Seizure frequency is associated with age at menopause in women with epilepsy

Patients and methods.

Women aged 45 years and older from four urban epilepsy center clinics in 2001 through 2002 were identified by using the computerized appointment schedules of 12 consecutive weekly epilepsy clinics. Subjects were briefly interviewed and medical charts were also reviewed. Subjects for whom an estimation or documentation of seizure frequency or other comprehensive information was not available were excluded. Information was obtained regarding seizure frequency, seizure treatment, age at onset of epilepsy, handedness, age at last menses, other severe medical illnesses, height and weight at the time of menopause to determine body mass index (BMI), cigarette use, number of children and pregnancies, age of subject at last childbirth, and history of treatment for depression. Age at last menses (in years) is used as the age at menopause herein. Women who were not at least 1 year without menses or who had cessation of menses due to hysterectomy were excluded. In order to be listed as a seizure treatment, at least 6 months of exposure to an AED was required. However, any exposure to an antidepressant medication for the treatment of depression was included, because it is unclear whether treatment for depression or the presence of severe depression itself is associated with early onset of menopause.9

The assessment of seizure frequency over the course of epilepsy was initially made by reviewing the chart. This information was then discussed with the subject to refine the information for accuracy and to get an idea of which seizure group would be most accurate. If it was not clear from chart review, subjects were asked if they had more or less than 20 seizures in their lifetime. Subjects were then categorized into one of three broad seizure frequency groups: Group 1 subjects had less than 20 seizures in their lifetime, Group 2 subjects had more than 20 seizures in their lifetime but less than one seizure per month on average, and Group 3 subjects averaged one or more seizures per month for most of their illness. All seizure types were included together within a subject, without separation of seizure types for this categorization. With the use of these broad categories, an attempt was made to group subjects into distinct seizure frequency categories, if not seizure severity or seizure type categories. This study was approved by the Human Studies Committee of the Weill Cornell Medical Center.

For the statistical analysis, a one-way analysis of variance was performed along with post hoc analysis (adjusted for multiple comparisons, Bonferroni approach) to calculate comparisons for seizure frequency groups and age at menopause, which is a normally distributed variable. The statistical analysis comparing seizure frequency groups regarding non-normally distributed variables (BMI, duration of epilepsy, age at onset of epilepsy, pack-years of cigarette use, number of children, age at bearing youngest child, years of exposure to each AED, years of exposure to enzyme-inducing AED) was performed using Kruskal-Wallis test for independent samples. Bivariate correlations were also calculated for continuous (Pearson) and ranked (Spearman) variables. Categorical variables (handedness, race, history of treatment for depression) were assessed between seizure groups using χ² analysis. Multivariable analysis was performed by linear regression to evaluate epilepsy-related variables. All p values are two-sided with significance evaluated at the 0.05 alpha level; 95% confidence intervals were calculated to assess the precision of the estimates.

Results.

Sixty-eight subjects were included. Twelve additional subjects were considered but not included because of inability to obtain accurate information in eight and unwillingness to provide information in four. Their mean age at the time of the survey was 53.8 years (SD ± 4.9, range 46 to 67 years). The mean age at last menses (menopause) was 47.8 years (SD ± 4.1, range 37 to 59 years) and the median age at menopause was 47 years. Fifteen subjects were in Group 1 (fewer than 20 seizures in lifetime), 25 subjects were in Group 2 (>20 seizures in lifetime but <1 seizure/month), and 28 subjects were in Group 3 (>1 seizure/month). The age at last menses was 49.9 years in Group 1, 47.7 years in Group 2, and 46.7 in Group 3. The difference between groups was approximately 3 years (p = 0.042). When comparing the three seizure frequency groups, post hoc tests reveal that seizure frequency Groups 1 and 3 are different regarding age at last menses (p = 0.037 using adjusted Bonferroni analysis) (table 1). A negative correlation was present between seizure group and age at menopause (p = 0.022, r = −0.277), with the higher seizure frequency groups associating with a lower age at menopause.

The number of years since menopause was 5.6 ± 5.6 in Group 1, 7.1 ± 5.4 in Group 2, and 5.6 ± 3.7 in Group 3. The difference in the number of years since menopause was not significant between groups.

Exact age at onset of seizures was available for 63 of 68 subjects. Fifty-five of 63 women developed epilepsy before menopause. The mean age at menopause in this subset was 47.6 years (SD ± 3.9, range 37 to 56 years) and the median age at menopause was 47 years. The age at menopause by seizure frequency group was 49.7 years in Group 1, 47.6 years in Group 2, and 46.8 years in Group 3, and this difference did not reach significance. Eight subjects had age at onset of epilepsy after menopause, and two in the same year as menopause. Six of these were in Group 1 with a mean age at menopause of 50.3 years, one was in Group 2 with age at menopause of 49 years, and one was in Group 3 with an age at menopause of 43 years.

Forty-nine subjects had onset of epilepsy before or at age 42 years, the age after which decline in ovarian function may be expected, used by Klein et al.8 in their definition of premature menopause. In this group, age at last menses was 49.7 years in Group 1, 47.6 years in Group 2, and 46.8 years in Group 3; this difference also did not reach significance. Fourteen subjects had onset of epilepsy after age 42 years. In this group, age at last menses was 50.0 years in Group 1, 48.3 years in Group 2, and 47.0 years in Group 3; this difference also did not reach significance. Four subjects became menopausal before age 42; one was in Group 2 and three were in Group 3.

The evaluation of the relationship between AED and age at menopause was restricted only to those subjects who had the onset of epilepsy before menopause. The total number of AED ever used, the number of enzyme-inducing AED ever used, and the number of non-enzyme-inducing AED ever used did not correlate with age at menopause. Additionally, the number of years of any AED and the total number of years of exposure to an enzyme-inducing AED did not correlate with the age at menopause. This variable of total number of years of exposure to an enzyme-inducing AED was derived in order to assess any differential effect of patient-years of exposure on age at menopause. The number of years of use of any AED and the number of patient-years of enzyme-inducing AED exposure was not different between seizure frequency groups. However, the number of years of exposure to carbamazepine approached significance at p = 0.068. The mean number of years of carbamazepine use in Group 1 was 1.4 years (SD ± 3.4), in Group 2 was 4.8 years (SD ± 7.2), and in Group 3 was 8.9 years (SD ± 10.4). A relevant observation is that only five subjects out of the entire study population had never used an enzyme-inducing AED; four were in Group 1 and one was in Group 3 (this subject had multiple drug allergies).

The age at onset of epilepsy positively correlated with the age at menopause (p = 0.022, r = 0.288). The age at menarche (p = 0.031, r = 0.272) and the age of the subject when she bore her last child also correlated with the age at menopause (p = 0.023, r = 0.343). Factors that did not correlate with the age at menopause were years of cigarette use, number of children, or BMI.

No difference between seizure frequency groups was found regarding race, handedness, or history of treatment for depression.

Age at onset of epilepsy was different in seizure frequency groups, with a mean age at onset of 41.7 years in Group 1, 25.4 years in Group 2, and 21.8 years in Group 3 (p = 0.001) (table 2). Post hoc analysis with adjustment for multiple comparisons showed that the differences were between Group 1 and Groups 2 (p = 0.006) and 3 (p = 0.001); Groups 2 and 3 were not different in age at onset. Duration of epilepsy was also different in groups as well (p = 0.008) (see table 2). Post hoc analysis with adjustment for multiple comparisons showed that the differences were between Group 1 and Groups 2 (p = 0.015) and 3 (p = 0.009); Groups 2 and 3 were not different in duration of epilepsy. Number of children born to each subject was not significantly different between groups. Other variables that did not differ between seizure groups were BMI, pack-years of cigarette smoking, and age at bearing youngest child.

The foregoing data indicate that the age at menopause in this population may be affected by several epilepsy-related factors, specifically the seizure frequency, age at onset of epilepsy, and a related factor, the duration of epilepsy. However, this data set has the peculiar feature of subjects categorized as Group 1, with few lifetime seizures, also having a later onset of epilepsy, and therefore shorter duration of epilepsy. This may be a bias produced by ascertaining subjects from epilepsy referral centers, where patients tend to either have severe epilepsy or recent onset of seizures.

Using the seizure frequency groups to approximate the number of seizures actually experienced by the subject may be skewed based on a short or long duration of epilepsy. Therefore, in order to further evaluate the effect of seizures on menopausal age, a method of estimating the number of seizures experienced by each subject was derived and then compared to menopausal age. This was performed by multiplying the years of duration of epilepsy by an estimated number of seizures per year. In Group 1, for example, who had <20 seizures in their lifetime, the mean duration of epilepsy was 14.4 years. Group 1 subjects may have had approximately one seizure per year; this was multiplied by the years of seizure duration, to estimate the number of lifetime seizures. For Group 2, the duration of epilepsy was multiplied by five (the estimated number of seizures per year), and for Group 3, the duration was multiplied by 18 (the estimated number of seizures per year).

Using this method, the range of lifetime seizures derived was from 1 to 882. Lifetime seizures were then grouped proportionately by the estimation of annual seizure frequency. This resulted in three groups of 1 to 49 lifetime seizures, 50 to 245 lifetime seizures, and 246 to 882 lifetime seizures. Age at menopause was different in these three lifetime seizure groups as well (p = 0.048), with the difference between the lowest and the highest lifetime seizure groups (table 3). There was a negative correlation between age at menopause and lifetime seizure groups as well (p = 0.014, r = −0.310).

When evaluating only those subjects with onset of epilepsy before menopause, the negative correlation between age at menopause and lifetime seizure groups (of groups 1 to 49, 50 to 245, or 246 to 882 seizures) (p = 0.036, r = −0.283) was maintained.

Four variables were used in the multivariate analysis: 1) lifetime number of seizures, which based upon previously described groups was inversely correlated with age at menopause; 2) years of carbamazepine use; 3) years of valproate use; and 4) years of phenytoin use. The influence of multiple variables revealed that the lifetime seizure groups were associated with age at menopause (p = 0.041), whereas years of carbamazepine (p = 0.583), years of valproate (p = 0.433), and years of phenytoin (p = 0.421) were not associated.

Discussion.

These data suggest that the age at natural menopause in women with epilepsy is lower than might be expected from statistics for the general population. The mean age at menopause was 47.6 years for women with onset of epilepsy before menopause, and this changed only slightly to 47.8 years when the few women with onset of epilepsy after menopause were included in the analysis. This age can be compared to a US population mean age at natural menopause of 51.3 years.10 For example, the mean age at menopause in a group of 5960 Australian twins was 51 years (95% CI 50 to 51).11 The median age at menopause in this epilepsy group was 47 years, compared to an expected median age of 51.4 years in the general American population.12

A caveat regarding the conclusions of this study is that the estimation of seizure frequency is inexact. The number of seizures in the lifetime of the subjects is an estimate based on reported seizure frequency and duration of epilepsy. This calculation is obviously biased by subject reporting, but should provide separation between those subjects who had many seizures over a long duration from those who had rare seizures. This evaluation suggests that lifetime number of seizures, and perhaps seizure frequency and duration of epilepsy, and the age at menopause are correlated, but the exact relationship is unclear. The magnitude of the effect appears to be about 3 years; women with a history of many seizures may undergo menopause at least 3 years earlier than those women who have had fewer seizures. Endocrine effects of subclinical seizure activity occurring before the onset of observable seizures may account for the early onset of menopause even in women with epilepsy diagnosed after menopause, but this remains speculative.

The analyses of the effect of seizures on age at menopause were limited by the inclusion of many women with frequent seizures and epilepsy of long duration, but a relative paucity of women with rare seizures and epilepsy onset before menopause as a comparator group. This may be a function of the epilepsy population for which the investigators have ready access and it is clear that the centers in this study serve intractable epilepsy patients to a greater proportion than would a general neurology clinic. Additionally, the study population does not include women who have discontinued AED after having prolonged seizure freedom. We have no reason to suspect a referral bias toward the inclusion of women having reproductive complaints, such as early onset of menopause.

The use of AED may impact the timing of menopause owing to their effect on endogenous hormonal metabolism by increasing levels of sex hormone binding globulin,13 but an independent effect of AED or enzyme-inducing AED cannot be discerned in this study population. This is because so few subjects did not take enzyme-inducing AED for at least several years. Further, the number of years of use of AED was not significant between seizure groups. Therefore, AED use may have uniformly affected the age at menopause in this population, but an independent influence cannot be determined.

In this fairly small sample, factors that clearly affect the age at menopause in the general population, such as smoking, BMI, or treated depression, were not obviously influential on the age at menopause. The most important environmental factor affecting the age at menopause, the number of pregnancies, was not significantly different between seizure frequency groups. Further, the reliability of reported age at menopause is important to consider in that the years since menopause significantly affect the reliability of the report.14 However, the number of years since menopause was very similar between seizure frequency groups; thus variation in reliability is not a factor in this data set.

There is little in the literature regarding neurologic disease and menopausal onset. In 42 women with Down syndrome, mean menopausal age was 44.7 years and correlated with age at onset of dementia.15 Left-handed women are reported to have an earlier age at menopause than right-handed women.16 It is generally accepted that medical illnesses (outside of medically treated depression) do not affect menopausal age, except for circumstances in which the ovaries are directly radiated or are exposed to chemotherapeutic agents.

The potential for seizures to produce premature failure of the reproductive system has been documented in animal models of epilepsy.17,18⇓ In female rats, right amygdala kindling caused polycystic ovaries and premature aging of the hypothalamopituitary axis, leading to anovulation and sustained, abnormal estrogen circulation.17 Effects on the reproductive system of male rats were also produced by repeated electroshock seizures; specifically, this resulted in chronic elevations in serum testosterone and estrogen levels as well as increased gonadal weight.18

Therefore, clinical and basic research suggests that seizures as well as chronic epilepsy can produce early cessation of normal reproductive cyclicity. The exact mechanism by which this might occur is not clear, especially because female rodents differ from women in that they undergo centrally mediated, rather than gonadally mediated, reproductive failure. From this, loss of reproductive cyclicity in rats cannot be directly extrapolated to a model for menopause. Nevertheless, reasonable hypotheses for the mechanism by which seizures produce early menopause in humans can be postulated, on the basis of available information. Natural menopause results primarily from a depletion of ovarian follicles that occurs throughout the lifespan. Without viable ovarian follicles, menstrual cyclicity ceases, and follicular stimulating hormone (FSH) rises and remains elevated. Women with epilepsy have chronic abnormalities of central hormone secretion (LH and FSH). Inadequate levels of LH and FSH produce anovulatory menstrual cycles and eventually secondary amenorrhea. Therefore, women with epilepsy may be experiencing early amenorrhea due to abnormal patterns of gonadotropin secretion, leading to earlier loss of reproductive hormonal cyclicity as the pool of ovarian follicles is depleted. This could be a consequence of frequent seizures occurring over a protracted time period. Seizures may act by disrupting normal limbic modulation of hypothalamic and pituitary function or may adversely affect neurally mediated trophic effects on the gonads.19 Frequent seizures, particularly when combined with long-term enzyme-inducing AED, may accelerate the rate of follicular atresia, resulting in premature depletion of the pool of primordial follicles.20

The potential effects on the general health of women with epilepsy resulting from earlier menopause include an earlier onset of bone loss and less cardioprotective lipid metabolism. Early menopause, from age 40 to 44 years, has also been associated with increased cancer-related mortality.21 For some women, earlier than expected ovarian failure will affect their childbearing plans. The association between epilepsy and early menopause deserves further exploration, both in terms of the possible health consequences for the affected patients and the underlying pathologic mechanisms.