Impact of pregnancy on the clinical activity of Crohn's disease

Authors


Professor L. Beaugerie, Service de Gastro-entérologie, Hôpital Saint-Antoine, 184 rue du faubourg Saint-Antoine, 75571 Paris Cedex 12, France.
E-mail: laurent.beaugerie@sat.ap-hop-paris.fr

Summary

Background : The impact of pregnancy on Crohn's disease activity has been poorly investigated.

Aim : To determine the effect of pregnancy on Crohn's disease activity from the retrospective analysis of a cohort of women who had a regular clinical follow-up.

Methods : Seventy pregnancies occurring in 61 women were studied. The Harvey–Bradshaw index was determined during the four quarters preceding each pregnancy, the three quarters of pregnancy and the four quarters following delivery.

Results : The mean Harvey–Bradshaw index during pregnancy [0.68 (0.18), mean (S.E.M.)] was significantly lower than that of the year preceding pregnancy [0.98 (0.16), P = 0.03] and that of the year following delivery [1.10 (0.17), P = 0.04]. In non-smoking women (48 pregnancies), there was no significant change of Harvey–Bradshaw index between these intervals. Whereas in those who smoked (22 pregnancies), most of whom reduced tobacco consumption during pregnancy, the mean Harvey–Bradshaw index during pregnancy was significantly reduced compared with that of the year following delivery [0.58 (0.20) vs. 1.60 (0.33), P = 0.01]. The use of drugs was significantly lower during pregnancy.

Conclusions : Crohn's disease activity is mildly but significantly lower during pregnancy. The reduction of tobacco consumption during pregnancy in smoking women may play an important role in this improvement.

Introduction

Crohn's disease (CD) often affects women in their reproductive years. These patients frequently ask questions about the relationships between CD and pregnancy. It is established that women with inactive disease have normal fertility although fertility is decreased in those with active disease, and that there is no increased risk of congenital malformations.1–4 It is also generally accepted that there is a significant risk of preterm delivery and low birth weight in mothers with CD, particularly when CD is active at conception.2–4

In contrast, it is not clear whether pregnancy affects the course of CD. In most series, it has been found that CD activity at conception correlates with CD activity during pregnancy and the postpartum. In these works, women with quiescent disease at conception usually remain in remission;5–7 if they occurred, relapses are observed during the first quarter of pregnancy6, 8, 9 and the postpartum.5, 8–11 Conversely, women with active CD at conception are likely to have an active disease throughout their pregnancy.1, 5, 6 In other studies, however, CD activity at conception did not correlate with disease activity during pregnancy11, 12 and the postpartum did not appear to be a period of increased risk of relapse.6, 12 Finally, it has been suggested in more recent studies that activity of CD could be reduced during pregnancy.13, 14

In these series, most of them published before the 1990s, disease activity was not regularly monitored using a validated clinical activity index. In addition, smoking status was not taken into account. In the recent years, it has been demonstrated that CD course is strongly influenced by smoking,15, 16 particularly in women.15 As women often change their smoking habits during pregnancy, this factor could be implicated in the interaction between pregnancy and CD activity.

We thus decided to study the impact of pregnancy on CD activity, taking into account smoking habits, in a cohort of women who were carefully followed-up during pregnancy, as well as during the year before and the year after pregnancy.

Methods

Patient selection

We considered for inclusion in the study all women with CD born between 1948 and 1984 regularly seen in our two inflammatory bowel disease (IBD) tertiary centres. In Rothschild hospital, selection of women was made by systematically examining the medical records, whereas in Saint Louis hospital, a questionnaire was sent to all the women fulfilling the preinclusion criteria. Pregnancies were analysed if all the following criteria were fulfilled: (i) onset of pregnancy at least 1 year after the diagnosis of CD; (ii) at least 1 year between pregnancies in women who had more than one pregnancy; (iii) complete record of clinical data at each visit in the unit, allowing the retrospective calculation of at least one Harvey–Bradshaw index (HBI)17 for each quarter of the study periods, including the year preceding pregnancy, pregnancy and the year following pregnancy; (iv) detailed information about tobacco consumption during all the study-periods, obtained by questionnaire mailed to the patients (Saint-Louis hospital) or phone call (Rothschild hospital).

Crohn's disease was considered as active in patients with a HBI above 4, inactive for a HBI between 1 and 3, and quiescent for a HBI = 0. Smoking status could be extracted from the medical charts for all patients.

Statistical methods

Non-parametric Friedman test was used for comparison of quantitative data between all study groups and Wilcoxon test was used for the comparison between two groups. McNemar's test was used to compare treatments used between periods. P-values < 0.05 were considered to denote significant differences. Calculations were performed using Statview statistical software (version 5.0, SAS Institute Inc., Cary, NC, USA). Results were expressed as mean with S.E.M.

Results

Patient population

Out of the 210 women's files reviewed at Rothschild hospital, 53 pregnancies occurring in 45 women fulfilled the inclusion criteria. Among the 71 women from Saint-Louis hospital who answered to the questionnaire, 17 pregnancies occurring in 16 women fulfilled these criteria. A total of 70 pregnancies that occurred in 61 women were thus studied. Characteristics of patients at the onset of pregnancy are shown in Table 1. Twenty-two pregnancies occurred in smoking women. Thirteen of them stopped smoking or reduced of more of 80% their tobacco consumption during pregnancy and restarted smoking in the weeks following delivery. In these women, the mean daily cigarette consumption was 16 cigarettes/day during the year preceding pregnancy, 1 cigarette/day during pregnancy and 14 cigarettes/day during the year following delivery. Three women stopped smoking during pregnancy and did not restart during the year following delivery. In these women, mean daily cigarette consumption was 7 cigarettes/day during the year preceding pregnancy. Six women did not stop smoking during pregnancy. In these women, mean daily cigarette consumption was 17 cigarettes/day during the year preceding pregnancy, 12 cigarettes/day during pregnancy and 17 cigarettes/day during the year following delivery.

Table 1.  Baseline characteristics of the patients at the onset of pregnancy (n = 70 pregnancies)
Age (years), mean (range)30 (23–42)
Duration of Crohn's disease (years), mean (range)7 (1–25)
Previous intestinal resection, n (%)34 (48.6)
Previous perianal lesions, n (%)18 (25.7)
Active perianal disease, n (%)1 (1.2)
Location of disease
 Small bowel only, n (%)17 (24.3)
 Colon only, n (%)18 (25.7)
 Small bowel and colon, n (%)35 (50.0)
Tobacco status
 Smoker (%)22 (31.4)
 Non-smoker (%)48 (68.6)

Data on pregnancy were available for 60 pregnancies. The rate of caesarean section was 38%. Caesarean section was indicated in six cases because of perianal disease. There was 32 (53%) boys and 28 (47%) girls. Two malformations were observed: one case of anencephalia leading to a miscarriage during the third trimester and one case of pylorus stenosis.

Crohn's disease activity

Mean HBI for each quarter of the year preceding pregnancy, for pregnancy and for the year following delivery, according to tobacco status, is shown in Figure 1. Mean HBI during pregnancy was significantly lower than that of the year preceding pregnancy and that of the year following delivery (Table 2). In the subgroup of non-smoking women, there was no significant change of HBI between the three periods. In smoking women, mean HBI during pregnancy was significantly lower than that of the year following delivery. Distribution of disease activity status for each quarter of the year preceding pregnancy, for pregnancy, and for the year following delivery, is shown in Figure 2. The number of women with asymptomatic disease was higher during the second quarter of pregnancy.

Figure 1.

Mean Harvey–Bradshaw index in the quarters of the year before pregnancy, of pregnancy, and of the year after pregnancy. Lines with diamonds, squares and triangles refer to total population (n = 70 pregnancies), smokers (n = 22) and non-smokers (n = 48), respectively. The rectangle figures the pregnancy period.

Table 2.  Comparison of mean Harvey–Bradshaw index between the three study periods according to tobacco status, mean (S.E.M.)
 Year before pregnancyPregnancyYear after pregnancy
  1. adifferent fromb, P = 0.03; bdifferent fromc, P = 0.04; ddifferent frome, P = 0.01.

Total population (n = 70 pregnancies)0.98a (0.16)0.68b (0.18)1.10c (0.17)
Smoking women (n = 22 pregnancies)1.14 (0.32)0.58d (0.20)1.60e (0.33)
Non-smoking women (n = 48 pregnancies)0.90 (0.19)0.73 (0.24)0.81 (0.19)
Figure 2.

Distribution of disease activity status in the quarters of the year before pregnancy, pregnancy and the year following pregnancy.

Treatment

The proportion of women treated with aminosalicylates, steroids or azathioprine during the year preceding pregnancy, pregnancy and the year following delivery, is shown in Table 3. The use of salicylates and azathioprine was significantly reduced during pregnancy compared with the year preceding pregnancy and the year following delivery. Indication for steroids was a flare-up of the disease, completion of treatment started during the year preceding pregnancy and steroid dependency in five, two and nine patients, respectively. Nine patients undergone surgery during the study period, three during the year preceding pregnancy and six during the year following delivery. Surgery consisted in ileal or ileocolic resection, coloproctectomy and drainage of a perianal abscess in five, two and two cases, respectively.

Table 3.  Treatments used during the year before pregnancy, pregnancy and the year following pregnancy (n = 70 pregnancies)
 Year before pregnancyPregnancyYear after pregnancy
  1. adifferent fromb, P < 0.001; bdifferent fromc, P < 0.001; ddifferent frome, P < 0.001; edifferent fromf, P = 0.01.

Salicylates, n(%)45 (64.3)a29 (41.4)b40 (57.1)c
Systemic steroids, n (%)20 (28.6)13 (18.6)16 (22.9)
Azathioprine, n (%)19 (27.1)d8 (11.4)e17 (24.2)f

Discussion

We have shown in this study that clinical CD activity is mildly but significantly reduced during pregnancy in the general population of women. We did not observe a significant change in the subset of pregnancies with no tobacco use throughout pregnancy. By contrast, mean clinical activity was significantly reduced in pregnancies occurring in smoking women. Reduction of tobacco consumption during pregnancy in most of these women could play an important role in the clinical improvement.

Our study is the first one devoted to the relationships between pregnancy and CD activity that takes into account smoking status. It is now well established that smoking alters the course of CD.18 Smokers have a more severe disease and require more steroids and immunosuppressive drugs.15 In patients who stop smoking for more than 1 year, the risk of flare up and the use of steroids or immunosuppressive drugs do not differ from that of non-smokers and are lower than in continuing smokers.16 In our study, CD activity during pregnancy was only reduced in the subgroup of women who significantly reduced their tobacco consumption during pregnancy. This result is consistent with the hypothesis of a beneficial short-term effect of smoking reduction on the course of CD during pregnancy. In addition, the tendency towards an exacerbation of CD activity in the postpartum state in women who reduced their tobacco consumption during pregnancy could be related to smoking resumption after delivery.

The methodological limits of our work must be underlined. We chose to use the HBI as this index has been validated as a ‘simple index’ that can be used in retrospective studies and that correlates well with the CDAI.19 In addition, the HBI does not include biological parameters, which is a good point, as the blood concentration of several biological components (haemoglobin, albumin), but not CRP, changes during normal pregnancies.20, 21 However, the HBI has not been specifically validated in pregnancy.

The overall reduction of disease activity during pregnancy that we observed cannot be attributed to an intensification of medical or surgical treatment during pregnancy. No woman underwent surgery during pregnancy and the overall use of treatment (salicylates, steroids or azathioprine) was lower during pregnancy.

In our study, there was no significant difference for disease activity between the year preceding pregnancy and the year following delivery. This suggests that pregnancy does not affect the short-term course of CD after delivery. Our data are consistent with the result of a study by Nielsen et al.9 These authors observed in 109 pregnancies no significant change in the risk of exacerbation per patient-year between the 6-month period preceding pregnancy and the period including pregnancy and the 6-month period following delivery. By contrast, Castiglione et al.14 found a significant reduction in the annual relapse rate between the 3-year period preceding pregnancy and the 3-year period following delivery. Of note, in this study, only 17 pregnancies were studied and there was no information about tobacco consumption.

The influence of pregnancy on the course of other diseases than IBDs has been reported in the literature. Pregnancy is frequently associated with a significant clinical improvement of the disease in rheumatoid arthritis, especially in cases of maternal–foetal disparity in HLA-DR or DQ alleles.22 In a recent work, Kane et al.23 have shown that improvement of IBD symptoms during pregnancy is associated with disparity in HLA class II antigens between mother and foetus. In rheumatoid arthritis, it has been postulated that foetal DQ alpha peptides might correct autoimmunity in patients with rheumatoid arthritis, either by inducing maternal regulatory T cells or by affecting the maternal T-cell receptor repertoire.22 Another hypothesis is the induction of regulatory cells during pregnancy, which would suppress T cells response by the production of interleukin-10 and tumour growth factor-beta.24 In multiple sclerosis, Confavreux et al.25 have shown that the rate of relapse declines during pregnancy. The authors suggest that this phenomenon could be related to the secretion by the foetal-placental unit of cytokines such as interleukin-10 that down regulates the production of other cytokines. In CD, modifications of the immune response related to pregnancy are unknown. We did not observe a significant alteration of CD activity in the subset of non-smoking women.

Women with CD have been discouraged for decades from procreating when disease active, as the risk of foetal loss and premature delivery is markedly increased in this context.2 As a consequence, in our study like in others, only a few women had active CD at conception, making not possible to specifically study the effect of pregnancy on CD course in this setting.

In conclusion, our study suggests that CD activity does not intrinsically increase during pregnancy and is reduced in smokers who significantly reduce their tobacco consumption during pregnancy.

Ancillary