FULL-LENGTH ORIGINAL RESEARCH
Mortality after temporal lobe epilepsy surgery
Address correspondence to Charles E. Polkey, Department of Neurosurgery, King’s College Hospital, Denmark Hill, London SE5 9RS, U.K. E-mail: firstname.lastname@example.org
Purpose: To report mortality, after a longer interval, in a cohort of patients with drug-resistant epilepsy treated by temporal lobe surgery between 1975 and 1995. A previous audit of these patients ending December 1, 1997 observed a standardized mortality ratio (SMR) of 4.5.
Methods: We analyzed mortality in a cohort of 306 patients with temporal lobe epilepsy (TLE) who underwent temporal lobe resections between December 1, 1975 and December 1, 1995. Deaths occurring after December 1,1997 and until December 1, 2009 were evaluated. Medical records, death certificates, postmortem examination reports, coroner officer’s reports, and coroner’s inquest reports were sought, and causes of death were ascertained. Sudden unexpected death in epilepsy (SUDEP) cases were identified.
Key Findings: In 3,569 person-years of follow-up 19 deaths occurred, [SMR 2.00, 95% confidence interval (CI) 1.27–3.13], 14 men (SMR 2.01, 95% CI 1.19–3.39) and 5 women (SMR 1.68, 95% CI 0.70–4.03). On analysis of subgroups, SMRs were significantly elevated in patients with mesial temporal sclerosis (MTS) (SMR 2.50, 95% CI 1.38–4.51), men with MTS (SMR 3.12, 95% CI 1.56–6.25), men with nonspecific lesions (SMR 2.68, 95% CI 1.00–7.09), and right-sided resections in MTS (SMR 3.33, 95% CI 1.39–8.00). During follow-up, six SUDEP cases were observed with a rate of 1/595 person-years.
Significance: In this cohort, the risk for premature death in patients undergoing TLE surgery decreased over time but remained above the standard population. Men had a slightly higher risk than women, as did right-sided resections in MTS, confirming this observation in the original cohort. Although lower, the risk of SUDEP remained. Without up-to-date information on seizure outcome, we were unable to directly relate this to mortality.
Patients with epilepsy have a two- to threefold increase in mortality compared to the general population (Zielinski, 1974; Hauser et al., 1980; Cockerell et al., 1994, 1997). Depending on various parameters such as gender, age, epilepsy syndrome, or age at onset of epilepsy, the overall standardized mortality ratio (SMR) can range from no increased risk to an SMR of 4 and higher, this being highest during the first years following diagnosis (Hauser et al., 1980; Cockerell et al., 1994, 1997; Lhatoo et al., 2001). Patients with remote symptomatic epilepsies and ongoing seizure activity are especially at risk: SMRs of 3.7–4.3 and 4.7–5.1, respectively (Nashef et al., 1995; Cockerell et al., 1997; Sperling et al., 1999; Lhatoo et al., 2001). In this subgroup, sudden unexpected death in epilepsy (SUDEP) remains the most common category of death. In patients referred for epilepsy surgery, the incidence of SUDEP ranges between 6.3 and 9.3/1,000 person-years (Tomson et al., 2008).
In patients with refractory epilepsy, the effect of surgery on mortality and the risk of premature death has been investigated using different methodologies. The data available from some of these studies suggest that those with good seizure outcome, that is, with no seizures or nondisabling seizures [Engel class I (Engel et al., 1993)] have a lower risk of premature death. Studies comparing mortality in patients who underwent epilepsy surgery, to those who were evaluated for but did not qualify for surgery, showed a higher risk for premature death in the unoperated patients (Hennessy et al., 1999; Sperling et al., 1999; Salanova et al., 2002; Nilsson et al., 2003; Ryvlin & Kahane, 2003; Bell et al., 2010).
In a cohort of 299 of 305 patients for whom survival status was available, and who underwent temporal lobe surgery for epilepsy between 1975 and 1995, an overall SMR of 4.5 [95% confidence interval (CI) 3.2–6.6] was reported (Hennessy et al., 1999). The present audit investigated late mortality in the same cohort.
The cohort consists of 306 patients with temporal lobe epilepsy (TLE), who underwent epilepsy surgery at the Maudsley Hospital between December 1, 1975 and December 1, 1995. Patients were identified through a database kept by one of the authors, who carried out the operations and conducted annual follow-up for the first 5 years after surgery. Analysis of mortality from surgery until December 1, 1997 has been reported (Hennessy et al., 1999). The original cohort of 305 patients did not include 21 patients, included here, who underwent surgery during the same time period. Excluding the 20 who died during the first period, this cohort thus includes 306 patients. This audit evaluated deaths occurring after December 1, 1997 and until December 1, 2009. All patients had been flagged previously by the Office of National Statistics (ONS), and we were notified of any deaths. Where possible, medical records, death certificates, postmortem examination reports, coroner’s officers’ reports, and coroner’s inquest reports, were obtained. The cause of death was established including whether death was epilepsy related. Cases were classified as SUDEP according to the Nashef (1997) definition. Person-years of follow-up were calculated for each patient in 5-year age bands, from the December 1, 1997 to the December 1, 2009, or to the date of death, if earlier. Age- and sex-specific death rates in the general population were derived from published mortality statistics from the ONS for England and Wales aggregated from 1998 to 2008. Deaths were expressed as rate per person-years of follow-up and SMR was calculated. SMR is the ratio of the number of observed deaths in the cohort to the number of expected deaths in the general population during the follow-up period, standardized for age and sex. SMRs were calculated for the whole cohort, and also for subgroups based on sex, side of surgery, and pathology of the resected specimen. Deaths were assumed to occur independently and randomly. All statistical analysis was performed in Stata 11 (StataCorp LP, College Station, TX, U.S.A.).
These 306 patients had a total follow-up of 3,569 patient-years (average 11.7 years per person; range 2.3–12). One hundred fifty-nine patients were male (1,836 person-years) and 147 were female (1,733 person-years). The average age at surgery was 25.8 years (range 3.1–58.7) and at the end of the study period 46.7 years (range 20–78.1, mean 45.8). One hundred forty-five patients (1,689 person-years) had resections on the right side and 161 (1,885 person-years) on the left. Age distribution was comparable in men and women. All patients: average age 46.7 years (range 20–78.1, mean 45.8); male patients: 46.7 years (range 20–78.1, mean 45.8); and female patients: 46.5 (range 21.6–74.6, mean 45.8).
Cases were grouped into three categories according to the pathology of the resected specimen: mesial temporal sclerosis (MTS), nonspecific pathologies (NS), and other lesions (e.g., low-grade glioneuronal tumors). There were 152 (1,781 person-years) patients with MTS, 80 (935 person-years) male and 72 (863 person-years) female, of whom 57 (675 person-years) had right-sided resections and 95 (1,122 person-years) had left-sided resections. There were 46 subjects (531 person-years) with NS lesions, 27 (307 person-years) male and 19 (230 person-years) female, 31 (357 person-years) of whom had right-sided resections and 15 (174 person-years) of whom had left. Of the remaining 108 patients (1,267 person-years) with other lesions, 52 (604 person-years) were male and 56 (667 person-years) were female; 57 (675 person-years) had a right-sided lesion and 51 (604 person-years) had a left-sided lesion.
Seizure outcome in the cohort
Seizure outcome is based on the Engel classification system (Engel et al., 1993). Information on seizure outcome was available for 291 patients (1,885.19 person-years), comprising 149 men (791.92 person-years), and 142 women (781.56 person-years) for an average of 6.5 (0.2–23.1) years after surgery as shown in Table 1.
Table 1. Details of seizure outcome with Engel classification (Engel et al., 1993)a
|Overall||291||107 (36.8)||153 (52.6)||138 (47.4)|
|Men||149||55 (36.9)||73 (49.0)||76 (51.0)|
|Women||142||42 (36.6)||80 (56.3)||62 (43.7)|
|MTS||158||56 (35.4)||85 (53.8)||73 (46.2)|
|Right side||128||46 (35.9)||68 (53.1)||60 (46.9)|
|Left side||155||57 (36.8)||80 (51.6)||75 (48.4)|
During the audit period, 19 deaths occurred with only 10 expected, giving an SMR of 2.00 (95% CI 1.27–3.13). The average age at death was 49.5 years (range 32.2–76.2). Fourteen patients were male, with an SMR of 2.01 (95% CI 1.19–3.39), and five were female, with an SMR of 1.68 (95% CI 0.70–4.03). An SMR of 2.01 (95% CI 1.12–3.64) was observed with operations on the right side and an SMR of 1.91 (95% CI 0.96–3.82) with operations on the left side.
In the subgroup of patients with MTS, 11 deaths occurred, with only four expected, leading to an SMR of 2.50 (95% CI 1.38–4.51). Male patients had an SMR of 3.12 (95% CI 1.56–6.25) and female patients had an SMR of 1.62 (95% CI 0.52–5.03). SMR was 3.33 (95% CI 1.39–8.00) in right-sided and 2.94 (95% CI 0.92–4.54) in left-sided resections.
For patients with NS pathology, five deaths occurred, with an SMR of 2.23 (95% CI 0.93–5.35), 2.68 (95% CI 1.00–7.13) in male patients and 1.15 (95% CI 0.16–8.19) in female patients. SMR was 2.09 (95% CI 0.78–5.57) in right-sided and 3.11 (95% CI 0.44–22.07) in left-sided resections.
In the remaining 109 patients with other pathologies, three deaths occurred, with an SMR of 0.95 (95% CI 0.31–2.95): 1.10 (95% CI 0.27–4.39) in male and 0.70 (95% CI 0.10–5.00) in female patients. SMR was 0.96 (95% CI 0.24–3.84) in right-sided and 0.79 (95% CI 0.12–5.57) in left-sided resections. SMRs for all subgroups are listed in Table 2.
Table 2. SMR values for pathologies depending on sex and resection side
Epilepsy-related deaths including SUDEP
Of the 19 deaths during the follow-up period, six were considered epilepsy related, all had postmortem examination, and all were classified as definite SUDEP cases. Of these, five were male patients, two had MTS, two had NS pathology, and two had other lesions. The cause of death stated on the death certificate in all these patients was epilepsy or SUDEP. The deaths all occurred at home and were unwitnessed (Table 3). Mean age at death in this group was 41.3 years (range 31–57) and postoperative survival time was 14.3 years (range 7–20). The overall SUDEP rate was 1/595 person-years. On last follow-up (average 6.6 years after surgery), but 5–10 years before death, two of six SUDEP patients only were seizure free. Of the other four, one had less severe seizures once a month, but a bitten tongue when found dead, suggesting a preceding tonic–clonic seizure. One patient had fewer seizures (frequency of about one per month), one patient continued to have one complex partial seizure per month as before surgery, and one patient relapsed with two to three complex partial seizures per month after 2 years of being seizure free after surgery.
Table 3. Summary of SUDEP cases
|M/35||R/MTS||IIIa||48||13 years (4)||1a epilepsy |
1b right temporal lobectomy
|M/24||L/MTS||Ia||37||13 years (7.2)||1a epilepsy|
|M/27||R/NS||IVb||57||20 years (13)||1a SUDEP|
|M/15||L/NS||Ic||31||16 years (8.5)||1a SUDEP|
|M/22||R/other lesions||IIIa||39||17 years (7)||1a epilepsy |
2 previous surgery to the right temporal lobe
|F/29||L/other lesions||IVa||36||7 years (7)||1a SUDEP|
Other causes of death
Thirteen deaths from other causes occurred. These included pulmonary edema (1), pneumonia (2), pulmonary embolism (2), cancer (2), ischemic heart disease (1), cerebral hemorrhage (1), and brain tumor (1). In one case the cause of death remained unascertained, but suicide was suspected.
In this extended audit, we evaluated mortality in a cohort of 306 patients who underwent epilepsy surgery for TLE between 1975 and 1995 during the period between December 1, 1997 and December 1, 2009. In 3,569 person-years of follow-up, 19 deaths occurred. The overall SMR of 2.00 (95% CI 1.27–3.13) is low compared to other operative series of patients with intractable epilepsy where SMRs of 2.5–5.27 were reported (Hennessy et al., 1999; Sperling et al.,1999; Salanova et al., 2002; Nilsson et al.,2003), although mortality remains higher than in the general population. SMR was significantly higher in male patients at 2.01 (95% CI 1.19–3.39), whereas in female patients the increased SMR of 1.68 (95% CI 0.70–4.03) did not reach statistical significance. The difference between genders was also observed in MTS and NS patients. The difference in mortality between male and female patients was not due to a difference in age distribution, which was almost identical between sexes.
Subgroup analysis by pathologic groups showed that MTS patients (1,781 person-years) had the highest mortality: an SMR of 2.50. Although patients with NS pathology (531 person-years) had an elevated SMR of 2.23, which did not reach levels of statistical significance, this group was the smallest and thus with reduced power. The lower SMR of 0.95 observed in the group with other lesions (1,267 person-years) may reflect their better seizure control. Further subgroup analysis revealed significantly elevated SMR in MTS patients with right-side operation at 3.33 (95% CI 1.39–8.00).
This mortality audit is an extension of the study that included largely the same patient cohort published in 1999 (Hennessy et al., 1999). That study identified 20 deaths in 2,729 person-years, three due to operative complications. The remaining 17 deaths may be compared to the 19 in 3,569 person-years in the present audit. The overall death rate in the Hennessy study was 1/136 or 1/161 person-years, excluding three postoperative deaths. In the present audit with extended follow-up, the death rate was 1/198 person-years. As expected, the mean age at death was higher in the present study (average 49.5 years) as the analysis was carried out at a later point after the surgery than the previous one by Henessey et al. A reduction in overall SMR from 4.5 in the previous study to 2.00 in this study was observed. A reduction in SMR is seen in all subgroups, with the increased SMR remaining significant only in specific subgroups. Similarly the incidence of SUDEP decreased from 1/455 to 1/595 person-years. A lower overall SMR would be expected in an older cohort given the increase in background mortality from all causes in the general population. The reduced overall death rate observed, however, as well as reduced SUDEP rates, suggest that the reduction in SMR is at least in part due to a true decrease in epilepsy-related mortality. The higher mortality in patients with MTS and right-sided resections was also reported earlier by Hennessy et al., an observation not confirmed in other cohorts. We are unable to offer an explanation for this.
Comparison of SMR to other previous studies
Jensen described reduced mortality in patients who survived unilateral temporal lobe resection in 1975. She observed “… an excess mortality compared to the general population, but a marked reduction compared to the estimated preoperative rate” (Jensen, 1975). More recent estimates of SMR in patients after TLE surgery are 2.5–5.27, which translates to a death rate of 1/132 to 1/206 in operated patients (Sperling et al., 1999; Hennessy et al., 1999; Nilsson et al., 2003; Salanova et al., 2002). The highest overall risk reported for patients with TLE was an SMR of 5.27, (Salanova et al., 2002), and Nilsson et al. reported similar results in 596 patients (2,455 person-years) where an SMR of 4.9 (95% CI 2.7–8.3) was found. This number was significantly higher in women (SMR 10.1; 95% CI 4.8–18.5) than men (SMR 2.2; 95% CI 0.6–5.5), a finding we have not observed. In Nilsson’s study, in those who underwent temporal lobe resection, the subgroup comparable to our cohort, an SMR of 4.0 (95% CI 1.6–8.2) was reported. The SMR was not significantly elevated when left- and right-sided operations were analyzed separately (left: SMR 3.6, 95% CI 0.7–10.5; right: SMR 3.3, 95% CI 0.7–9.7) (Nilsson et al., 2003). An overview of results from studies on mortality after epilepsy surgery is given in Table 4.
Studies suggest that seizure outcome is a strong factor for premature death. Sperling et al. described 393 patients following epilepsy surgery; 194 experienced a recurrence in seizures, whereas 199 remained seizure free. Deaths occurred only in the group with recurrent seizures, resulting in a highly significant difference in survival in favor of the seizure-free patients (Sperling et al., 1999). Similar results were reported by Salanova et al., where 148 of 215 operative patients were seizure free and 67 experienced a relapse in seizure activity. There were three deaths among the seizure-free patients, not significantly above the expected number, whereas eight deaths in the nonseizure-free group resulted in an SMR of 7.4 (Salanova et al., 2002). Nilsson’s group, however, did not confirm significant differences between seizure free (SMR 3.8, 95% CI 1.3–9.0) and non–seizure-free patients (SMR 4.3, 95% CI 1.6–9.4) in their cohort of 596 patients during the first 2 years following surgery in the 483 patients where outcome information was available (Nilsson et al., 2003). In the most recent study on mortality in epilepsy surgery, outcome was available for 515 patients. Mortality analysis revealed a 4.8 times higher risk of dying for those who were not seizure free compared to seizure-free patients (Engel class I) (Bell et al., 2010). Information on seizure outcome was limited in this audit, not allowing direct comparisons to other studies. However, more than one-half of the patients had a good outcome with no seizures or only nondisabling seizures, an average of 6.5 years after surgery. Seizure control has the potential to reduce morbidity and mortality in patients with epilepsy.
A SUDEP rate of 1/595 is lower than other previously described rates in epilepsy surgery cohorts, where rates varied between 1/133 and 1/455 (Hennessy et al., 1999; Sperling et al., 1999; Salanova et al., 2002; Nilsson et al., 2003). However, this audit reports only on late mortality and does not include the first years following epilepsy surgery where a higher SMR was found (Hennessy et al., 1999). Of the six epilepsy-related deaths observed, all were classified as SUDEP, two had MTS, two had other lesions, and two had nonspecific findings; the number is too small for subgroup analysis. Even with this later follow-up, SUDEP cases still occurred. Of note is that only two of the SUDEP cases had been previously recorded as seizure free. Without recent data on seizure outcome, we are unable to directly relate reduced mortality to freedom from seizures.
In this extended audit of mortality after TLE surgery, we observed that mortality risk in general, as well as the risk of SUDEP, decreased significantly compared to the earlier years following epilepsy surgery. Nevertheless, even at extended follow-up, SMRs remained significantly raised, albeit to a limited extent, above the general population, an effect that is more pronounced in men than in women, and significantly higher in patients with a right-sided resection than in those with left-sided surgery.
This work was partly supported by Epilepsy Bereaved through the SUDEP Research Initiative with King’s College London.
None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.