Abstract
Background and Objectives Several studies have shown an increased risk of relapse after in vitro fertilization (IVF) in women with multiple sclerosis (MS), especially when a gonadotrophin-releasing hormone (GnRH) agonist stimulation protocol was used. Our objective was to investigate the risk of relapse after IVF in women with MS, overall and according to stimulation protocol (GnRH agonists vs antagonists), using data from the French national health insurance database.
Methods This retrospective cohort study included all women with MS who have benefited from IVF between January 1, 2010, and December 31, 2015, in France. Three-month exposed periods after IVF were compared with unexposed periods before IVF, each woman being her own control. Four outcomes were considered: annualized relapse rate (ARR), proportion of IVF with relapse, difference in the number of relapses “after–before,” and the delay from IVF to the first relapse. Relapses were identified by an algorithm based on MS-related hospital admissions and the use of corticosteroid therapy. Stimulation protocols and disease-modifying therapies (DMTs) were identified using drug claims. Zero-inflated Poisson regression models adjusted for age at IVF and the presence of DMT were used. A random effect on women was included because women may undergo multiple IVF procedures. Subgroup analyses by stimulation protocol and IVF outcome (pregnancy or failure) were conducted.
Results A total of 225 women accounting for 338 IVF procedures were included (the mean age at the first IVF 34.6 ± 4.5 years; 36% of women underwent at least 2 IVF procedures during the period). No increase in the risk of relapse after IVF was found overall (before vs after IVF: 0.20 vs 0.18 relapse per patient-year; 7.7% vs 7.1% of IVF with women having at least one relapse) and in subgroups. A lower ARR before and after IVF was observed among women who remained treated until IVF.
Discussion The maintenance of DMT until IVF seemed to be a determining factor in reducing the risk of relapse. Women with MS should be reassured because we did not show an increased risk of relapse requiring the use of corticosteroid therapy after IVF neither with GnRH agonists nor with GnRH antagonists.
Glossary
- ARR=
- annualized rate of relapse;
- ART=
- assisted reproduction technology;
- DMT=
- disease-modifying therapy;
- GnRH=
- gonadotrophin-releasing hormone;
- ICD-10=
- International Classification of Diseases, 10th Revision;
- IVF=
- in vitro fertilization;
- LTD=
- long-term disease registration;
- MS=
- multiple sclerosis;
- ZIP=
- zero-inflated Poisson regression
Multiple sclerosis (MS) is a chronic autoimmune disease of the CNS that particularly affects women and usually starts in young adulthood.1 There is no cure, but MS-specific disease-modifying therapies (DMTs) help to reduce the risk of relapse and delay disability progression in patients with MS.2,3 Pregnancy is also known to decrease relapse occurrence, especially in the second and third trimesters.4,-,6
Assisted Reproduction Technology (ART) is a medical procedure used primarily to address infertility, such as in vitro fertilization (IVF).7 IVF starts with ovarian stimulation, with gonadotrophin-releasing hormone (GnRH) analogs that can be either agonists or antagonists.8,9 Several transfers can take place for the same IVF (1or more frozen embryo(s) + the first fresh transfer) to increase the chance of success.
To date, only a few studies have considered the risk of MS relapses after IVF. All these studies10,-,15 showed an increase in the annualized rate of relapses (ARRs) after IVF compared with that before, except the most recent one15 that found for the first time a nonsignificant difference in ARR after IVF compared with that before. Those studies had several limitations, especially small sample sizes and risks of selection bias. The main objective of this study was to investigate the risk of relapse after IVF in women with MS, overall and according to the stimulation protocol (GnRH agonists vs GnRH antagonists), using data from the French national health insurance database.
Methods
Study Design and Data Source
This retrospective cohort study compared an exposed period of 3 months after IVF with an unexposed period of 3 months before IVF, each woman being her own control. Moreover, we considered a control period of 3 months 1 year before IVF because we hypothesized that patients may have been in a better phase of the disease during IVF, i.e., a period with fewer relapses. Because some women did not have a 1-year follow-up before IVF, the study population was smaller when considering this control period.
The French national health insurance database16 (Système National des Données de Santé [SNDS]) covers 98% of the French general population without any age or wealth criteria. It compiles exhaustive anonymous individual prospective data regarding the reimbursement of ambulatory health care such as consultations, drugs prescriptions, and exhaustive hospital activity of all public and private hospitals.
Each individual is identified by a unique lifelong identifier. The following characteristics are available: gender, year of birth, date of death, insurance scheme (general scheme, agricultural workers, self-employed, and other schemes), and long-term disease registration (LTD, which allows 100% reimbursement), coded according to ICD-10 codes17 and its corresponding starting year, if applicable.
Study Population
Women of childbearing age (15–49 years) with MS were identified from January 1, 2010, to December 31, 2015, using a 3-criteria algorithm: the presence of a LTD status for MS, or MS-related hospital admissions (ICD-10 code G35), or reimbursement for MS-specific drugs (beta-interferon, glatiramer acetate, natalizumab, fingolimod, dimethyl fumarate, and teriflunomide).18,-,20 Only data regarding patients with MS were accessed and stored in a secure environment. In this study, only women who have benefited from IVF, followed by at least 1 embryo transfer during the study period (to get a chance of being successful), were included.
IVF and Exposure to Ovarian Stimulation Protocols
Women may experience a different IVF history (1 or several IVF procedures, 1 or several transfers after IVF, and 1 or several pregnancies, either related to IVF or spontaneous) over the study period. In the French health insurance database (SNDS), we identified procedures of interest among the procedures related to infertility21 as follows: embryo transfers (coded JSEC001 and JSED001), ovulation induction (coded YYYY032), embryo warming (coded 0083), and embryo thawing (coded 0063). We considered both standard IVF and IVF called intracytoplasmic sperm injection, coded 0060 and 0061, respectively. The date of IVF corresponds to the date of fertilization. IVF procedures performed after October 1, 2015, were excluded because the post-IVF 3-month period was not available.
Each IVF procedure includes an initial stage of ovarian stimulation by GnRH agonists or antagonists, whose identification in the database was based on drug claims. Nafarelin acetate and triptorelin were classified as GnRH agonists, whereas ganirelix and cetrorelix were classified as GnRH antagonists.21
Relapse Outcomes
The event of interest was the occurrence of relapse, which is usually defined as new symptoms or worsening of preexisting symptoms for at least 24 hours and occurring at least 30 days after a previous relapse.22,23 Because this clinical event was not directly available in the database, we used a recently published specific algorithm based on hospital admissions and/or the use of corticosteroids (either oral or intravenous).24,-,26
Four outcomes were considered: the ARR and the proportion of IVF where a woman had at least 1 relapse after IVF were compared with those before IVF; the difference in the number of relapses after vs before IVF was also calculated; and the delay from IVF to the first relapse within 1 year.
Other Data
Exposure to DMT was defined using dates and quantities of drug issued by pharmacists (deliveries made on a monthly basis). An IVF procedure was considered as exposed if the date of IVF was before the end of treatment.
An IVF procedure was defined as successful when a pregnancy occurred less than 14 days after an embryo transfer. Outside this period, a pregnancy was considered as spontaneous. Pregnancy was identified through its issue, and the start of pregnancy was calculated using a specific algorithm.27 Pregnancies of less than 37 weeks were considered as premature.
Statistical Analyses
Descriptive analyses were performed using mean ± SD for quantitative variables and proportions for qualitative variables. First, the ARR was defined as the sum of the number of relapses in the considered period divided by the sum of the follow-up times. To compare ARR between the 3-month period before IVF and the 3 following months, a zero-inflated Poisson regression (ZIP) was used (R software packages lme4 and ggeffects). Because the occurrence of relapses was not frequent (based on a descriptive analysis), i.e., the expected frequency of zero was high, a counting model with zero inflation was the most appropriate. Because the data were not overdispersed, this model was preferred to a negative binomial model. An offset on the follow-up time was added to obtain the ARR. IVF was the statistical unit of analysis, and because a woman can undergo several IVF procedures (repeated measures), a random effect on the woman was included. Incidence rate ratios and their CIs were calculated and summarized in a forest plot.
Second, regarding the proportion of IVF where a woman had at least 1 relapse in the 2 periods of interest, a McNemar test was used to check whether it was the same woman who experienced relapses before and after IVF. Then, a logistic regression was used (mclogit package), and a random effect on IVF was added to compare the 2 periods.
Third, the difference between the number of post-IVF relapses and the number of pre-IVF relapses was calculated, categorized into 3 ordered classes: decrease, no change, and increase and modeled using an ordinal multinomial logistic regression. Fourth, to measure the time from IVF to the first relapse within 1 year, a Kaplan-Meier survival curve was estimated overall and according to subgroups of interest and compared with log-rank tests (survminer package).
All models were adjusted for age at IVF and the presence/absence of DMT at IVF. Subgroup analyses were performed according to the stimulation protocol (GnRH agonists vs antagonists) and IVF outcome (success vs failure).
Finally, 2 sensitivity analyses were conducted. First, a duration of 6 months was taken instead of 3 months as period of interest before and after IVF. Second, an IVF procedure was assumed as exposed if the patient had stopped treatment for less than 3 months (to consider a possible persistent effect of the treatment after its cessation). Results were considered significant when p < 0.05. All analyses were conducted using R (v.3.6.0).
Standard Protocol Approvals, Registrations, and Patient Consents
Ethical and data access approvals for the study were obtained in accordance with the current French legislation (IDS approval decision n°191, May 25, 2016, and CNIL decision DE-2017-026, March 21, 2017).
Data Availability
According to data protection and the French regulation, the authors cannot publicly release the data from the French national health data system (SNDS). However, a request for data reuse may be made and would require the obtention of previous approval of the French regulatory authorities (snds.gouv.fr/SNDS/Processus-d-acces-aux-donnees).
Results
Study Population
Between 2010 and 2015, among the 46,294 women of reproductive age who experienced MS, 244 women had IVF with at least 1 embryo transfer. After excluding inconsistencies, our study population was composed of 225 patients who benefited from a total of 338 IVF procedures during the study period, with a systematic follow-up duration of 3 months before and after IVF (Figure 1).
The mean age at the first IVF was 34.6 ± 4.5 years, and 36% of women had at least 2 IVF procedures during the period (Table 1). In case of repeated IVF procedures, the mean interval between 2 procedures was 267 ± 283 days (median 182 days). Regarding the stimulation protocols, 205 IVF procedures (60.7%) were performed with a GnRH antagonist protocol and 125 (37.0%) with a GnRH agonist protocol, and the 8 remaining (2.3%) could not be determined (Table 1). Overall, 142 pregnancies (33 spontaneous and 109 IVF pregnancies) occurred in 121 different women, i.e., a pregnancy rate of 42.0% (Figure 1) over the study period. The IVF success rate was 32.2% (27.8% if agonist stimulation protocol vs 34.4% if antagonist protocol). Regarding outcomes, 108 pregnancies (76.1%) resulted in live births, of which 17 (15.7%) were premature (Table 2). When we focused on the period of interest of the 3 months after IVF, we observed 92 IVF pregnancies, i.e. 92 IVF successes and 246 IVF failures (no IVF pregnancy within 3 months). However, we also observed 8 cases of spontaneous pregnancy among the IVF failures, which were excluded in the subgroup analysis according to IVF outcome. Finally, the subgroup analysis was performed on 92 IVF successes and 238 failures. Of the 338 IVF procedures, one quarter (82, 24.3%) was performed among women under DMT (71 on first-line treatment and 11 on second-line treatment). Twelve women stopped DMT in the 3 months preceding IVF and 9 women in the 3–6 previous months (Table 1).
Characteristics of the Study Population
Pregnancy Outcomes and Mean Gestational Age
ARR Before and After the IVF Procedure
The ARR in the year before IVF was 0.35 relapses per patient-year. Among the 338 IVF procedures, 314 (92.9%) did not result in any relapse, leading to an overall ARR of 0.18 in the 3 months after IVF, compared with 0.20 in the 3 months before IVF. As shown in Figure 2 and summarized in Table 3, there was no statistically significant increase in the ARR in the 3 months after IVF compared with that in the 3 months before, overall and in subgroups. The ARR was lower in treated patients (0.15 before IVF; 0.13 after) than in untreated patients (0.28 before IVF; 0.25 after). The protective effect of the pregnancy was also observed (the ARR decreased from 0.30 to 0.07). Among the 238 IVF failures, 181 women (76.1%) were not treated during IVF, and only 6 (3.3%) women restarted the treatment within 3 months of IVF.
*3 months before IVF, **3 months 1 year before IVF.
Annualized Relapse Rate and Incidence Rate Ratio 3 Months After IVF Compared With That 3 Months Before and Then With 3 Months, 1 Year Before
We found the same findings when using the 3-month control period 1 year before IVF (0.13 before IVF; 0.09 after) (Figure 2 and Table 3). Only the GnRH agonists and IVF failure subgroup reached statistical significance but toward a decreased ARR after IVF, probably due to a high ARR before IVF (0.72).
Because natalizumab and fingolimod are associated with a risk of rebound after stop, we checked how many women were in that case. Among the 4 women who stopped natalizumab in the 6 months before IVF, 3 of them (75.0%) had a relapse after stop, which occurred in the 3 months before IVF. In addition, 2 of the 4 women switched to interferon within 1 month and then were considered as treated during IVF, whereas the 2 other women remained untreated. Among the 2 women who stopped fingolimod treatment in the 6 months before IVF, none of them had a relapse. One woman switched to glatiramer acetate and the other was untreated when she had IVF.
The Proportion of Relapses
Among the 338 IVF, 26 women (7.7%) had at least 1 relapse in the 3 months before and 24 women (7.1%) had at least 1 relapse in the 3 months after IVF (6 women had relapses both before and after IVF). The multivariate logistic regression did not show any difference in the proportions of women having relapse before vs after IVF, which was consistent with the results from the ZIP.
Among the few women who had a relapse after IVF, only 2 were treated (2/82 = 2.4% vs 22/256 = 8.6% among untreated women). The same findings were found when using the 3-month control period 1 year before IVF (9.1% vs 7.0% of women with at least 1 relapse, before and after IVF, respectively).
The Difference in the Number of Relapses After and Before
Most IVF procedures were associated with a stable number of relapses (88.8%), some had a decrease (5.9%) and others had an increase (5.3%). In this multivariate analysis where the reference class was no change, no significant association was found (Table 4). Only increasing age was associated with a lower risk of having an increased number of relapses after IVF.
Results of the Ordinal Multinomial Regression Over 2 Control Periods: Main Analysis and Control Period
By contrast, using the control period 1 year before IVF, women who had a GnRH antagonist protocol had a lower risk (OR <1) of having a decrease in their ARR after IVF than no change, compared with those who had a GnRH agonist protocol.
Time to First Relapse Within 1 Year of IVF
The Kaplan-Meier analyses did not reveal any significant difference in the time to first relapse in the year after IVF between the 2 stimulation protocols despite a p value near significance (p = 0.052) (Figure 3). When restricting to the first 3 months after IVF, no significant difference (p = 0.99) was found.
The 2 sensitivity analyses confirmed all these results regarding the 4 outcomes.
Discussion
In this study, we did not find any significant increase in the risk of relapse requiring the use of corticosteroid therapy after IVF compared with that before IVF in women with MS in France, between 2010 and 2015, overall and according to the stimulation protocol and the outcomes of IVF. These results differ from several previously published studies and do not support the hypothesis that GnRH agonist protocol is responsible for the ARR increase. Thus, our findings may have direct implications in the management of women with MS who are pursuing fertility treatment.
The ARR in the year before IVF was 0.35 relapse per patient-year, which was comparable with that in recent studies about pregnancy and MS,28,29 and probably reflects the current risk of relapse in the therapeutic area. Among studies that focused on the risk of relapse after IVF, it was comparable with those in the studies conducted by Hellwig et al.11 (0.29) in 2008 and Bove et al.15 (0.27) in 2019, but lower than those in the studies conducted by Correale et al.13 (0.42) in 2012 and Michel et al.14 (0.80) in 2012. In the study conducted by Laplaud et al.,10 no woman had a relapse in the 3 months before IVF. Treatment epoch may explain such differences. Indeed, the arrival of new DMTs combined with a better knowledge of prognostic factors in MS has led to changes in practices from neurologists toward an earlier treatment in the disease course. It also applies to pregnancy, where more and more women with MS keep their treatment until conception, or even during pregnancy for women with active disease.20 Indeed, the proportion of women having a DMT during IVF was almost 25% in our study, whereas it was null in the previously published series. We do expect that the use of a DMT until IVF (mainly first-line treatments) helps preventing the occurrence of relapse, as highlighted by the very low number of treated women having a relapse in the 3 months after IVF compared with untreated ones (2.4% vs 8.6%, respectively). Among the 24 women who had a relapse within 3 months before IVF, 3 of them had recently stopped natalizumab and may experience a disease rebound. Owing to the limited number of women who restarted DMT earlier after an IVF failure (n = 6), we cannot make a link between the absence of relapse and the immediate restart of a DMT.
To identify MS relapses, we relied on a recent algorithm based on 5 criteria and which allowed the identification of relapses requiring hospital admission or corticosteroid therapy. It means that benign relapses (i.e., not seen by neurologists or not needing the use of corticosteroid therapy) were not considered in the outcomes, which may imply that ARR may be underestimated. However, in a study using this algorithm of identification of relapses,25 the annual relapse rate and the proportion of patients having at least 1 relapse were close to other studies using clinical data. Finally, because this study was based on health insurance data, we can know when an MRI was performed, but we cannot access to MRI results to confirm relapses. Regarding the treatment of relapses during pregnancy, there is no specific recommendation,30 but we cannot rule out that neurologists may use less corticosteroid therapy during pregnancy. Indeed, several studies (outside MS) highlighted an excess risk of cleft lip and palate with high-dose corticosteroid exposure in the first trimester of pregnancy,31,-,33 but this was not systematically observed.34,-,37 If so, the risk of relapse after IVF may have been underestimated in pregnant women. This limitation should not question our findings because the number of pregnant women was less than half, and above all, subgroup analyses offered the opportunity to consider separately IVF successes (i.e., pregnant women) and IVF failures.
In this study, the date of identification of relapses corresponded to the date of care, which necessarily occurred later than first symptoms. In the previously published studies, the relapses were identified by a neurologist, who was able to retrace the history and note the date of relapse onset. Such discrepancy means that we may have missed some relapses, which started at the third month after IVF, but were treated in the fourth month, for example. Similarly, it was possible that post-IVF relapses in our analysis may have started before IVF and were therefore misclassified. After verification, only 1 woman had a relapse identified 14 days after IVF, which could have started before the procedure. Finally, the values of ARR and the dynamics of the ARR around pregnancy (data not shown) led us think that the algorithm was rather efficient. It has also been recently shown that administrative databases detect less relapses on average than clinical databases38 (the mean ARR of 0.44 for clinical data, 0.53 for claims data, and 0.62 for pregnancy-specific registry). To deal with this potential limitation regarding relapse ascertainment, we planned to replicate this analysis on data from the French Observatory of MS (OFSEP)39 linked to the French health insurance database. Such a linkage will have the advantage of combining strengths of both data sources and will also offer the opportunity to assess the properties of the relapse algorithm.
Regarding calendar period, the use of IVF stimulation protocols in women with MS has also evolved over time, probably linked to previous studies showing an increased risk of relapse after GnRH agonist stimulation. Thus, the GnRH antagonist protocol was the most frequent in our study (60.7%) in contrast to the other studies. We did not find any increase in the ARR after IVF, whatever the protocol.
Moreover, in the previous studies, patient recruitment was based on neurologists, which may have favored the inclusion of more active patients, whereas the French health insurance system used in this study is almost exhaustive and free from any selection bias. Risk of recruitment bias combined with small sample sizes (6–32 patients and 10–78 IVF procedures) may explain the statistical significance of previous results.
Indeed, 1 strength of this study is the number of patients included, 225 women and 338 IVF procedures, and data source, which is nationwide and almost exhaustive (98% coverage). In addition, when comparing the spontaneous pregnancy rate (between 17% and 24%) in the general population who have benefited from IVF40 with that in women with MS in this study (23.2%), we find comparable results. However, such French health insurance database has also limitations. For instance, for drugs (DMT and stimulation treatments), we have the date of delivery from the pharmacist and not the date of effective use, whereas dates of medical procedures such as IVF are exact. Therefore, there is a risk of overcounting exposed pregnancies. In addition, the database did not provide the context of infertility.
From a methodological point of view, we tested different statistical models to ensure the robustness of the results. IVF was the unit of analysis, and the inclusion of a random effect was essential because women may undergo several IVF procedures. It also means that analyses needed to be adjusted for other characteristics such as age and the presence of DMT. In addition, before and after design offers the opportunity to fully adjust for individual characteristics, including unmeasured confounding factors. Moreover, we compared the ARR at 3 months after IVF with a control period 1 year before IVF because IVF procedures are expected to be scheduled outside an inflammatory period.
Despite several limitations, we do think that women with MS should be reassured about the absence of an increased risk of relapse requiring the use of corticosteroid therapy after IVF, whether performed with GnRH agonists or GnRH antagonists. Thus, there should be no hesitation to use IVF among women with MS who have difficulties conceiving. Moreover, the stimulation protocol should be adapted to the patient and to the context of infertility and not to MS, contrary to what may be performed today. To conclude, we did not find any increase in the risk of MS relapse requiring hospitalization or the use of corticosteroid therapy after IVF in a nationwide exhaustive data set over 6 years.
Study Funding
This work was funded by Fondation pour l'aide à la recherche sur la sclérose en plaques (ARSEP).
Disclosure
M. Mainguy and H. Tillaut were funded by the Fondation pour l'Aide à la Recherche sur la Sclérose en Plaques (ARSEP). The other authors report no relevant disclosures. Go to Neurology.org/N for full disclosures.
Acknowledgment
The authors thank Soline Leblanc (MSc), Mathilde Lefort (PhD), and Jonathan Roux (PhD), EA 7449 REPERES (Pharmacoepidemiology and health services research), Univ Rennes, EHESP, Rennes, France, for their helpful comments.
Appendix Authors
Footnotes
Go to Neurology.org/N for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article.
Submitted and externally peer reviewed. The handling editor was Olga Ciccarelli, MD, PhD, FRCP.
This Null Hypothesis article is published as part of a collaborative effort between Neurology® and CBMRT.
Editorial, page 737
Null Hypothesis: NPub.org/Null
- Received January 10, 2022.
- Accepted in final form June 13, 2022.
- © 2022 American Academy of Neurology
References
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
