Abstract
Background: Pooled data from New Drug Applications (NDAs) submitted to the U.S. Food and Drug Administration (FDA) provide an opportunity to study the incidence of and risk factors for rare events.
Objective: To examine the incidence and causes of mortality in patients with epilepsy participating in clinical trials of antiepileptic drugs (AEDs); and to examine the incidence of and risk factors for sudden unexplained death in such patients.
Methods: Exposure data and death narratives were obtained from the NDAs of five recently reviewed AEDs. Deaths were classified as sudden unexplained, accidental, or other cause using the 1993 Burroughs–Wellcome expert panel criteria, and mortality rates were calculated for each category. Add-on trials were analyzed separately from monotherapy initiation trials.
Results: Among 9,144 patients in the add-on trial database, the all-cause and sudden unexplained mortality rates were 9.1 and 3.8 deaths per 1,000 person-years (124 and 52 deaths in 13,617.1 person-years of drug exposure). Sixty-five percent of all deaths were related to the underlying epilepsy. Of the examined risk factors, only age was associated with the incidence of sudden unexplained death. Among 1,293 patients in the monotherapy initiation trials, the all-cause and sudden unexplained mortality rates were 7.1 and 0 deaths per 1,000 person-years (7 and 0 deaths in 982.5 person-years of drug exposure).
Conclusions: A large proportion of the deaths in the add-on cohort was attributable to epilepsy-related causes. Mortality due to sudden death in the add-on cohort falls into the high end of the reported range for patients with epilepsy. The difference in mortality due to sudden death between the add-on and monotherapy initiation cohorts suggests that disease severity is the primary determining factor for risk of sudden unexplained death.
As early as 1910,1 sudden death was linked with epilepsy. More recently the phenomenon has been referred to as sudden unexpected death in epilepsy (SUDEP). The rate of SUDEP has been examined in a variety of populations of patients with epilepsy including the general population,2,3⇓ epilepsy clinics,4 a residential program for children with epilepsy,5 candidates for epilepsy surgery,6 and participants in development programs for a new antiepileptic drug (AED)7 or device.8
During the 1990s, as the number of potential AEDs under development dramatically increased, high rates of SUDEP were observed in cohorts of patients in AED development programs. Concern arose because these rates of SUDEP exceeded the highest published population-based rates of SUDEP. As SUDEP rates have now been shown to be similar across all recent drug development programs, it appears most likely that the disease, rather than the drugs, is responsible for the apparent increase in SUDEP.
In AED development, the “add-on” trial is the study design most frequently submitted to the US Food and Drug Administration (FDA) to demonstrate the efficacy of new AEDs. In fact, until some recent development programs included monotherapy initiation studies that enrolled patients with newly diagnosed epilepsy, almost all data in drug development programs in the 1990s had been derived from add-on studies. The add-on study design, by definition, enrolls patients who do not have adequate control of their epilepsy and who are already treated with at least one AED.9 Patients must also have disease that conforms to specific inclusion/exclusion criteria, in effect creating a cohort of patients with a well-defined, relatively narrow range of illness severity. Following enrollment, patients continue their current therapeutic regimen and are randomized to add either the study drug or a control (usually placebo).
Using data pooled predominantly from add-on studies of several new AEDs approved in the 1990s, we examined the effects of age, gender, duration of epilepsy, and number of concomitant AEDs on SUDEP rates. This pooled database allows for evaluation of patient risk factors for SUDEP, the study of which has been limited until now by the small size of the study populations.
Although monotherapy initiation trials in newly diagnosed patients with epilepsy account for only a small proportion of recent AED development programs, we have also pooled the data from these studies to look at SUDEP rates and risk factors. However, because this subpopulation of patients with epilepsy likely differs from those entering add-on trials, we analyzed this database separately.
Methods.
Data procurement and characteristics of patients included in the drug development programs.
Sponsors of four FDA-approved AEDs (called Drugs A, B, C, D) provided data from randomized controlled add-on trials, open extension trials, and open safety and efficacy trials from each drug’s development program. The data consisted of patient-specific information on the extent of drug exposure during the study, as well as selected baseline information on demographic characteristics, duration of epilepsy, and the use of concomitant AEDs. Sponsors also provided a narrative summary of the circumstances surrounding each death. Data from the development program of Drug A and from a marketing application currently under review (Drug E) was used in the analysis of monotherapy initiation trials.
In general, patients included in the add-on studies (Drugs A, B, C, D) had epilepsy that was uncontrolled by standard antiepileptic therapies, experiencing an average of four seizures per month at baseline. The development programs enrolled patients with partial seizures, generalized tonic-clonic seizures, and Lennox–Gastaut syndrome. Patients with structural CNS abnormalities (e.g., brain tumors) were excluded from these trials. We also excluded healthy volunteers, patients who were exposed to the AED for nonepilepsy indications, and patients enrolled in compassionate use protocols from the analysis.
SUDEP definition.
In 1993, Burroughs–Wellcome convened a panel of experts to evaluate the occurrence of SUDEP in the lamotrigine development program. The panel considered the following six criteria essential in the diagnosis of definite or probable SUDEP7:
-
1. The victim had epilepsy, defined as recurrent unprovoked seizures.
-
2. The victim died unexpectedly while in a reasonable state of health.
-
3. If observed, the death occurred within minutes.
-
4. The death occurred during normal activities and benign circumstances, thus excluding accidental deaths (ACC) such as drownings, motor vehicle accidents (where the patient was the driver), and falls with immediate death due to trauma.
-
5. An obvious medical cause of death was not found. An autopsy was necessary to establish a definite SUDEP; if an autopsy was not performed and an obvious medical cause of death was also not established, the case was considered a probable SUDEP.
-
6. The death did not occur in the setting of status epilepticus.
We adopted the same classification. A death occurring during or after a seizure could be considered a SUDEP, but a death in the setting of status epilepticus was not considered to be a SUDEP. We considered the presence of substantial aspiration of gastric contents on autopsy as an obvious medical cause of death, therefore such a case was not considered a SUDEP.
In a few cases, when a death narrative provided a summary statement without supporting documentation, we took a pragmatic approach. Statements such as “the patient aspirated,” were assumed credible and classified as non-SUDEP. A more uncertain statement, such as “the patient may have aspirated,” in the absence of supporting documentation, was not considered a strong enough pronouncement to reclassify the death from probable SUDEP to non-SUDEP.
Rate calculation.
“ALL” refers to the all-cause mortality rate. “SUDEP + ACC” includes what would be considered definite or probable SUDEP, with the addition of accidental deaths. “SUDEP” rates include only the definite or probable SUDEP cases. The numerator represents all deaths of interest, while the denominator is the sum total of the person-time exposure to the AEDs included in the cohort (i.e., drugs A, B, C, and D for the add-on cohort; drugs A, E, and E comparators for the monotherapy initiation cohort).
We used the Poisson regression module of HiroSoft’s (Seattle, WA) Epicure software to calculate mortality rates and 95% CI. Covariate information was not available for every patient in the database. Drug D’s database lacked the duration of epilepsy history for its entire cohort, and Drug C’s database lacked the number of concomitant AEDs for its entire cohort.
Results.
Add-on trials.
In the pooled database, there were 9,144 patients, accounting for 13,617.1 person-years of drug exposure. The median length of epilepsy history in the cohort was 19 years, and the median number of concomitant AEDs at enrollment was two. Table 1 displays the cause-specific mortality for the add-on cohort. When SUDEP and accidental deaths were combined, they accounted for more than half of the deaths in the group. When other seizure-related causes were added, including aspiration and status epilepticus, 65% of deaths in the cohort were potentially disease-related.
Cause-specific mortality in the add-on trial cohort
Of 124 total deaths, we classified 52 as SUDEP and 20 as accidental. Table 2 summarizes the all-cause mortality, SUDEP + ACC, and SUDEP rates for each of the four AEDs. Because the populations and rates were similar across the four drugs, we pooled the data for the four drugs for further analysis.
Exposure and mortality experience in add-on trials
Table 3 shows the mortality rates by demographic and disease characteristics. There was no substantial difference in SUDEP rate between men and women. The SUDEP rate tended to increase with age, and while the CI overlapped across all strata, the test for trend was significant. There was no relationship between longer epilepsy history and SUDEP rate, even after length of epilepsy history was controlled for age. There was a trend toward a higher SUDEP rate in patients with two or more concomitant AEDs, as compared with those patients taking only one concomitant AED.
All, SUDEP + ACC, and SUDEP rates by demographic and disease characteristics
Monotherapy initiation trials.
Among seven deaths occurring in the monotherapy initiation cohort, two were potentially epilepsy-related (status epilepticus [n = 1] and traffic accident [n = 1]), accounting for 29% of the deaths. The other five deaths were due to cardiovascular disease (n = 2), cancer (n = 1), suicide (n = 1), and political violence (n = 1).
Person-time experience and mortality rates in the monotherapy initiation trials of drugs A and E are summarized in table 4. Altogether, the seven deaths in 982.3 person-years of experience accounted for an all-cause mortality rate of 7.1 cases per 1,000 person years. No cases of SUDEP occurred. Assuming that there was no true difference in SUDEP rates between the patients in the add-on trials and those in the monotherapy initiation trials, the probability of observing no SUDEP in 982.3 person-years would be 3%.
Exposure and mortality experience in monotherapy initiation trials
By definition, the monotherapy initiation trial participants could not have any history of epilepsy and were not taking any concomitant AEDs. When mortality rates were examined by demographic characteristics, the all-cause mortality rates increased with increasing age (0 in the 1 to 14 age group; up to 21.7 cases per 1,000 person-years in the 55 to 72 age group). When calculated by gender, the all-cause mortality rate in men exceeded that of women five-fold (11.2 versus 2.2 cases per person-year).
Discussion.
Our study, encompassing nearly 14,000 person-years of experience on recently approved AEDs, is the largest examination of mortality in the refractory patient population participating in add-on trials. We found that a large proportion of the mortality in the add-on cohort was attributable to epilepsy, with 65% of deaths due to SUDEP, accidents, aspiration, or status epilepticus. Forty-two percent of all deaths were attributed to SUDEP alone. Prior studies of mortality in similar epilepsy populations have also identified a similar pattern of cause-specific mortality.4,5⇓ In contrast, only 29% of the monotherapy initiation cohort deaths were potentially epilepsy-related.
Our pooled database allowed an examination of the incidence of and risk factors for sudden unexplained death in patients with epilepsy participating in clinical trials of AEDs. The SUDEP rate in the add-on cohort was 3.8 per 1,000 person-years, which falls into the high end of the range for patients with epilepsy reported in the medical literature.8 In general, SUDEP rates have varied from 0.35 to 0.5 cases per 1,000 person-years in general population-based studies1,2⇓, to 3.5 to 4.5 cases per 1,000 person-years in clinical trials of drugs7 or devices.8
Notably, there was a marked difference in SUDEP rates between the add-on and monotherapy initiation cohort (3.8 versus 0 cases per 1,000 person-years). If disease severity is the determining factor in the occurrence of SUDEP, then a group of patients with newly diagnosed epilepsy, who would be expected to represent a range of disease severity, would likely have a lower overall incidence than that of the cohort of patients with refractory chronic epilepsy.
One weakness in our analysis of the mortality across the AED development programs was the absence of primary documentation describing the circumstances of death. We had to make our determination of SUDEP status based on narrative summaries provided by the sponsors. Even with primary documentation, the determination of SUDEP status can be controversial, and the dilemmas involved are well described in the description of SUDEP incidence in the lamotrigine development program.7 As such, the possibility of misclassification always exists and may have been enhanced by the absence of primary records. Additionally, our study was not able to assess the possibility that the AED itself increases the risk of SUDEP. However, it is reassuring that similar SUDEP rates were observed in each of the four development programs.
Among the risk factors for SUDEP examined in our study, only age was associated with SUDEP incidence. Because of strict inclusion/exclusion criteria, patients with active cardiovascular disease are generally excluded from participating in clinical trials of AEDs. The relatively healthy cohort of older patients in this clinical trial population would not be expected to experience as much cardiovascular morbidity as their counterparts in the general population. Despite the expectation of lower cardiovascular morbidity in this age group, the SUDEP rate still increased with age. In contrast, a population-based study from Olmsted County, MN, found that the incidence of SUDEP increased from birth up to 44 years old, decreased by half in patients 45 to 64 years old, and finally decreased to zero in patients older than age 65.2 Similarly, Annegers et al.8 hypothesized that all or at least the majority of sudden unexplained deaths in elderly patients with epilepsy are accounted for by cardiovascular deaths, and SUDEP per se occurs infrequently, if at all, in this age group.8 It is possible that the different conclusion on the relationship of age and SUDEP incidence between our study and those in the literature stems from a difference in the method by which a death was designated as “unexpected.” For example, if an elderly patient who died suddenly had evidence of coronary artery disease on autopsy, one group of investigators might consider death to be expected and attribute it to myocardial infarction or coronary atherosclerosis. However, our approach in such a patient would have been to characterize the death as a SUDEP unless there was evidence of an acute myocardial infarction (e.g., fresh coronary thrombosis).
In a large case-control study of risk factors for SUDEP, the relative risk of SUDEP increased with use of more than one AED.10 Other studies have also shown a relationship between number of AEDs and incidence of SUDEP. In a cohort of more than 5,000 patients followed at three epilepsy centers, the relative risk of SUDEP in patients taking three or more AEDs, compared with two or fewer AEDs, was 3.4 (CI 1.1 to 10.2).11 In contrast, our study did not show a strong relationship between number of concomitant AEDs and increasing incidence of SUDEP. One study in the medical literature also did not find an increased risk of SUDEP associated with number of concurrent AEDs in users of two or more AEDs.12 Its study population was similar to our add-on cohort in that it was limited to patients with refractory epilepsy. The investigators of that study suggested that there may have been too few patients to demonstrate a correlation between SUDEP risk and number of concomitant AEDs in users of at least two AEDs. Our add-on cohort is considerably larger and also failed to demonstrate such a relationship. The lack of a relationship between number of AEDs and SUDEP risk in patients with refractory epilepsy suggests that in those studies that did find a relationship3,10,11⇓⇓, the number of AEDs may be a measure of the severity of epilepsy.
Our study is the first to consider duration of epilepsy as a risk factor for SUDEP. Across the three development programs for which data were available, duration of epilepsy history was not associated with the rate of SUDEP. It may be that disease severity, regardless of disease duration, is the most important predictor of SUDEP incidence. Whereas other studies have not addressed the relationship between duration of disease and the rate of SUDEP directly, populations with chronic epilepsy have been shown to have high rates of SUDEP.4,5⇓ A large case-control study of risk factors for SUDEP examined the relative risk of SUDEP with regard to age at epilepsy onset and found a significant elevation in relative risk in patients who were 5 years old or younger at onset when compared with those who had an older age at onset.10 Although a younger age at onset may indicate a longer history of epilepsy, this variable is not a direct surrogate measure of length of epilepsy history.
Our examination of mortality in recent AED development programs confirms that epilepsy-related events remain a common cause of death in this population. With regard to SUDEP, our data suggests that severity of epilepsy is the primary determinant of SUDEP risk and likely accounts for the difference in SUDEP rates between the add-on and monotherapy initiation cohorts. Length of epilepsy history, number of concomitant AEDs, and gender did not influence the SUDEP rate in this chronic epilepsy population. SUDEP rates did increase with age, though, suggesting that there is still a risk of sudden unexpected death due to epilepsy in older patients. In our study, NDA databases provided a rich source of risk factor information. We hope that in the future, additional information collected as part of clinical trials, such as concomitant medications and concurrent illnesses, can be examined as potential risk factors for SUDEP.
Disclaimer.
This article expresses the views of the authors, and does not necessarily represent those of the US Food and Drug Administration.
Acknowledgments
Acknowledgment
The authors thank Jacqueline Ware, PharmD, for her assistance in obtaining the NDA safety data.
- Received March 10, 2000.
- Accepted November 8, 2000.
Disputes & Debates: Rapid online correspondence
NOTE: All authors' disclosures must be entered and current in our database before comments can be posted. Enter and update disclosures at http://submit.neurology.org. Exception: replies to comments concerning an article you originally authored do not require updated disclosures.
- Stay timely. Submit only on articles published within the last 8 weeks.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- 200 words maximum.
- 5 references maximum. Reference 1 must be the article on which you are commenting.
- 5 authors maximum. Exception: replies can include all original authors of the article.
- Submitted comments are subject to editing and editor review prior to posting.