The reproductive effects of beta
interferon therapy in pregnancy
A longitudinal cohort
R. Boskovic, MD; R. Wide, BcS; J. Wolpin, PhD; D.J. Bauer, PhD; and G. Koren, MD
Abstract—Objective: To determine whether interferon therapy during human pregnancy increases reproductive risks in
women. Methods: This longitudinal, controlled cohort study consisted of three groups of women: an exposed group, a
disease matched unexposed group, and a healthy comparative group. Subjects were selected from women contacting the
Motherisk Program regarding maternal beta interferon exposure, mostly for multiple sclerosis during pregnancy, from
1997 to 2004. After delivery all of the women were re-contacted for a follow-up interview regarding maternal health,
pregnancy outcome, and neonatal health. Results: The study group (n ? 16 women, 23 pregnancies) were exposed to
interferon beta-1a (Avonex, Rebif) and interferon-1b (Betaseron). There was a decrease in mean birth weight in the
exposed group (3,189 ? 416 g) as compared to healthy controls (3,783 ? 412 g, p ? 0.002). Women exposed to beta
interferon had a higher rate of miscarriages and stillbirths (39.1%) vs healthy controls (5%) (p ? 0.03), even after
correction for potential confounders. There were two major malformations (abnormality in the X chromosome, Down’s
syndrome) among exposed fetuses. Conclusions: Beta interferon therapy in the first trimester of pregnancy appears to be
associated with an increased risk for fetal loss and low birth weight.
First suggested by Jacobs et al. in 1987, interferon
beta is currently the most widely used therapy for
multiple sclerosis (MS).1-3Beta interferon has been
shown to reduce the relapse rate, disease activity,
and disease progression of active-relapsing MS1,2,4,5
by decreasing human leukocyte antigen (HLA) ex-
pression. While pregnant women with MS do not
appear to be more likely to have pregnancy and de-
livery complications, they exhibit higher relapse
Experimental studies have shown that high doses
of beta interferon therapy in pregnant rhesus mon-
keys (40 times the human recommended dose based
on surface area) do not result in teratogenicity but
dose-dependent abortive effect after three to five
Similarly, it has been reported that alpha interfer-
ons are unlikely to be teratogenic, but may increase
the risk of spontaneous abortions and intrauterine
growth retardation (IUGR).9
The limited human experience with beta inter-
feron is based on case reports only. During phase III
of the clinical trials, seven women with MS became
pregnant while on interferon, resulting in five nor-
mal pregnancies and two miscarriages. In all of the
cases interferon was discontinued when pregnancy
was recognized.1An additional case report described
normal pregnancy outcome in a 25-year-old woman
treated with interferon beta for retinal necrosis.
However, interferon therapy started at 25 weeks of
gestation, beyond the period of organogenesis.10
Pregnant women receiving interferon are typically
advised to discontinue therapy due to lack of repro-
ductive information.11However, it has been sug-
gested that interferon use during early pregnancy
should not be an indication for voluntary pregnancy
termination.11We sought to determine whether in-
terferon therapy during human pregnancy is associ-
ated with increased reproductive risks among
women with MS. This is the first prospective,
disease-matched controlled human cohort study on
the reproductive effects of beta interferon therapy,
primarily for MS.
and counseling service at the Hospital for Sick Children in To-
ronto. The service provides counseling to members of the public
and health professionals regarding the reproductive safety of
drugs, chemicals, radiation, and infection during pregnancy and
The Motherisk Program is a teratogen information
Editorial, see page 788
See also page 802
From the Motherisk Program (Drs. Boskovic, Wolpin, and Koren, and R. Wide), Division of Clinical Pharmacology-Toxicology, The Hospital for Sick Children,
Toronto, and University of Toronto; The Ivey Chair in Molecular Toxicology (Dr. Koren), University of Western Ontario, Canada; and Quantitative
Psychology Program (Dr. Bauer), Department of Psychology, University of North Carolina, Chapel Hill.
Disclosure: The authors report no conflicts of interest.
Received August 10, 2004. Accepted in final form July 28, 2005.
Address correspondence and reprint requests to Dr. Gideon Koren, Division of Clinical Pharmacology, The Hospital for Sick Children, 555 University Avenue,
Toronto, Ontario, M5G 1X8 Canada; e-mail: firstname.lastname@example.org
Copyright © 2005 by AAN Enterprises, Inc.
lactation. After the expected date of confinement standard ques-
tioners are used for follow-up of pregnancy outcome. The Hospi-
tal’s Research Ethics Board approved the study protocol. The
participating women gave informed consent for the follow-up
This longitudinal, controlled, cohort study consisted
of three groups of women: a beta interferon exposed group, a
disease-matched unexposed group, and a healthy control group.
Subjects selected for the study were pregnant women who con-
tacted the Motherisk Program regarding maternal beta interferon
or Copaxone exposure during pregnancy, between 1997 and 2004.
Women using beta interferon in pregnancy formed the study
group, while those that discontinued beta interferon or Copaxone
prior to conception formed the disease-matched group. A healthy
control group was formed from callers to the Motherisk Nausea
and Vomiting (NVP) helpline, which counsels pregnant women on
the treatment of NVP symptoms.
During the initial contact, pregnancy information was col-
lected. This included general information regarding maternal
health and any possible xenobiotic exposure, including drug ther-
apy. After the expected date of confinement the women participat-
ing in this study were re-contacted for a follow-up interview. The
interviews gathered information regarding maternal health, preg-
nancy outcome, and neonatal health. All xenobiotic exposures
throughout gestation were recorded.
The follow-up interviews documented maternal health prob-
lems and their severity, all medications used during pregnancy,
prescription and over the counter drugs, alcohol, tobacco, and
recreational drug use during gestation. For all exposures, the
indication, dose, frequency, and side effects were recorded. We
also recorded pre-pregnancy weight, weight at delivery, method of
delivery, vaginal or caesarean section, and total length of labor.
Women were asked to provide information regarding pregnancy
outcome in terms of live birth, miscarriage, stillbirth, gestational
age at birth, birth weight, neonatal health, birth defects, and
The information obtained from the mother was verified,
through a written document, by the child’s pediatrician or family
physician, upon the mother’s verbal consent. The physicians were
requested to complete a questionnaire stating the baby’s birth
weight and head circumference, as well as any malformations or
As a first step, univariate analyses were
performed. These procedures included frequency counts, cross-
tabulations, and analyses of variance.
Subsequently, multivariate analyses, such as linear and logis-
tic regressions with standard and mixed models, were employed.
Because we were interested in accounting for the clustering of
births within women and modeling differences among all three
groups we used a mixed model with a random intercept for moth-
ers in PROC GLIMMIX and PROC NLMIXED on SAS. GLIMMIX
uses a linearization of the model (PQL) to obtain estimates and is
more stable and quicker but less accurate. NLMIXED approxi-
mates the integral in the likelihood numerically using quadra-
ture to obtain a more accurate solution, but optimization is more
tations were followed up. The exposed study group (Group
1) consisted of 16 women with 23 gestations. One woman
gave information regarding three gestations, 5 women
gave information regarding two gestations, and 10 women
gave information regarding one gestation (see table 1).
Fourteen women received beta interferon for MS, one for
the treatment of thrombocytosis, and one for the treatment
of essential thrombocythemia. Beta interferon-1a (Avonex)
was administered IM, 30 mcg once a week and Rebif rang-
ing from 22 to 44 mcg three times a week, subcutaneously.
Beta interferon-1b (Betaseron) was administered subcuta-
neously ranging from 2 MIU to 6 MIU every other day. The
mean gestational duration of beta interferon therapy was 9
weeks, ranging from 2 to 38 weeks. In 21 gestations (all
with MS) interferon therapy was terminated during the
first trimester of pregnancy. The woman with essential
A total of 46 women were recruited and 64 ges-
thrombocythemia discontinued therapy at 21 weeks of ges-
tation, while the woman with thrombocytosis discontinued
therapy at 38 weeks of gestation.
The disease control group (Group 2) consisted of 12
women with MS with 21 gestations. Six of these women
gave information regarding three gestations, three women
gave information regarding two gestations, and three
women with one gestation. Ten patients had discontinued
either interferon therapy or Copaxone for the treatment of
MS at least 1 month prior to conception, while one woman
discontinued interferon for the treatment of hepatitis C
and one woman discontinued drug taken for treatment of a
wart on her foot.
The healthy control group (Group 3) consisted of 18
women with 20 gestations. Two women gave information
regarding two gestations while 16 women gave information
regarding one gestation (see table 1). None of these women
had any diseases or major health concerns.
The mean age of exposed women (35.6 ? 5.3 years) was
higher than both the disease-matched (30.2 ? 4.8 years)
and healthy controls group (33.5 ? 5.6 years) (p ? 0.002,
table 2). The mean pre-pregnancy weight in the exposed
group was similar among the three groups (see table 2).
Women exposed to interferon beta had nine spontaneous
abortions and one fetal death yielding a significantly
higher rate of pregnancy loss and lower rate of live birth
than the healthy control group (see table 1).
Subsequently, initial analyses for the prediction of non-
live births involved the use of a logistic mixed-effects
model. However, both PQL and ML estimation with nu-
merical integration failed document variance component
due to mothers’ clustering once maternal age, gravidity,
prior history of lost pregnancies, and study group variables
were controlled for. More specifically, the estimate for this
variance component obtained by both methods was zero.
Given the lack of evidence of dependence due to clustering
of multiple births in mothers, a standard logistic regres-
sion model was used to evaluate the effects of interest.
Maternal age was positively associated with non-live
births: with each year of age a non-live birth is 1.22 times
more likely than the year before. No significant effects of
gravidity or prior history of lost pregnancies were ob-
served. Of greater interest, mothers in the interferon beta
study group were 6.9 times more likely to produce a non-
live birth, after controlling for age, gravidity, and prior
pregnancy history, an effect that was significant (see table
3). While a higher rate of non-live births was also observed
in the discontinued interferon group, this effect did not
reach significance (see table 3).
In predicting child birth weight the random intercept,
reflecting dependence due to clustering, had a variance of
71,729, which is marginal by the Wald test (z ? 1.52, p ?
0.06). However, the Wald test is conservative and power is
low at this sample size. The intraclass correlation, control-
ling for gestational age, maternal weight gain, and expo-
sure group, is 0.53, indicating a relatively strong
clustering effect—the birth weights of children born to the
same mother are correlated at 0.53 even after controlling
for other variables in the model. Hence, it is probably
important to retain the random intercept in the model,
despite its marginal significance, for the fixed effects esti-
mates to have accurate standard errors and CIs. This anal-
September (2 of 2) 2005
ysis revealed that interferon has an independent and
significant negative effect on birth weight (table 4).
There were two major chromosomal malformations
among exposed fetuses. The first was an offspring of a
woman with MS, with abnormality in the X chromosome
and a spontaneous abortion.
The second major malformation was Down’s syndrome
detected prenatally and the family decided to abort the
pregnancy. The woman giving birth to a child with Down
syndrome was 29 years old with no previous history or
known risk factors for Down syndrome.
case reports have suggested that beta interferon is
unlikely to cause major malformations, but may in-
crease the risk of spontaneous abortions and IUGR.9
Our study was conducted to determine whether beta
interferon therapy for MS increases the risk of repro-
In planning this study it was deemed important to
try to control for potential confounding effects of
maternal characteristics, and hence we also re-
Previous animal studies and human
Table 1 Characteristics of women among the study and control groups
Women with disease (mostly MS)Healthy women
IFN No IFNNVP
Distribution of characteristicsGroup 1Group 2 Group 3
Total number of women*16 1218
No. of gestations in the study
Reporting 1 gestation103 16
Reporting 2 gestations532
Reporting 3 gestations160
Total no. of gestations2321 20
End result of reported pregnancy, n (%)†
Live birth 12 (54.5)17 (81.0)18 (90.0)
Non-live birth10 (45.5) 4 (19.0)2 (10.0)
Description of non-live birth, n (%)
No. of spontaneous abortions9 (39.1)4 (19.0)1 (5.0)
No. of fetal deaths1 (4.3)1 (5.0)
End result of pregnancies, history, n (%)
All non-live births7 (30.4) 4 (19.0)2 (10.0)
Combinations—live and non-live9 (39.1)5 (23.8) 3 (45.0)
All live births 7 (30.4)12 (57.1)15 (75.0)
No. (%) of major malformations2 (8.7)1 (4.8)1 (5.0)
No. (%) of neonates born prematurely‡2 (8.7) 3 (4.3) 1 (5.0)
No. (%) of smokers 2 (8.7) 1 (4.8)0 (0.0)
No. (%) of alcohol users 7 (30.4)2 (9.5)1 (5.0)
* Forty-five women participated in the study. One woman appears in two groups, Group 1 (1 gestation) and Group 2 (3 gestations). The
last gestation for this woman appears in Group 1.
† One missing value case in Group 1.
‡ ?37 weeks gestation (postconception).
MS ? multiple sclerosis; IFN ? interferon; NVP ? Nausea and Vomiting helpline.
Table 2 Comparison of maternal and neonatal characteristics and pregnancy outcome among the three groups
Characteristics/outcomeGroup I: Interferon Group II: Disease matchedGroup III: Healthy controls Significant pair
Age at time of gestation, y, mean SD 35.6 ? 4.5 30.2 ? 4.833.5 ? 5.6 1 & 2
Pre-pregnancy weight, lb, mean SD137.4 ? 27.9 139.6 ? 16.9150.2 ? 29.3
Weight gain in pregnancy, lb, mean SD225.7 ? 15.030.2 ? 16.844.9 ? 27.9
Gestational age at time of call, wk 4.2 ? 4.42.2 ? 3.8 9.3 ? 3.0 1 & 2, 2 & 3
Child’s birth weight, g 3,189 ? 4163,498 ? 470 3,783 ? 4121 & 3
Gestational age, wk 37.8 ? 2.337.4 ? 2.338.2 ? 1.0
The last three variables refer to live birth only.
September (2 of 2) 2005
cruited a disease-matched group of women with
MS who discontinued their interferon therapy
prior to conception.
The prospective longitudinal design of this study
aimed to ensure that selection and recall bias are not
confounding the interpretation. In particular, it has
been shown that retrospective studies exhibit higher
rates of adverse effects, because women with adverse
fetal outcome may be more motivated to report their
experience than those who experience normal preg-
nancy outcome.14Yet, not being a randomized trial,
the group of women with MS using interferon could
be different from the disease-control group in impor-
tant confounders that may affect the outcome of
interest. For this reason we also conducted multivar-
iate analyses to try to address the effects of con-
founders. Although the mean birth weight of 3,189 g
is only 200 g below typical baby birth weight of
healthy women, it is worth noting that this is similar
to the effect size of heavy maternal smoking on
birth weight. Because women in the three groups
had clustering of different numbers of offspring,
our statistical analyses aimed at addressing this
potential bias (i.e., that not every child is an inde-
Our analysis suggests that interferon therapy for
MS during pregnancy may decrease birth weight
even after controlling for important confounders.
Our study agrees with previous findings9with inter-
feron alpha. Consistent with studies in primates, our
study suggests higher rates of pregnancy loss,
mainly due to spontaneous abortions, among women
treated with beta interferon.
Here again, the association persisted after logistic
regression for correction of potential confounders,
and after addressing potential effect of clustering.
Although ethanol consumption was significantly
more common among women who used beta inter-
feron therapy during pregnancy, the exposure was
very mild in all cases (several drinks in pregnancy)
and this type of exposure has not been associated
with an increased risk of spontaneous abortions.15
Kesmodel et al. have reported that women consum-
ing ?5 drinks/week are at increased risk of first
trimester abortion.16Another study has suggested
that moderate alcohol consumption (?3 to 7 drinks/
Table 3 Logistic regression for prediction of non-live births
OROR 95% CI
Predictors Coefficient B SEWald ?2
Exp (B)Lower Upper
Maternal age0.20 0.09 4.90 0.0271.221.02 1.45
?0.530.381.890.169 0.59 0.28 1.25
No. of prior lost pregnancies0.560.57 0.97 0.3261.75 0.57 5.33
IFN vs healthy group 1.940.904.59 0.0326.94 1.1840.79
Disease vs healthy group1.361.07 1.63 0.2023.910.4831.67
?8.29 3.196.76 0.009———
IFN ? interferon.
Table 4 Linear mixed model for prediction of child birth weight (g)
Unstandardized coefficient 95% CI
Fixed effectB SE*
t df †
Gestational age at birth, wk 1,230.38310.3930.93310.90.00459.44 187.32
Mother’s weight gain, lb30.9620.97 10.33330.5 0.192
IFN vs healthy group*
Disease vs healthy group†
IFN vs disease group‡
Random effectVarianceSE Wald z
* Computed using the Kackar–Harville method.
† Computed using the Kenward–Rogers method.
‡ Computed from coefficients from preceding two lines of the table.
IFN ? interferon.
September (2 of 2) 2005
week) may be associated with increased risk of spon-
taneous abortion.17None of the women in our study
reported consumption of alcohol in similar amounts.
As women in the healthy control group called later
in pregnancy, there is a possibility that some miscar-
riages were missed. Our study is limited by its small
sample size and by potential confounders not con-
trolled for in our analysis. However, pregnancy expo-
sure to beta interferon is a relatively uncommon
event. Often women are advised to discontinue inter-
feron prior to conception, making it challenging to
recruit women who have taken the drug in preg-
nancy. Furthermore, we did not have measured dis-
ease activity and, although we tried to address a
variety of confounders, it is possible that women re-
ceiving beta interferon differed from those not receiv-
ing the drug in confounders not controlled by us. It
does not appear likely, though, that a randomized
trial of interferon vs no therapy will be conducted in
The results of our study suggest that women who
become pregnant while taking beta interferon should
not terminate pregnancy but rather discontinue the
drug until delivery. Discontinuing beta interferon
therapy during gestation should not necessarily in-
crease the risk of relapse of MS, as pregnancy tends
to reduce such risk.7,18
It appears that labor is followed by an increase in
the relapse rate 6 months postdelivery. Women
should be therefore advised to resume interferon
therapy very soon after delivery if they do not intend
Future studies should aim at recruiting larger co-
horts of women for both the study and disease-
1. Panitch H, Miller A, Paty D, Weinshenker B. North American Study
Group on Interferon beta-1b in Secondary Progressive MS. Interferon
beta-1b in secondary progressive MS: results from a 3-year controlled
study. Neurology 2004;63:1788–1795.
2. Jacobs L, Rudick R, Simon J. Extended observations on MS patients
treated with IM interferon-beta1a (Avonex): implications for modern
MS trials and therapeutics. J Neuroimmunol 2000;107:167–173.
3. Jacobs L, Salazar AM, Herndon R, et al. Intrathecally administered
natural human fibroblast interferon reduces exacerbations of multiple
sclerosis. Results of a multicenter, double blind study. Arch Neurol
4. Piper JM, Wen T, Xenakis E. Interferon therapy in primary care. Pri-
mary Care Update Obstet Gynaecol 2001;8:163–169.
5. Sibley WA, Ebers GC, Panitch HS, Reder AT. Interferon beta-1b is
effective in relapsing-remitting multiple sclerosis. Neurology 1993;43:
6. Mueller BA, Zhang J, Critchlow CW. 2002 Birth outcomes and need for
hospitalization after delivery among women with multiple sclerosis. Am
J Obstet Gynecol 2002;186:446–452.
7. Confavreux C, Hutchinson M, Hours MM, Cortinovis-Tourniaire P, Mo-
reau T. Rate of pregnancy-related relapse in multiple sclerosis. Preg-
nancy in Multiple Sclerosis Group. N Engl J Med 1998;339:285–291.
8. Avonex (Interferon beta 1-a). Biogen, Canada, 1998. Product mono-
9. Mitsuhiro H, Minakami H, Koshizuka S, Sato I. Administration of
interferon-alpha during pregnancy: effects on fetus. Perinat Med 2000;
10. Shiraki K, Moriwaki M, Ataka S, Henmi K, Miki T, Kanaoka Y. Acute
retinal necrosis late in the second trimester. Am J Ophthalmol 1998;
11. Ferero S, Pretta S, Ragni N. Multiple sclerosis: management issue
during pregnancy. Eur J Obstet Gynecol Reprod Biol 2004;115:3–9.
12. McCullagh P Nelder JA. Generalized linear models (second edition).
London: Chapman and Hall, 1989.
13. Hox J. Multilevel analysis: techniques and applications. Mahwah, NJ:
Lawrence Erlbaum Associates, 2002.
14. Bar Oz B, Moretti ME, Mareels G, Van Tittelboom T, Koren G. Report-
ing bias in retrospective ascertainment of drug-induced embryopathy.
15. Armant DR, Saunders DE. Exposure of embryonic cells to alcohol: con-
trasting effects during preimplantation and postimplantation develop-
ment. Semin Perinatol 1996;20:127–139.
16. Kesmodel U, Wisborg K, Olsen SF, Henriksen TB, Secher NJ. Moderate
alcohol intake in pregnancy and the risk of spontaneous abortion. Alco-
hol Alcohol 2002;37:87–92.
17. Windham GC, Von Behren J, Fenster L, Schaefer C, Swan SH. Moder-
ate alcohol consumption and risk of spontaneous abortion. Epidemiol-
18. Lorenzi AR, and Ford HL. 2002. Multiple sclerosis and pregnancy.
Postgrad Med J 2002;78:460–464.
September (2 of 2) 2005