Introduction
Epilepsy is a common neurological disorder, affecting about 50 million people worldwide [1]. Despite many available therapeutic options, one third of patients continue to have seizures despite trials of at least two antiseizure medications [2]. In people with medication-resistant epilepsy, seizures are unpredictable but commonly occur in regular cycles [3].
Catamenial seizure patterns are defined as increased seizure rates during certain phases of the menstrual cycle (perimenstrual (C1), peri-ovulatory (C2), or luteal (C3) in anovulatory cycles). Catamenial patterns have been described in up to one third of women with epilepsy (WWE), with some literature demonstrating an even higher proportion [4]. Progesterone is generally thought to be anti-convulsant, and estrogen proconvulsant [5].
Administration of progesterone in rodent models produced a significant decrease in seizures [6], while in similar models, estrogen has been shown to have proconvulsant effects [7,8]. Finally, animal studies suggest the C1 perimenstrual pattern may be in part due to the withdrawal of the progesterone metabolite allopregnanolone, which is a potent positive allosteric GABA-A receptor modulator neuroactive steroid [9].
The C1 pattern of catamenial epilepsy corresponds with the increase in the estrogen/progesterone ratio in the perimenstrual period as the serum progesterone and allopregnanolone levels decline rapidly prior to menstruation [10]. When people become pregnant, there are linear increases in both estrogen and progesterone (and allopregnanolone) that otherwise would not occur [11]. It has been hypothesized that these alterations during pregnancy may lead to improved seizure control in women with catamenial epilepsy patterns, who are presumably more sensitive to sex steroid hormonal changes [12].
However, limited and conflicting information exists regarding if these women have different seizure patterns in pregnancy. Cagnetti and colleagues reported that WWE with catamenial patterns were more likely to have improved seizures during pregnancy [12]. Theoretically, this could be attributed to increased sensitivity to changing sex steroid hormone and neuroactive steroid concentrations.
They reported that pre-pregnancy seizure freedom was a reliable indicator for continued seizure-freedom during pregnancy irrespective of catamenial pattern [12], consistent with other studies [13,14]. However, in the Maternal Outcomes and Neuro-developmental Effects of Anti-epileptic Drugs (MONEAD) cohort, a history of catamenial patterns was not associated with seizure improvement during pregnancy [14]. Collection of high-quality prospective seizure and menstrual data from people with epilepsy is exceptionally challenging.
The Women in Epilepsy: Pregnancy Outcomes and Deliveries (WEPOD) study was designed with the primary aims of comparing pregnancy rates between WWE and healthy women [15]. The study demonstrated that WWE generally have a similar rates of achieving pregnancy compared to their peers without epilepsy [15]. This cohort was examined further regarding catamenial epilepsy, and it was determined that there was no difference in likelihood of demonstrating a catamenial pattern between those with a reported history of a pattern and those with no reported history [16].
Furthermore, more than 50 % of women fit into some type of catamenial pattern, most being the C1 pattern, with substantial overlap between the C1 and C2 patterns [16]. In our analysis, we wanted to examine this cohort for the change in their seizure patterns during pregnancy.
We utilized data from this well-characterized cohort of women with epilepsy with prospective daily tracking of seizures and menses via an electronic customized diary app to evaluate how seizure patterns immediately before pregnancy can inform pregnancy seizure control. We hypothesized that patients with catamenial patterns would be more likely to have seizure improvement in pregnancy.
Material and methods
Women with the diagnosis of epilepsy, ages 18-40 years, and seeking pregnancy were enrolled in a longitudinal, prospective cohort study, to compare fertility and birth outcomes to age-matched healthy controls [15,17]. Once enrolled, participants used a daily diary customized app for menstrual and seizure tracking from preconception through delivery. For this analysis, we included participants who had seizures in the preconception period, had at least one menstrual cycle of tracking in the preconception period to define catamenial patterns, and who became pregnant and tracked seizures during at least one day of pregnancy.
Participants were excluded if they chose to discontinue the study, if they did not become pregnant, if their diary data was inadequate (defined as <80 % of the days tracked), or if they had <1 month of seizure data tracked prior to conception.
Pre-conception prospective diary data was used to determine if participants met criteria for any of the three catamenial patterns, by Herzog criteria [18]. Presence of catamenial pattern was defined by a twofold or greater increase in daily seizure frequency during different phases of the menstrual cycle corresponding to 3 types of catamenial patters: C1, C2, or C1 and C2 [16]. The categories were defined as follows where day 1 was the start of menses: C1 perimenstrual as an increase in days −3 through +3 and C2 peri-ovulatory as an increase in days +10 through −13.
Anovulatory cycles were included as the C3 luteal pattern.
Irrespective of catamenial pattern, we calculated each participant’s average daily seizure frequency (ADSF) during the pre-conception period. When WWE became pregnant, we calculated the ADSF during pregnancy from conception through delivery, if they completed pregnancy. We calculated the percent change in ADSF during pregnancy compared to pre-conception.
If the percent change in seizure frequency during pregnancy was less than or equal to 0 %, the WWE’s seizures were improved or stable during pregnancy.
To evaluate if the percent change in seizure frequency was different for WWE with catamenial patterns compared to WWE without catamenial patterns, we used a rank ANCOVA, controlling for pre-conception ADSF and seizure type (generalized- versus focal-onset). A rank ANCOVA is commonly used to evaluate change in seizure frequency during clinical trials of antiseizure medications. We used Wald tests to evaluate if there was sufficient evidence for an association with a significance threshold of p < 0.05 [19].
Results
A total of 89 WWE were recruited into WEPOD. Participants tracked a median of 96.5 days (SD 47, range 30–156) prior to pregnancy and 258 days (SD 9.4, range 247–280) during pregnancy.
31 participants had seizure activity in the preconception period, and 28 of these completed the study with three discontinuing early. 23 participants had at least one menstrual cycle tracked in the preconception period. Finally, 13 participants became pregnant and had at least one day of tracking in pregnancy (Fig.
1).
Of 13 participants (Table 1), 6 (46 %) had catamenial epilepsy pattern: 2 (33 %) with C1 only, 1 (17 %) with C2 only, and 3 (50 %) with both C1 and C2. None met criteria for a C3 pattern, however, only 11 participants experienced one or more anovulatory cycles, among the 89 total participants, limiting the ability to evaluate for C3. The mean age was 30 years [IQR: 29–32 years] in non-catamenial group and 30 years [IQR: 26–33 years] in catamenial group.
All participants were white and 77 % non-Hispanic. In the catamenial group, 2 participants (33 %, 2/6) had focal seizures, compared with 6 participants (86 %, 6/7) in the non-catamenial group. No participants had unclassified seizures.
A higher proportion of participants in the catamenial group had generalized seizures compared with the non-catamenial group, but this difference did not reach statistical significance (p = 0.10), likely due to the small sample size.
This analysis focused on the change in seizure frequency during pregnancy as compared to the pre-conception period (Fig. 2). In the catamenial group, 2/6 (33 %, 95 % Confidence Interval [CI] 0–67 %) became seizure free during pregnancy, whereas 3/7 (43 %, 95 % CI 14–86 %) in the non-catamenial group became seizure free (Fisher exact p-value 0.99).
In the catamenial group, 4/6 (67 %, 95 % Confidence Interval [CI] 33–100 %) participants showed seizure improvement or stability during pregnancy, whereas 6/7 (86 %, 95 % CI 57–100 %) had improved or stable seizures in the non-catamenial group (Fisher exact p-value 0.56). When looking across all levels of seizure frequency change, the rank ANCOVA approach did not observe a significant difference in seizure frequency change in the catamenial compared to non-catamenial group (p = 0.42). That approach also controlled for the potential association of more improvement in generalized-onset seizures compared to focal-onset seizures (p = 0.058).
The overall rank ANCOVA model did not explain more variation than was expected by chance.
Discussion
This study of WWE seeking pregnancy who had seizures soon before pregnancy demonstrated that 77 % (10/13) had an improvement of seizures and 38 % (5/13) became seizure free during pregnancy. Catamenial patterns were not associated with more or less improvement in seizure frequency during pregnancy. However, only 15 % (13/86) of enrolled participants had sufficient seizures to evaluate that association.
These results build upon the MONEAD study, which also demonstrated no significant association between a historical catamenial seizure pattern and change in seizure frequency during pregnancy. These differ from the results of Cagnetti and colleagues’ analysis, which concluded that people with catamenial seizures had more improvement in seizure frequency than people who had non-catamenial seizures. There are two main differences between the study by Cagnetti, et al. and the WEPOD and MONEAD studies and their approach.
The WEPOD and MONEAD studies defined catamenial seizure patterns based on seizure diaries within 9 months of pregnancy (WEPOD: up to 5 months before pregnancy; MONEAD 9 months after pregnancy). Cagnetti and colleagues used at least 24 months of pre-conception seizure reporting to identify catamenial patterns. However, 39 % of participants with catamenial patterns were seizure free in the 9-months before conception in the Cagnetti and colleagues’ study.
Additionally, Cagnetti, et al. reported that in comparison to the participants with catamenial patterns, participants with non-catamenial patterns had a higher seizure frequency (p = 0.018) and slightly lower rate of seizure freedom (51 %) in the 9 months before conception. While Cagnetti and colleagues used multivariable regression models to account for those differences, there may have been other unmeasured factors that contributed to less improvement in non-catamenial patterns than catamenial patterns. For example, patients classified as having catamenial patterns one year may have transitioned to no longer having a catamenial pattern in the next year (e.g., when seizure free, which could have been due to changes in ASMs or natural progression of their epilepsy).
Therefore, there may be a difference between the long-term catamenial patterns observed in the Cagnetti and colleagues’ study compared to the periods immediately before and after pregnancy evaluated in WEPOD and MONEAD, respectively.
In this analysis of the WEPOD study, we excluded participants with pre-conception seizure freedom. Additionally, the shorter pre-conception seizure diary required a higher seizure rate to conclude that a catamenial pattern was present, as compared to a non-catamenial pattern although there was insufficient data to test if seizure rate was different. Based on these complex relationships between the timing of seizure freedom and the requirement of observing seizures to define a catamenial pattern, the prognostic value of a pre-conception catamenial seizure pattern was unclear.
Rigorous prospective evaluation of pregnancy in people with epilepsy can be challenging because of effective management of seizures [20]. Of the participants recruited into WEPOD, only 36 % (31/86) had seizures in the pre-conception period and, of those, 74 % (23/31) provided enough seizure and menstrual data to differentiate between catamenial and non-catamenial patterns. Of those 23 participants seeking pregnancy, 56 % (13/23) became pregnant during the study.
This funnel of participant inclusion and exclusion where only 15 % (13/81) of recruited participants provided enough data to evaluate this clinical question highlighted the challenge of rigorously studying seizure frequency and patterns prior to and during pregnancy in WWE. Potentially consequentially, the WEPOD study was one of the few studies with sufficiently high-quality data in these patients to evaluate our hypotheses.
Seizure frequency prior to and during pregnancy also is influenced by active management of and adherence to antiseizure medication [15,17,21]. In these participants, seizure frequency and seizure freedom may have been associated with changes in doses and clearance of antiseizure medications, which can substantially change over the course of pregnancy [21,22]. During this prospective study, participants’ antiseizure medications were managed according to the best practices at subspecialty centers with expertise in epilepsy during pregnancy.
Current recommendations and best practices at the time do not recommend altering antiseizure medications based on pre-conception catamenial seizures [22].
There are inherent limitations in the accuracy and reliability of seizure diaries, which is a common challenge in prospective studies and clinical trials [23,24]. These challenges are compounded by recall bias and other inaccuracies in retrospective studies [25], therefore the WEPOD and MONEAD studies and work of Cagnetti and colleagues provide uniquely high-quality data. Our analysis of the percent change of seizure frequency from preconception to pregnancy indicated that participants with generalized-onset seizures had more improvement in seizure frequency (p = 0.058).
That association was not observed in the larger MONEAD study. Future studies should include prospective seizure tracking beginning well prior to conception along with menstrual flow, continuing through delivery and postpartum in a larger number of participants.
Conclusions
In patients with seizures before pregnancy, the prognosis for seizure control during pregnancy was good. Previous work demonstrated that 73 % of patients who were seizure free immediately before or after pregnancy remained seizure free in pregnancy [14]. In patients who were not seizure free immediately before pregnancy, seizure frequency improved in 76 % of patients and 38 % of patients became seizure free.
There was insufficient evidence to support that a pre-conceptional catamenial pattern immediately before pregnancy would alter prognosis during pregnancy or counseling. More information is needed to improve the counseling and treatment of seizures during and around pregnancy.