Professional Documents
Culture Documents
CLINICAL—LIVER
Gerardo, Monza, Italy; §§Department of Surgical and Gastroenterological Sciences, University of Padova, Padova, Italy; and 储 储UPRES EA 4266 Université de Franche
Comté et Service d’Hépatologie, CHRU de Besançon, Besançon, France
This article has an accompanying continuing medical education activity on page e13. Learning Objective: Upon
completion of the CME activity, successful learners will be able to prescribe a medical treatment for intrahepatic
cholestasis of pregnancy (ICP).
CLINICAL LIVER
tients to improve pruritus, prevent fetal complications, UDCA alone. Control groups could include several interventions
such as placebo, cholestyramine, dexamethasone, S-adenosyl-
and delay the delivery that definitely eliminates the pru-
methionine, or no specific treatment.
ritus and the risk of IUFD. Such pharmacologic treatment
must be safe for the mother and the baby. In this context, Criteria for Exclusion
ursodeoxycholic acid (UDCA) appeared to be the most Nonrandomized trials and trials that did not use UDCA
efficient pharmacologic treatment.3,11–14 The aim of this in at least 1 studied group were excluded.
systematic review was to evaluate the efficacy of UDCA in
improving pruritus and liver tests, as well as preventing Selection and Data Extraction Bias
poor fetal outcomes in women suffering from ICP. All RCTs considered for inclusion were analyzed inde-
pendently by 3 observers (Y.B., V.D.M., and L.S.), with any
disagreement being discussed by all 3 observers. The decision as
Patients and Methods to inclusion or exclusion was independent of the results.
Meta-analyses were conducted according to a predeter-
mined protocol following the recommendations of Sacks et al15 Statistical Methods
but were limited to English language randomized controlled The response rate was calculated based on the intention-
trials (RCTs). To collect data for the meta-analyses, a standard- to-treat method and using the data collected by the answers to
ized questionnaire was sent to the corresponding author of each the questionnaire. When we failed to contact the corresponding
of the selected RCTs. author, we used the published data. For the 4 end points related
to the fetal outcome (fetal distress, RDS, Apgar score, and
Study Objective and Definition of End Points hospitalization in NICU), the calculations were based on live
births, ie, after excluding cases of IUFD. For each end point, we
The study aimed to identify the effects of UDCA on performed pooled analysis comparing UDCA versus all controls
maternal and fetal conditions in patients suffering from ICP. To including placebo and pooled analysis restricted to RCTs com-
avoid heterogeneity in the definitions of the end points and paring UDCA versus placebo. We used the DerSimonian and
allow further analyses, we sent to each corresponding author Laird model for random effects meta-analysis to obtain sum-
queries regarding each outcome measure, including details of mary estimates across studies.16 Results were expressed by the
their definitions. The queries were related to the number of combined weight-adjusted odds ratios (OR), with their 95%
mothers or newborns in UDCA and control groups and con- confidence interval (95% CI). A P value (from directional zero
cerned the following end points: (1) total resolution of pruritus, effect test) of .05 or less was considered statistically significant.
(2) improvement of pruritus (total resolution or reduction of We tested for heterogeneity by using the Cochran’s Q test, which
pruritus score of more than 50%), (3) normalization of serum follows a 2 distribution to make inferences about the null
ALT, (4) more than 50% decrease in serum ALT levels, (5) more hypothesis of homogeneity (considered significant at P ⬍ .100).
than 50% decrease in serum total bile acid concentrations, (6) For each end point, in case of heterogeneity, the study, or
total prematurity (spontaneous or iatrogenic delivery before 37 studies, showing the greatest difference between the 2 groups in
weeks gestation), (7) spontaneous prematurity, (8) fetal distress favor of UDCA was/were removed from the meta-analysis. This
(meconium staining and/or fetal asphyxia), (9) IUFD, (10) neo- report follows the PRISMA guidelines17 and the Cochrane col-
natal RDS, (11) Apgar score ⬍7 at 5 minutes, and (12) hospi- laboration guidelines for reporting meta-analyses (available at
talization in neonatal intensive care unit (NICU). Maternal and www.cochrane.org).
fetal adverse effects reported in the UDCA group or control
groups were also collected. For pruritus and liver tests (serum Results
ALT and bile acids), the meta-analysis was based on changes
Studied Population
between baseline and day 14 or between baseline and day 21 if
data were not available at day 14. For some outcome measures, Nine RCTs including 489 patients with ICP were
the corresponding authors allowed us to record and report selected and analyzed.18 –26 Out of the 489 patients, 35
unpublished data from their trials. patients who received an association of UDCA and S-
1494 BACQ ET AL GASTROENTEROLOGY Vol. 143, No. 6
adenosyl-methionine were excluded from the meta-analy- 213), and 25.7% (18/70) of cases, respectively. In pooled
sis. Out of the remaining 454 patients, 207 received analyses, UDCA significantly improved pruritus when
UDCA alone, 70 received the placebo alone, 42 received compared both with all controls (OR, 0.27; 95% CI, 0.13–
cholestyramine, 36 received dexamethasone for 1 week 0.55; P ⬍ .0001) and placebo (OR, 0.21; 95% CI, 0.07– 0.62;
and then placebo for 2 weeks, 65 received S-adenosyl- P ⬍ .01) (Figure 1).
methionine, and 34 received no treatment. We sent our Serum ALT. A normalization of serum ALT levels
standardized questionnaire to each corresponding author was observed with UDCA, all controls, and placebo in
of the 9 RCTs, and we received 8 answers. For the remain- 27.8% (48/173), 9.4% (20/213), and 14.3% (10/70) of cases,
ing RCT, we used the data available in the published respectively. The pooled analysis demonstrated that the
article.26 For 2 studies, the data regarding the pruritus rate of serum ALT normalization was higher in the UDCA
outcome and liver tests were available at day 21 (instead of groups than in both all control (OR, 0.23; 95% CI, 0.10 –
day 14).18,19 0.50; P ⬍ .001) and placebo (OR, 0.18; 95% CI, 0.06 – 0.52;
The main characteristics of these 9 RCTs are given in P ⬍ .001) groups.
Table 1. Three RCTs were double blinded.18,20,23 The di- A decrease in serum ALT levels was observed with
agnosis of ICP was always based on the presence of pru- UDCA, all controls, and placebo in 65.9% (114/173),
ritus and abnormal liver tests (ie, serum ALT and bile acid 25.4% (54/213), and 20.0% (14/70) of cases, respectively.
levels). Four RCTs included patients with multiple preg- Weight-adjusted pooled analysis showed that a decrease
nancies (Table 1). There was no difference in the fre- in serum ALT levels was more frequent in the UDCA
quency of multiple pregnancies among the UDCA (7/ groups than in both all control (OR, 0.24; 95% CI, 0.11–
207; 3.4%), all control (7/247; 2.8%), or placebo (2/70; 0.52; P ⬍ .0001) and placebo (OR, 0.12; 95% CI, 0.05– 0.31;
2.9%) groups. The results regarding each outcome mea- P ⬍ .0001) groups (Figure 2).
sure given by the selected studies, including the studies Serum total bile acids. A decrease in serum total
secondarily removed because of heterogeneity, are given bile acid concentrations (with or without normalization)
in Tables 2 and 3. The results of pooled analyses, was observed with UDCA, all controls, and placebo in
including the test of heterogeneity, are summarized in 54.3% (94/173), 24.4% (52/213), and 18.6% (13/70) of
Table 4. cases, respectively. Weight-adjusted pooled analysis show-
Effect of UDCA on Maternal Clinical and ed that this decrease was more frequent in the UDCA
Biochemical Parameters groups than in both all control (OR, 0.37; 95% CI, 0.19 –
Pruritus. A total resolution of pruritus was observed
0.75; P ⬍ .001) and placebo (OR, 0.30; 95% CI, 0.12– 0.73;
with UDCA, all controls, and placebo in 41.6% (72/173), P ⬍ .01) groups.
6.1% (12/213), and 8.6% (6/70) of cases, respectively. It was
more frequently observed with UDCA than in all controls Effect of UDCA on Fetal Outcome
(weight-adjusted OR, 0.23; 95% CI, 0.07– 0.74; P ⬍ .01). In Prematurity. The rate of total prematurity was
studies comparing UDCA versus placebo, the total resolu- 15.9% (33/207), 33.6% (83/247), and 40.0% (28/70) for
tion of pruritus was not more frequent in the UDCA groups UDCA, all control, and placebo groups, respectively.
(OR, 0.39; 95% CI, 0.17–1.32; P ⫽ .13). Weight-adjusted pooled analysis showed this rate to be
An improvement of pruritus was observed with UDCA, lower in UDCA groups than in all control groups (OR,
all controls, and placebo in 61.3% (106/173), 26.8% (57/ 0.44; 95% CI, 0.24 – 0.79; P ⬍ .01) (Figure 3). The total
December 2012 UDCA FOR ICP 1495
Table 2. Meta-analytical Data of Mother Outcome Parameters From RCTs Comparing UDCA vs All Controls and UDCA vs
Placebo
Meta-analyses comparing Meta-analyses comparing
No. of patients UDCA vs all controls UDCA vs placebo
Outcome parameters and UDCA All control Random effect Random effect
studies groups treatments Placebo OR (95% CI) weight for OR (%) OR (95% CI) weight for OR (%)
Total resolution of pruritus
Diafera (1996)18 0/8 0/8 0/8 1.00 (0.02⫺56.46) 7.07 1.00 (0.02⫺56.47) 9.06
Palma (1997)20 4/8 0/7 0/7 0.07 (0.00⫺1.55) 10.70 0.07 (0.00⫺1.55) 14.88
Nicastri (1998)19 6/8 10/16 5/8 0.62 (0.11⫺3.60) 23.66 0.60 (0.08⫺4.38) 37.56
Glantz (2005)23 3/47 2/83 1/47 0.39 (0.07⫺2.06) 25.07 0.41 (0.06⫺2.90) 38.50
Roncaglia (2004)22 4/24 1/22 — 0.32 (0.05⫺2.22) 21.05 — —
Binder (2006)25 22/26 0/25 — 0.00 (0.00⫺0.08) Removed — —
Floreani (1996)21 10/10 0/10 — 0.00 (0.00⫺0.13) Removed — —
Kondrackiene (2005)24 23/42 0/42 — 0.00 (0.0⫺0.17) 12.46 — —
Liu (2006)26 NA NA — NA NA — —
CLINICAL LIVER
Improvement of pruritus
Diafera (1996)18 8/8 8/8 8/8 1.00 (0.02⫺56.46) 3.08 1.00 (0.02⫺56.46) 6.89
Palma (1997)20 8/8 1/7 1/7 0.01 (0.00⫺0.39) Removed 0.01 (0.00⫺0.39) 9.79
Nicastri (1998)19 7/8 10/16 5/8 0.32 (0.04⫺2.41) 10.28 0.31 (0.03⫺2.86) 21.38
Glantz (2005)23 14/47 12/83 4/47 0.40 (0.17⫺0.96) 27.63 0.24 (0.08⫺0.75) 61.94
Roncaglia (2004)22 7/24 5/22 — 0.73 (0.20⫺2.65) 18.83 — —
Binder (2006)25 24/26 11/25 — 0.08 (0.02⫺0.37) 15.38 — —
Floreani (1996)21 10/10 2/10 — 0.01 (0.00⫺033) Removed — —
Kondrackiene (2005)24 28/42 8/42 — 0.13 (0.05⫺0.33) 24.80 — —
Liu (2006)26 NA NA — NA NA — —
Normalization of serum ALT
Diafera (1996)18 7/8 4/8 4/8 0.20 (0.02⫺1.78) 12.30 0.20 (0.02⫺1.78) 21.30
Palma (1997)20 5/8 0/7 0/7 0.04 (0.00⫺1.00) 6.12 0.04 (0.00⫺1.00) 10.85
Nicastri (1998)19 8/8 6/16 2/8 0.04 (0.00⫺0.74) 6.70 0.02 (0.00⫺0.56) 10.58
Glantz (2005)23 11/47 10/83 4/47 0.45 (0.18⫺1.14) 51.12 0.33 (0.10⫺1.06) 57.28
Roncaglia (2004)22 3/24 0/22 — 0.14 (0.01⫺2.80) 6.68 — —
Binder (2006)25 1/26 0/25 — 0.33 (0.01⫺8.57) 5.82 — —
Floreani (1996)21 0/10 0/10 — 1.00 (0.02⫺55.27) 3.85 — —
Kondrackiene (2005)24 13/42 0/42 — 0.03 (0.00⫺0.45) 7.41 — —
Liu (2006)26 NA NA — NA NA — —
Decrease of serum ALT
Diafera (1996)18 8/8 5/8 5/8 0.09 (0.00⫺2.16) 5.39 0.09 (0.00⫺2.16) 8.48
Palma (1997)20 7/8 0/7 0/7 0.01 (0.00⫺0.38) Removed 0.01 (0.00⫺0.38) 7.48
Nicastri (1998)19 8/8 10/16 4/8 0.10 (0.00⫺1.94) 5.82 0.06 (0.00⫺1.36) 8.56
Glantz (2005)23 20/47 17/83 5/47 0.35 (0.16⫺0.77) 32.25 0.17 (0.06⫺0.50) 75.48
Roncaglia (2004)22 6/24 3/22 — 0.51 (0.12⫺2.17) 17.96 — —
Binder (2006)25 24/26 3/25 — 0.02 (0.00⫺0.09) Removed — —
Floreani (1996)21 8/10 7/10 — 0.63 (0.09⫺4.20) 12.40 — —
Kondrackiene (2005)24 33/42 9/42 — 0.08 (0.03⫺0.22) 26.18 — —
Liu (2006)26 NA NA — NA NA — —
Decrease of serum bile acid
concentrations
Diafera (1996)18 8/8 2/8 2/8 0.02 (0.00⫺0.56) Removed 0.02 (0.00⫺0.56) 7.52
Palma (1997)20 4/8 2/7 2/7 0.45 (0.06⫺3.31) 10.21 0.45 (0.06⫺3.31) 19.18
Nicastri (1998)19 7/8 11/16 4/8 0.41 (0.06⫺3.17) 9.87 0.20 (0.02⫺1.78) 15.91
Glantz (2005)23 11/47 15/83 5/47 0.72 (0.30⫺1.70) 31.22 0.41 (0.14⫺1.24) 57.39
Roncaglia (2004)22 5/24 3/22 — 0.64 (0.14⫺2.80) 16.17 — —
Binder (2006)25 24/26 6/25 — 0.03 (0.00⫺0.16) Removed — —
Floreani (1996)21 10/10 5/10 — 0.05 (0.00⫺1.03) 4.73 — —
Kondrackiene (2005)24 25/42 8/42 — 0.17 (0.06⫺0.44) 27.81 — —
Liu (2006)26 NA NA — NA NA — —
NOTE. Empty cases (—) are not applicable. Removed, study removed from the meta-analysis to avoid heterogeneity.
UDCA, ursodeoxycholic acid; NA, data nonavailable after queries to authors; OR, odds ratio.
prematurity rate was not different between UDCA and control, and placebo groups, respectively. Weight-adjusted
placebo groups (OR, 0.59; 95% CI, 0.19 –1.77; P ⫽ .43). pooled analysis showed that there was no difference in the
The rate of spontaneous prematurity was 8.7% (15/ rate of spontaneous prematurity between UDCA and all
173), 19.3% (41/213), and 18.6% (13/70) for UDCA, all control groups (OR, 0.51; 95% CI, 0.22–1.20; P ⫽ .15) or
1496 BACQ ET AL GASTROENTEROLOGY Vol. 143, No. 6
Table 3. Meta-analytical Data of Fetal Outcome Parameters From RCTs Comparing UDCA vs All Controls and UDCA vs
Placebo
Meta-analyses comparing Meta-analyses comparing
No. of patients or babies UDCA vs all controls UDCA vs placebo
Total prematurity
Diafera (1996)18 0/8 8/8 8/8 0.00 (0.00⫺0.20) Removed 0.00 (0.00⫺0.20) Removed
Palma (1997)20 2/8 5/7 5/7 0.17 (0.02⫺1.41) 6.81 0.17 (0.02⫺1.14) 21.29
Nicastri (1998)19 12/8 7/16 4/8 0.49 (0.01⫺2.79) 9.09 0.38 (0.05⫺2.73) 23.46
Glantz (2005)23 2/47 22/83 11/47 0.96 (0.43⫺2.15) 23.20 1.12 (0.44⫺2.82) 55.25
Roncaglia (2004)22 3/24 8/22 — 0.28 (0.68⫺1.14) 12.40 — —
Binder (2006)25 2/26 11/25 — 0.13 (0.03⫺0.58) 11.23 — —
Floreani (1996)21 5/10 4/10 — 1.44 (0.27⫺7.79) 9.58 — —
Kondrackiene (2005)24 3/42 5/42 — 0.60 (0.15⫺2.48) 12.38 — —
Liu (2006)26 4/34 13/34 — 0.24 (0.07⫺0.78) 15.31 — —
Spontaneous prematurity
Diafera (1996)18 0/8 4/8 4/8 0.06 (0.00⫺1.36) 6.39 0.06 (0.00⫺1.36) 9.93
Palma (1997)20 1/8 1/7 1/7 0.87 (0.07⫺10.51) 9.38 0.87 (0.07⫺10.51) 15.16
CLINICAL LIVER
Nicastri (1998)19 1/8 3/16 1/8 0.77 (0.09⫺6.37) 12.15 1.00 (0.08⫺11.93) 15.34
Glantz (2005)23 9/47 16/83 7/47 1.01 (0.41⫺2.46) 31.42 1.33 (0.46⫺3.82) 59.58
Roncaglia (2004)22 0/24 8/22 — 0.03 (0.00⫺0.65) 7.21 — —
Binder (2006)25 0/26 4/25 — 0.09 (0.00⫺1.77) 6.99 — —
Floreani (1996)21 1/10 0/10 — 3.32 (0.12⫺91.60) 5.79 — —
Kondrackiene (2005)24 3/42 5/42 — 0.60 (0.15⫺2.48) 20.67 — —
Liu (2006)26 NA NA — NA NA — —
Fetal distressa
Diafera (1996)18 0/8 3/8 3/8 0.09 (0.00⫺2.16) 3.46 0.09 (0.00⫺2.16) 9.12
Palma (1997)20 1/9 3/6 3/6 0.17 (0.02⫺1.73) 6.03 0.17 (0.02⫺1.73) 15.93
Nicastri (1998)19 2/8 3/16 2/8 1.48 (0.23⫺9.66) 8.24 1.00 (0.13⫺7.87) 18.75
Glantz (2005)23 18/49 30/85 17/48 1.07 (0.52⫺2.21) 23.14 1.06 (0.47⫺2.40) 56.20
Roncaglia (2004)22 2/25 5/22 — 0.38 (0.08⫺1.89) 10.27 — —
Binder (2006)25 3/26 5/25 — 0.56 (0.13⫺2.40) 11.67 — —
Floreani (1996)21 0/10 0/10 — 1.00 (0.02⫺55.27) 2.22 — —
Kondrackiene (2005)24 8/46 14/44 — 0.46 (0.18⫺1.23) 18.47 — —
Liu (2006)26 6/34 21/34 — 0.14 (0.05⫺0.43) 15.50 — —
Respiratory distress
syndromea
Diafera (1996)18 0/8 1/8 1/8 0.29 (0.01⫺8.37) 7.33 0.29 (0.01⫺8.37) 28.50
Palma (1997)20 0/9 0/6 0/6 0.68 (0.01⫺39.07) 5.03 0.68 (0.01⫺39.07) 19.52
Nicastri (1998)19 1/8 1/16 1/8 2.07 (0.18⫺23.36) 13.97 1.0 (0.08⫺11.93) 51.97
Glantz (2005)23 NA NA NA NA NA NA NA
Roncaglia (2004)22 0/25 2/24 — 0.18 (0.01⫺3.3.87) 8.61 — —
Binder (2006)25 2/26 9/25 — 0.18 (0.04⫺0.82) 35.26 — —
Floreani (1996)21 0/10 0/10 — 1.00 (0.02⫺55.27) 5.10 — —
Kondrackiene (2005)24 NA NA — NA NA — —
Liu (2006)26 1/34 7/34 — 0.16 (0.03⫺1.02) 24.70 — —
Apgar score ⬍7 at 5
minutesa
Diafera (1996)18 0/8 4/8 4/8 0.06 (0.00⫺1.36) 5.51 0.06 (0.00⫺1.36) 22.60
Palma (1997)20 0/9 0/6 0/6 0.68 (0.01⫺39.07) 3.31 0.68 (0.01⫺39.07) 13.60
Nicastri (1998)19 0/8 2/16 2/8 0.34 (0.01⫺7.98) 5.45 0.15 (0.01⫺3.77) 21.68
Glantza (2005)23 1/49 3/85 1/48 0.73 (0.10⫺5.10) 14.32 0.98 (0.10⫺9.75) 42.12
Roncaglia (2004)22 2/24 1/22 — 1.67 (0.20⫺13.66) 12.25 — —
Binder (2006)25 2/26 5/25 — 0.38 (0.08⫺1.90) 21.01 — —
Floreani (1996)21 0/10 0/10 — 1.00 (0.02⫺55.27) 3.37 — —
Kondrackiene (2005)24 4/46 7/44 — 0.53 (0.16⫺1.85) 34.77 — —
Liu (2006)26 NA NA — NA NA — —
Hospitalization in neonatal
intensive care unita
Diafera (1996)18 0/8 1/8 1/8 0.29 (0.01⫺8.37) 4.21 0.29 (0.01⫺8.37) 37.24
Palma (1997)20 0/9 0/6 0/6 0.68 (0.01⫺30.07) 2.88 0.68 (0.01⫺30.07) 25.51
Nicastri (1998)19 0/8 1/16 1/8 0.61 (0.02⫺16.61) 4.31 0.29 (0.01⫺8.37) 37.24
Glantz (2005)23 NA NA NA NA NA NA NA
Roncaglia (2004)22 3/25 7/24 — 0.36 (0.09⫺1.49) 23.57 — —
Binder (2006)25 3/26 4/25 — 0.71 (0.16⫺3.24) 20.56 — —
Floreani (1996)21 0/10 0/10 — 1.00 (0.02⫺55.27) 2.93 — —
Kondrackiene (2005)24 6/46 11/44 — 0.47 (0.16⫺1.36) 41.55 — —
Liu (2006)26 NA NA — NA NA — —
NOTE. Empty cases (—) are nonapplicable data. Removed, study removed from the meta-analysis to avoid heterogeneity.
UDCA, ursodeoxycholic acid; NA, data nonavailable after queries to authors; OR, odds ratio.
aFor these parameters, the analysis was based on the number of live newborns (or live fetuses) and did not include the 2 cases of IUFD.
UDCA and placebo groups (OR, 0.88; 95% CI, 0.32–2.43; justed pooled analysis showed this rate was lower in
P ⫽ .93). UDCA groups than in all control groups (OR, 0.46; 95%
Fetal distress. The rate of fetal distress was 18.6% CI, 0.25– 0.86; P ⬍ .01) (Figure 3), but there was no
(40/215), 33.30% (84/252), and 35.7% (25/70) for UDCA, difference between UDCA and placebo groups (OR, 0.63;
all control, and placebo groups, respectively. Weight-ad- 95% CI: 0.23–1.72; P ⫽ .50).
December 2012 UDCA FOR ICP 1497
Heterogeneity Heterogeneity
Outcome parameter Combined OR (95% CI) (Cochran’s Q test) Combined OR (95% CI) (Cochran’s Q test)
Total resolution of pruritus
UDCA vs all controls 0.09 (0.02⫺0.40); P ⬍ .0001 18.94; P ⬍ .01 0.23 (0.07⫺0.74); P ⬍ .01 7.45; P ⫽ .19
UDCA vs placebo 0.39 (0.12⫺1.32); P ⫽ .13 1.61; P ⫽ .66
Improvement of pruritus
UDCA vs all controls 0.19 (0.08⫺0.45); P ⬍ .0001 14.42; P ⫽ .04 0.27 (0.13⫺0.55); P ⬍ .0001 8.38; P ⫽ .14
UDCA vs placebo 0.21 (0.07⫺0.62); P ⬍ .01 3.30; P ⫽ .35
Normalization of serum ALT
UDCA vs all controls 0.23 (0.10⫺0.50); P ⬍ .001 7.47; P ⫽ .38
UDCA vs placebo 0.18 (0.06⫺0.52); P ⬍ .001 3.39; P ⫽ .34
Decrease in serum ALT levels
UDCA vs all controls 0.13 (0.05⫺0.36); P ⬍ .0001 19.26; P ⬍ .01 0.24 (0.11⫺0.52); P ⬍ .0001 8.13; P ⫽ .15
UDCA vs placebo 0.12 (0.05⫺0.31); P ⬍ .0001 2.34; P ⫽ .51
CLINICAL LIVER
Decrease in serum bile acid
concentrations
UDCA vs all controls 0.22 (0.09⫺0.54); P ⬍ .0001 17.72; P ⫽ .01 0.37 (0.19⫺0.75); P ⬍ .001 7.05; P ⫽ .22
UDCA vs placebo 3.32 (1.38⫺8.06); P ⬍ .01 3.11; P ⫽ .38
Total prematurity
UDCA vs all controls 0.37 (0.18⫺0.75); P ⬍ .001 16.02; P ⫽ .04 0.44 (0.24⫺0.79); P ⬍ .01 10.35; P ⫽ .17
UDCA vs placebo 0.25 (0.04⫺1.45); P ⫽ .20 9.55; P ⫽ .02 0.59 (0.19⫺1.77); P ⫽ .43 3.03; P ⫽ .22
Spontaneous prematurity
UDCA vs all controls 0.51 (0.22⫺1.20); P ⫽ .15 9.74; P ⫽ .20
UDCA vs placebo 0.88 (0.32⫺2.43); P ⫽ .93 3.42; P ⫽ .33
Fetal distress
UDCA vs all controls 0.46 (0.25⫺0.86); P ⬍ .01 12.70; P ⫽ .12
UDCA vs placebo 0.63 (0.23⫺1.72); P ⫽ .50 3.97; P ⫽ .26
Respiratory distress syndrome
UDCA vs all controls 0.30 (0.12⫺0.74); P ⬍ .01 3.93; P ⫽ .69
UDCA vs placebo 0.66 (0.11⫺3.91); P ⫽ .64 0.33; P ⫽ .85
Apgar score ⬍7 at 5 minutes
UDCA vs all controls 0.53 (0.25⫺1.10); P ⫽ .09 3.48; P ⫽ .84
UDCA vs placebo 0.33 (0.07⫺1.47); P ⫽ .15 2.37; P ⫽ .50
Hospitalization in neonatal intensive
care unit
UDCA vs all controls 0.49 (0.25⫺0.98); P ⫽ .04 0.67; P ⫽ .10
UDCA vs placebo 0.36 (0.05⫺2.81); P ⫽ .33 0.12; P ⫽ .94
RDS. The rate of RDS was 3.3% (4/120), 16.3% NICU to be less frequent in UDCA groups than in all control
(20/123), and 9.1% (2/22) for UDCA, all control, and groups (OR, 0.49; 95% CI, 0.25– 0.98; P ⫽ .046) (Figure 3).
placebo groups, respectively. Weight-adjusted pooled There was no difference in this rate between UDCA and
analysis showed this rate was lower in UDCA groups than placebo groups (OR, 0.36; 95% CI, 0.05–2.81; P ⫽ .33).
in all control groups (OR, 0.30; 95% CI, 0.12– 0.74; P ⬍ IUFD. Two cases of IUFD were observed in the
.01) (Figure 3) but was not different between UDCA and selected RCTs, each of them in a placebo group.20,23 The
placebo groups (OR, 0.66; 95% CI, 0.11–3.91; P ⫽ .64). small number of events did not allow specific analysis on
Apgar score. The rate of Apgar score ⬍7 at 5 this point.
minutes was 5.0% (9/181), 10.1% (22/218), and 10.0%
(7/70) for UDCA, all control, and placebo groups, respec- Maternal and Fetal Adverse Effects
tively. Weight-adjusted pooled analysis showed that this Maternal adverse effects were observed in 1% (2/
rate was not significantly different between UDCA and all 207), 5.7% (14/247), and 1.4% (1/70) of cases in UDCA, all
control groups (OR, 0.53; 95% CI, 0.25–1.11; P ⫽ .09) or control, and placebo groups, respectively. Weight-adjusted
between UDCA and placebo groups (OR, 0.33; 95% CI, pooled analysis showed that the number of maternal
0.07–1.47; P ⫽ .15). adverse effects was not different between UDCA and all
Hospitalization in NICU. The rate of hospitalization control groups (OR, 0.70; 95% CI, 0.23–2.12; P ⫽ .53) or
in NICU was 9.1% (12/132), 18.1% (24/133), and 9.1% (2/22) between UDCA and placebo groups (OR, 1.27; 95% CI,
for UDCA, all control, and placebo groups, respectively. 0.26 – 6.19; P ⫽ .76). The adverse effects reported in UDCA
Weight-adjusted pooled analysis showed hospitalization in groups were diarrhea (n ⫽ 1) and transient nausea and
1498 BACQ ET AL GASTROENTEROLOGY Vol. 143, No. 6
vomiting (n ⫽ 1). No fetal adverse effects were observed in RCTs are difficult to carry out with pregnant women, par-
any group of these RCTs. ticularly in a relatively rare disease such as ICP, we decided to
perform a new systematic review with meta-analyses using
Discussion data from a standardized questionnaire sent to each corre-
Although UDCA appears as the most efficient sponding author of the published RCTs. This enabled us to
treatment of ICP, the evidence of its benefit on pruritus is collect exhaustive information with homogenized defini-
still debated, and the only previous meta-analysis failed to tions of end points from 8 of the 9 published RCTs to date
provide significant results.27 Moreover, the role of UDCA and to perform specific meta-analyses on 12 outcome mea-
in preventing severe fetal outcome has never been dem- sures, in an individual data-based fashion.
onstrated.28 In the previously published systematic review Our study provides 3 relevant findings: (1) it demonstrates
on this topic led by Burrows et al (update available on that UDCA is efficient for improving pruritus and liver tests
www.cochrane.org), the authors concluded that “inconsis- in women with ICP; (2) it provides specific analyses on fetal
tent and inadequate reporting of results precluded pool- outcome that strongly suggest a benefit of UDCA; and (3) it
ing the results of small studies.”27 However, this review confirms the good safety profile of UDCA.
was based only on data from the published papers and Our results clearly demonstrated the benefit of UDCA
was hampered by missing information and by the hetero- on both liver tests studied (serum ALT and serum bile
geneity of the end points used between trials.27 Because large acid concentrations). These results were robust because
CLINICAL LIVER
Figure 3. Effects of UDCA ther-
apy on fetal outcome in ICP:
RCTs comparing UDCA vs all
controls. (A) Effects on total pre-
maturity. (B) Effects on fetal dis-
tress. (C) Effects on respiratory
distress syndrome. (D) Effects
on hospitalization in neonatal in-
tensive care unit.
the benefit of UDCA was significant when compared with deed, although ICP is a benign disease for the mother (ie,
all control treatments, as well as with placebo, despite the with no risk of liver failure), it is worthwhile for the
small sample size in the latter comparison. The effect of clinician to have efficient treatment at their disposal for
UDCA on total serum bile acid levels should be particu- this very painful symptom. The need for efficient treat-
larly relevant for the prognosis of the fetus.29 The effect of ment against pruritus may have made practitioners reluc-
UDCA on pruritus (both for total resolution and im- tant to conduct large studies of UDCA versus placebo,
provement) was significant when comparing UDCA ver- thus hampering a definitive evidence-based demonstra-
sus all control treatments. The improvement of pruritus tion of the positive effect of UDCA.
was also significant when comparing UDCA versus pla- Despite this difficult context for providing evidence-
cebo. However, the effect of UDCA versus placebo was no based results, our meta-analysis has the merit to strongly
longer significant when considering the total resolution suggest that UDCA is beneficial for fetal outcome. Indeed,
of pruritus, which is a less subjective but more stringent in this meta-analysis, the use of UDCA was associated
outcome measure. This result was possibly the conse- with a decrease of total prematurity; fetal distress; RDS;
quence of a type 2 error. This beneficial effect of UDCA on and, as a consequence, a decrease of the number of hos-
pruritus is of great importance for clinical practice. In- pitalizations in NICU in comparison with other treat-
1500 BACQ ET AL GASTROENTEROLOGY Vol. 143, No. 6
ments. Among these fetal outcome parameters, the de- 2 error, and the small number of women receiving placebo
creases of total prematurity is of importance because it is can be explained by the context of pregnancy, the uncom-
likely that it contributes to the improvement of the other fortable symptoms, and the reluctance of the obstetrical
fetal outcome parameters that are expected consequences team to leave the mother without efficient treatments.
of the prematurity. Moreover, although no statistical dif- Second, for some outcome measures, pooled analyses pro-
ference was found regarding the rate of Apgar score ⬍7 at vided heterogeneity for which we failed to find 1 relevant
5 minutes, the data showed there was a trend in favor of source (as available in Table 1). Thus, in case of hetero-
UDCA for this parameter, which is a well-known predictor geneity, we pragmatically chose to exclude the outlying
of the neurologic development in infants.30 The benefit of studies (ie, with the more marked positive differences)
UDCA was not significant for any of these items in studies favoring UDCA for each end point considered in the
comparing UDCA versus placebo, probably because of a analysis. By using this method, we avoided, as much as
type 2 error. possible, the impact of positive studies with ORs that may
Finally, this systematic review confirms that UDCA is be considered as overestimated, perhaps resulting from
well tolerated and that maternal adverse effects are rare imperfect study design, and participating in publication
and minor. Moreover, no fetal adverse effects were ob- biases. The combined ORs before and after removing
CLINICAL LIVER
served in any group of these RCTs included in this meta- these outlying studies from analyses had the same statis-
analysis and, to our knowledge, in any ICP studies or cases tical significance (Table 4). Third, we failed to demon-
reported. Long-term data regarding the safety of UDCA in strate a positive effect of UDCA on IUFD, which is the
later childhood are lacking, but a follow-up of 26 children most feared consequence of ICP but which, fortunately, is
(1 infant, 12 preschool children, and 13 school children a rare event. In this systematic review, IUFD was indeed
aged up to 12 years) showed that these children developed too rare to allow any specific analyses. It only occurred in
normally.13 women receiving placebo, but no conclusion can be drawn
The mechanism of this beneficial effect of UDCA in concerning this dramatic event because some cases of
patients suffering from ICP and their fetus remains spec- IUFD have also been reported in women with ICP receiv-
ulative. As with chronic liver disease, UDCA, a hydrophilic ing UDCA.38
bile acid, may decrease liver damage in the mother by In conclusion, this systematic review supports the use
affording a cytoprotection against the hepatotoxic effects of UDCA as initial treatment in patients with ICP because
of the hydrophobic bile acids and by improving the trans- UDCA improves liver tests and is currently the most
port and secretion of hepatobiliary bile acid.31 In ICP, the efficient treatment of pruritus in these patients. The re-
most important mechanism seems to be the stimulation sults of this meta-analysis also suggest that UDCA is
of impaired hepatocellular secretion.32,33 UDCA may also beneficial for fetal outcome.
have a specific effect by improving the transport of bile
acid across the placenta.34,35 The improvement of pruritus
References
under UDCA therapy in patients with ICP contrasts with
1. Bacq Y, Sapey T, Brechot MC, et al. Intrahepatic cholestasis of
the absence of a significant effect in patients with primary pregnancy: a French prospective study. Hepatology 1997;26:
biliary cirrhosis.36 However, ICP is a disease with a short 358 –364.
duration, whereas primary biliary cirrhosis is a chronic 2. Reyes H. Review: intrahepatic cholestasis. A puzzling disorder of
disease with inflammatory and fibrotic damage of the pregnancy. J Gastroenterol Hepatol 1997;12:211–216.
interlobular bile ducts; additionally, the mechanism of 3. Geenes V, Williamson C. Intrahepatic cholestasis of pregnancy.
World J Gastroenterol 2009;15:2049 –2066.
pruritus may be different between these 2 cholestatic liver
4. Reyes H, Gonzalez MC, Ribalta J, et al. Prevalence of intrahepatic
diseases. The improvement of fetal outcome could directly cholestasis of pregnancy in Chile. Ann Intern Med 1978;88:487–
result in a decrease of maternal bile acid concentrations 493.
related to UDCA because a relationship has already been 5. Reyes H, Taboada G, Ribalta J. Prevalence of intrahepatic choles-
found between maternal bile acid concentrations and tasis of pregnancy in La Paz, Bolivia. J Chronic Dis 1979;32:499 –
504.
some fetal complications like fetal distress or RDS.29,37
6. Lee RH, Goodwin TM, Greenspoon J, et al. The prevalence of
This improvement may also be, at least in part, secondary intrahepatic cholestasis of pregnancy in a primarily Latina Los
to the improvement of the maternal condition. Indeed, Angeles population. J Perinatol 2006;26:527–532.
when the the pruritus and liver test results improved 7. Lammert F, Marschall H-U, Glantz A, et al. Intrahepatic cholestasis
under medical treatment, the obstetrical team may have of pregnancy: molecular pathogenesis, diagnosis and manage-
ment. J Hepatol 2000;33:1012–1021.
chosen to delay the delivery, which may, in turn, contrib-
8. Royal College of Obstetricians and Gynaecologists. Obstetric
ute to the decrease of the rate of prematurity. cholestasis. RCOG 2011;43:1–14.
Our study acknowledges several limitations. First, for 9. Saleh MM, Abdo KR. Consensus on the management of obstetric
some specific outcome measures, the positive effect of cholestasis: National UK survey. BJOG 2007;114:99 –103.
UDCA compared with control treatments was no longer 10. Sentilhes L, Bacq Y. La Cholestase Intrahépatique Gravidique. J
Gynecol Obstet Biol Reprod (Paris) 2008;37:118 –126.
significant when restricting the comparison to placebo.
11. Pusl T, Beuers U. Intrahepatic cholestasis of pregnancy. Orphanet
This was observed for the total resolution of pruritus as J Rare Dis 2007;2:26.
well as for all the studied parameters regarding fetal 12. Pathak B, Sheibani L, Lee RH. Cholestasis of pregnancy. Obstet
outcome. The discrepancies were possibly because of type Gynecol Clin North Am 2010;37:269 –282.
December 2012 UDCA FOR ICP 1501
13. Zapata R, Sandoval L, Palma J, et al. Ursodeoxycholic acid in the 28. Gurung V, Williamson C, Chappell L, et al. Pilot study for a trial of
treatment of intrahepatic cholestasis of pregnancy. A 12-year ursodeoxycholic acid and/or early delivery for obstetric cholesta-
experience. Liver Int 2005;25:548 –554. sis. BMC Pregnancy Childbirth 2009;9:19.
14. European Association for the Study of the Liver. EASL Clinical 29. Glantz A, Marschall HU, Mattsson LA. Intrahepatic cholestasis of
Practice Guidelines: management of cholestatic liver diseases. pregnancy: relationships between bile acid levels and fetal com-
J Hepatol 2009;51:237–267. plication rates. Hepatology 2004;40:467– 474.
15. Sacks HS, Berrier J, Reitman D, et al. Meta-analyses of random- 30. Casey BM, McIntire DD, Leveno KJ. The continuing value of the
ized controlled trials. N Engl J Med 1987;316:450 – 455. Apgar score for the assessment of newborn infants. N Engl J Med
16. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control 2001;344:467– 471.
31. Paumgartner G, Beuers U. Ursodeoxycholic acid in cholestatic
Clin Trials 1986;7:177–188.
liver disease: mechanisms of action and therapeutic use revis-
17. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for
ited. Hepatology 2002;36:525–531.
systematic reviews and meta-analyses: the PRISMA statement.
32. Beuers U. Drug insight: mechanisms and sites of action of ur-
BMJ 2009;339:b2535.
sodeoxycholic acid in cholestasis. Nat Clin Pract Gastroenterol
18. Diaferia A, Nicastri PL, Tartagni M, et al. Ursodeoxycholic acid Hepatol 2006;3:318 –328.
therapy in pregnant women with cholestasis. Int J Gynaecol Obstet 33. Arrese M, Macias RI, Briz O, et al. Molecular pathogenesis of
1996;52:133–140. intrahepatic cholestasis of pregnancy. Expert Rev Mol Med 2008;
19. Nicastri PL, Diaferia A, Tartagni M, et al. A randomised placebo- 10:e9.
CLINICAL LIVER
controlled trial of ursodeoxycholic acid and S-adenosylmethionine 34. Azzaroli F, Mennone A, Feletti V, et al. Clinical trial: modulation of
in the treatment of intrahepatic cholestasis of pregnancy. Br J human placental multidrug resistance proteins in cholestasis of
Obstet Gynaecol 1998;105:1205–1207. pregnancy by ursodeoxycholic acid. Aliment Pharmacol Ther 2007;
20. Palma J, Reyes H, Ribalta J, et al. Ursodeoxycholic acid in the 26:1139 –1146.
treatment of cholestasis of pregnancy: a randomized, double-blind 35. Serrano MA, Brites D, Larena MG, et al. Beneficial effect of
study controlled with placebo. J Hepatol 1997;27:1022–1028. ursodeoxycholic acid on alterations induced by cholestasis of
21. Floreani A, Paternoster D, Melis A, et al. S-adenosylmethionine pregnancy in bile acid transport across the human placenta.
versus ursodeoxycholic acid in the treatment of intrahepatic J Hepatol 1998;28:829 – 839.
cholestasis of pregnancy: preliminary results of a controlled trial. 36. Gong Y, Huang ZB, Christensen E, et al. Ursodeoxycholic acid for
Eur J Obstet Gynecol Reprod Biol 1996;67:109 –113. primary biliary cirrhosis. Cochrane Database Syst Rev 2008:
22. Roncaglia N, Locatelli A, Arreghini A, et al. A randomised con- CD000551.
trolled trial of ursodeoxycholic acid and S-adenosyl-l-methionine in 37. Laatikainen T, Tulenheimo A. Maternal serum bile acid levels and
the treatment of gestational cholestasis. BJOG 2004;111:17–21. fetal distress in cholestasis of pregnancy. Int J Gynaecol Obstet
1984;22:91–94.
23. Glantz A, Marschall HU, Lammert F, et al. Intrahepatic cholestasis of
38. Sentilhes L, Verspyck E, Pia P, et al. Fetal death in a patient with
pregnancy: a randomized controlled trial comparing dexamethasone
intrahepatic cholestasis of pregnancy. Obstet Gynecol 2006;107:
and ursodeoxycholic acid. Hepatology 2005;42:1399 –1405.
458 – 460.
24. Kondrackiene J, Beuers U, Kupcinskas L. Efficacy and safety of
ursodeoxycholic acid versus cholestyramine in intrahepatic
cholestasis of pregnancy. Gastroenterology 2005;129:894 –901. Received November 23, 2011. Accepted August 7, 2012.
25. Binder T, Salaj P, Zima T, et al. Randomized prospective compar-
ative study of ursodeoxycholic acid and S-adenosyl-L-methionine Reprint requests
in the treatment of intrahepatic cholestasis of pregnancy. J Peri- Address requests for reprints to: Yannick Bacq, MD, Service
nat Med 2006;34:383–391. d’Hépatogastroentérologie, CHRU de Tours, 37044 Tours Cedex,
26. Liu Y, Qiao F, Liu H, et al. Ursodeoxycholic acid in the treatment of France. e-mail: bacq@med.univ-tours.fr; fax: (33) 2 47478428.
intraheptic cholestasis of pregnancy. J Huazhong Univ Sci Tech- Acknowledgments
nolog Med Sci 2006;26:350 –352. The authors thank Adam Smith for reviewing the English text.
27. Burrows RF, Clavisi O, Burrows E. Interventions for treating choles-
tasis in pregnancy. Cochrane Database Syst Rev 2001: Conflicts of interest
CD000493. The authors disclose no conflicts.