special considerations that have been highlighted in the outcome literature for those with a low IQ [24–26] or older patients [27, 28] have yet to be explored in the long-term cognitive literature.

Generating large samples in longitudinal follow-up studies is challenging. Presurgical neuropsychological protocols may change over time, together with the clinical characteristics of patients who are offered surgery. Cumulative numbers of patients become lost to follow-up over time. These are not usually random losses. Patients who do well postoperatively, that is, those who become seizure free, and who eventually cease medication following surgery are less likely to keep in contact with specialist epilepsy services and may be more likely to be lost to follow-up than patients who continue to experience seizures.

In order to ensure that surgical candidates are able to make as informed a choice as possible with respect to epilepsy surgery, we need to provide a longitudinal perspective on cognitive change. At present the evidence base for this advice is thin and limited to temporal lobe resections. In order to make a clinically meaningful contribution to the long-term literature, future studies should meet a number of minimum requirements. These are outlined in Table 5.2.