Hum. Reprod. 2006 Costello 1387 99

8/3/2019 Hum. Reprod. 2006 Costello 1387 99

http://slidepdf.com/reader/full/hum-reprod-2006-costello-1387-99 1/13

Human Reproduction Vol.21, No.6 pp. 1387–1399, 2006 doi:10.1093/humrep/dei5

Advance Access publication January 31, 2006.

© The Author 2006. Published by Oxford Universi ty Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved. 13For Permissions, please email: journals. [email protected]

A systematic review and meta-analysis of randomizedcontrolled trials on metformin co-administration duringgonadotrophin ovulation induction or IVF in women withpolycystic ovary syndrome

Michael F.Costello1,2,3,4, Michael Chapman1,3 and Una Conway1,3

1Division of Obstetrics and Gynaecology, School of Women’s and Children’s Health, University of New South Wales,2Department of Reproductive Medicine, Royal Hospital for Women and, 3IVFAustralia, Sydney, NSW, Australia

4To whom correspondence should be addressed at: Division of Obstetrics and Gynaecology, School of Women’s and Children’s Health

Level 1 Women’s Health Institute, Royal Hospital for Women, Locked Bag 2000, Randwick, Sydney, NSW 2031, Australia. E-mail:

[email protected]

BACKGROUND: A systematic review of randomized controlled trials (RCTs) comparing whether metformin c

administration with gonadotrophins for ovulation induction (OI) with timed intercourse or IVF improves outcome women with polycystic ovary syndrome (PCOS). METHODS: The quality of reporting of meta-analyses (QUOROM

guidelines were followed. A systematic computerized literature search of three bibliographic databases was performe

RESULTS: Eight RCTs were included in the overall review. Meta-analysis demonstrated that the co-administratio

of metformin to gonadotrophin OI does not significantly improve ovulation [odds ratio (OR) = 3.27; 95% confiden

interval (95% CI) = 0.31–34.72] or pregnancy (OR = 3.46; 95% CI = 0.98–12.2) rates. Metformin co-administration

IVF treatment does not improve pregnancy (OR = 1.29; 95% CI = 0.84–1.98) or live birth (OR = 2.02, 95% CI = 0.98–4.14

rates but reduces the risk of ovarian hyperstimulation syndrome (OHSS) (OR = 0.21; 95% CI = 0.11–0.41, P < 0.00001

CONCLUSIONS: Current data on the use of metformin in the gonadotrophin OI or IVF treatment settings ar

inconclusive because of the review’s failure to exclude an important clinical treatment effect. Further RCTs ar

necessary to definitively clarify whether metformin co-administration during gonadotrophin OI or IVF w

improve the efficacy of these treatments in PCOS women.

Key words: FSH/gonadotrophins/IVF/metformin/ovulation induction/polycystic ovary syndrome

Introduction

Polycystic ovary syndrome (PCOS) is characterized by chronic

anovulation and hyperandrogenism and affects approximately

5–10% of women of reproductive age (Knochenhauer et al.,

1998; Diamanti-Kandarakis et al., 1999). PCOS is probably the

most prevalent endocrinopathy in women and by far the most

common cause of anovulatory infertility (Homburg, 1996).

The primary aetiology of PCOS is unknown (Balen, 2004).

However, insulin resistance with compensatory hyperinsuli-

naemia is a prominent feature of the syndrome and appears tohave a pathophysiologic role in the hyperandrogenism of the

disorder for both lean and obese women with PCOS (Dunaif

et al., 1989). Hyperinsulinaemia results in increased ovarian

androgen biosynthesis in vivo and in vitro (Adashi et al., 1985;

Barbieri et al., 1986) and decreased sex hormone-binding glob-

ulin (SHBG) synthesis from the liver (Nestler et al., 1991), lead-

ing to increased bioavailability of free androgens.

Women with PCOS who are anovulatory and wishing to

become pregnant have traditionally been treated with the

anti-estrogen clomiphene citrate as first-line medical therapy

induce ovulation. Ovulation induction (OI) with gonadotroph

therapy usually follows for those women who fail to eith

ovulate or conceive with a course of clomiphene citrate trea

ment (Balen, 1998). IVF is an effective therapy for PCO

patients who are refractory to OI or who have co-existing infe

tility factors (Buyalos and Lee, 1996).

The association of insulin resistance contributing to anov

lation has led to the novel and promising therapy of administe

ing insulin-sensitizing drugs to women with PCOS to restoovulation and enhance pregnancy. Of the insulin-sensitizin

drugs, metformin has been the one studied most widely and h

the most reassuring safety profile (Nestler et al., 2002). Me

formin enhances insulin sensitivity in both the liver, where

inhibits hepatic glucose production, and the peripheral tissu

where it increases glucose uptake and utilization into musc

tissue. By increasing insulin sensitivity, metformin reduc

insulin resistance, insulin secretion and hyperinsulinaem

(Dunn and Peters, 1995).

mailto:[email protected]






M.F.Costello, M.Chapman and U.Conway

1388

Patients with PCOS undergoing gonadotrophin OI or IVF

usually show an increased response to gonadotrophins and

consequently produce large numbers of follicles and oocytes

with high serum estradiol (E2) levels, resulting in an increased

risk of ovarian hyperstimulation syndrome (OHSS) (Aboulghar

and Mansour, 2003; Anonymous, 2003; Yarali and Zeyneloglu,

2004). PCOS women with insulin resistance undergoing gona-

dotrophin OI have a longer duration of treatment, use a higher

total FSH dose, have an elevated cancellation rate and a lower

conception rate (Dale et al., 1998). On the other hand, insulin

resistance in PCOS women undergoing IVF has been not been

shown to be independently related to IVF outcome (Fedorcsak

et al., 2001).

There is physiologic rationale for believing that suppression

of insulin levels, with insulin-sensitizing agents such as met-

formin, in women with PCOS undergoing gonadotrophin OI or

IVF might ameliorate the adverse effects on ovarian stimula-

tion and consequently improve treatment outcomes such as

ovulation and pregnancy rates (Dunaif et al., 1989). In addi-

tion, metformin may also act directly on ovarian thecal cells to

decrease androgen production (Attia et al., 2001).

Three recent systematic reviews have demonstrated that met-formin co-administration with clomiphene citrate OI improves

both ovulation and pregnancy rates in both unselected and clo-

miphene citrate-resistant PCOS women (Costello and Eden,

2003; Lord et al., 2003, 2004; Kashyap et al., 2004). Our pre-

vious systematic review analysed both observational studies

and randomized controlled trials (RCTs) and located only two

published studies (both RCTs) on the use of metformin in FSH

OI, and no meta-analysis was performed. In addition, only a

single observational study was identified assessing the effect of

co-administering metformin during IVF treatment of PCOS

patients (Costello and Eden, 2003). The systematic reviews

and meta-analyses of RCTs by Lord et al. (2003, 2004) and

Kashyap et al. (2004) did not evaluate the effect of metforminco-administration with either FSH OI or IVF.

The objective of this review was to investigate the effective-

ness of metformin used in combination with gonadotrophins

for OI or IVF in restoring ovulation and achieving pregnancy

and live birth in women with PCOS.

Materials and methods

The electronic search strategy involved conducting a literature search

for all pertinent published RCTs on the use of metformin in combina-

tion with gonadotrophins for OI with timed intercourse or IVF for

restoring ovulation and achieving pregnancy and live birth in women

with PCOS using the Cochrane central register of controlled trials(Cochrane Library, 3rd Quarter, 2005) and the bibliographic databases

MEDLINE (January 1966 to August 2005) and EMBASE (January

1980 to August 2005). References of selected articles identified were

hand-searched for additional relevant citations. Experts and specialists

in the field were also contacted for additional relevant studies. There

was no restriction on the language of publication (Moher et al., 2000;

Juni et al., 2002). Only RCTs were considered as these studies gener-

ally provide the least biased estimates of treatment effect and are

therefore more suitable for meta-analysis (Hughes, 1996).

Search terms included those for patient selection (‘polycystic ovary

syndrome’, ‘ovary polycystic disease’), exposure (‘metformin’, ‘insulin

sensitising drugs’, ‘insulin sensitising agents’, ‘gonadotrophins’, ‘fol-

licle stimulating hormone’, ‘follitropin’, ‘ovulation induction’ and

‘fertilization in vitro’), primary outcomes of interest (‘ovulation’,

‘pregnancy’, ‘pregnancy outcome’ and ‘live birth’) and study type

(‘randomized controlled trial’, ‘randomized’ and ‘randomised’).

Inclusion criteria for selecting RCTs were based on patient popula-

tion (PCOS), treatment intervention versus comparison (metformin

and gonadotrophin OI with timed intercourse versus gonadotrophin

OI with timed intercourse and no treatment or placebo, metformin and

IVF versus IVF and no treatment or placebo) and primary (restorationof ovulation, pregnancy and live birth) and secondary (ovarian

response and OHSS) outcomes of interest. There were no restrictions

on the language of publication. Only parallel-group designed trials or

pre-crossover data from crossover trials were included because cross-

over trials may exaggerate estimates of effectiveness when compared

to parallel-group designed trials (Khan et al., 1996).

The methodological standards of all studies were assessed accord-

ing to standards most likely to provide valid results (Guyatt et al.,

1993). Down-weighting of studies of doubtful quality for the purpose

of meta-analysis was not performed because of such weighting being

too arbitrary, even though quality criteria have been published

(Chalmers et al., 1981; Thompson and Pocock, 1991). In addition, the

incorporation of such weighting by quality scores lacks statistical or

empirical justification (Detsky et al., 1992; Juni et al., 2001). The ana-lysis of individual components of trial quality overcomes many of the

shortcomings of composite scores (Juni et al., 2001). Institutional

review board approval was not sought for this systematic review

because only previously published data were used.

Statistical analysis

Statistical analysis was performed using RevMan 4.2 software pro-

vided by the Cochrane Collaboration. For dichotomous data, results

for each study were expressed as odds ratios (OR) with 95% confid-

ence intervals (95% CIs) and combined for meta-analysis to calculate

a pooled estimate of treatment effect for each outcome across studies.

For continuous data, the mean post-treatment/intervention values and

standard deviation for each group were measured, and weighted mean

differences (WMD) with 95% CI were calculated.The combined results of each study for meta-analysis and statistical

homogeneity between trials were assessed using the fixed-effects

model where Cochran Q-test (chi-squared test) P ≥ 0.05 represents

statistical homogeneity. The random-effects model of meta-analysis

was used in the presence of unexplained statistical heterogeneity.

P values <0.05 or 95% CI not containing 1.00 (OR) or 0 (WMD) were

considered statistically significant.

Results

Trial flow

Eight RCTs that met the inclusion criteria were identified and

thus consulted in the review (De Leo et al., 1999; Yarali et al.,2002; Fedorcsak et al., 2003; Visnova et al., 2003; Kjøtrød

et al., 2004; Tasdemir et al., 2004; Onalan et al., 2005; Tang

et al., 2005) (Figure 1). All the eight RCTs were included in

the review.

Study characteristics and validity assessment

Tables I and II summarize the characteristics of the included

RCTs evaluating the effectiveness of metformin co-administration

with gonadotrophin OI and IVF, respectively, in women with

PCOS. Only one of the eight RCTs completely fulfilled the



Metformin and gonadotrophins in PCO

13

validity criteria in terms of their design and execution, but this

is seldom achieved in any RCT (double blind, placebo con-

trolled, large sample size, long duration, few dropouts and

intention-to-treat analysis) (Tang et al., 2005). Our approach

was to exclude only trials with gross deficiencies in design as

excluding trials that fail to meet some arbitrary standard of

quality could exclude studies that might contribute valid

information (Juni et al., 2001). All of the identified studieswere considered to be of a reasonable methodological standard

to be included in the review (Guyatt et al., 1993).

One of the eight trials was published in abstract format only

(Tang et al., 2005). This trial was included, despite limited

methodological information being provided in the abstract for-

mat, to have a total perspective on this review. Additional rel-

evant information not published in the abstract was sought and

received by personal communication. One of the included stud-

ies was printed in Czech and required translation into English

for this review (Visnova et al., 2003).

Meta-analysis results Metformin and gonadotrophin OI versus gonadotrophin OI

and placebo or no treatment

Primary outcomes

Ovulation. A single RCT was identified assessing the effect

on ovulation of metformin co-administration with gonado-

trophin OI in PCOS (Table I). This trial by Yarali et al. (2002)

demonstrated that metformin did not improve the ovulation

rate in clomiphene citrate-resistant PCOS women undergoing

gonadotrophin OI (Figure 2).

Pregnancy. Figure 3 demonstrates the forest (or CI) plot

RCTs reporting on pregnancy rate as an outcome measure. N

improvement in pregnancy rates was seen in any of the indivi

ual trials or meta-analysis. However, the meta-analysis demo

strated a trend towards an improvement in pregnancy rat

with metformin co-administration (28 versus 10%; OR = 3.4

P = 0.05, 95% CI = 0.98–12.2).

Live birth. There were no trials identified comparing me

formin combined with gonadotrophin OI versus gonadotrophOI and placebo or no treatment reporting live birth as an ou

come measure (Table I).

Secondary outcomes

Length of ovarian stimulation. Figure 4 demonstrates the fo

est plot of RCTs reporting on the length of ovarian stimulatio

as an outcome measure. The meta-analysis of the two RCT

shows that metformin significantly reduces the length

ovarian stimulation at gonadotrophin OI (WMD = –4.14 day

P = 0.0002, 95% CI = –6.36 to –1.93).

Total FSH dose. The co-administration of metformin wi

gonadotrophin OI significantly decreases the total do

of FSH used during the treatment cycle (WMD = –425.

IU, P < 0.00001, 95% CI –507.08 to –343.03) (Figure 5).

Serum E 2 level on the day of HCG trigger. Metformin sign

icantly reduces the serum E2 level on the day of HCG trigg

(WMD = –1.24 nmol/l, P < 0.00001, 95% CI = –1.5 to –0.9

(Figure 6).

OHSS. There were no RCTs identified comparing metform

combined with gonadotrophin OI versus gonadotrophin OI and p

cebo or no treatment reporting on the outcome measure of OHSS.

Metformin and IVF versus IVF and placebo or no treatmen

Primary outcomesPregnancy. The meta-analysis of the combined data of the fi

trials evaluating the effectiveness of metformin combined with IV

treatment on pregnancy rates in PCOS patients demonstrates th

metformin co-administration did not improve pregnancy rates

IVF (34 versus 29%; OR = 1.29, P = 0.25, 95% CI = 0.84–1.9

(Figure 7). Only one of the five individual trials showed

improvement in pregnancy rates with metformin (Tang et al., 200

Live birth. Two RCTs were identified assessing the effect

metformin co-administration during ovarian stimulation

IVF on live birth rates in PCOS women (Table II) (Kjøtrø

et al., 2004; Tang et al., 2005). One of the individual RCT

showed a significant improvement in live birth rates with meformin (Tang et al., 2005). The meta-analysis of the combine

data of the two trials showed a non-significant improveme

in live birth rates with metformin (36 versus 22%; OR = 2.0

P = 0.06, 95% CI = 0.98–4.14) (Figure 8).

Secondary outcomes

Length of ovarian stimulation. The addition of metformin

IVF treatment has no effect on the length of ovarian stimul

tion (WMD = –0.09 days, P = 0.66, 95% CI = –0.49 to 0.3

(Figure 9).

Figure 1. Flow diagram showing the number of publications identi-fied in the literature search. RCTs, randomized controlled trials.

10

11

12

Potentially relevant RCTs identified and

screened for retrieval (n = 567)

Potentially relevant non duplicated

RCTs screened for retrieval (n = 309)

Remove and check for duplication (n = 258)

Relevance filter according to inclusion criteria listed

in materials and methods section applied to titles and

abstracts. RCTs excluded with reasons (n = 301):• Review articles (n = 58)

• Not relevant to patient population, treatment

intervention versus comparison, or outcome

of interest (n = 242)

• Cohorts not randomized (n = 1)

RCTs retrieved for more detailed

evaluation (n = 8)

Potentially appropriate RCTs to be

included in the meta-analysis (n =8)

RCTs included in the meta-analysis

(n = 8)

RCTs excluded (n = 0)

RCTs excluded from meta-analysis (n = 0)






139

T a b l e I I .

C h a r a c t e r i s t i c s o f p u b l i s h e d R C

T s o n t h e e f f e c t o f m e t f o r m i n c o m b i n e d w i t h I V

F v e r s u s I V F a l o n e i n w o m e n w i t h P C O S

R e f e r e n c e a n d

l o c a t i o n

R C T t y p e

M e t h o d o f

r a n d o m i z a t i o n

( a l l o c a t i o n

c o n c e a l m e n

t )

O u t c o m e s

r e p o r t e d

O t h e r i n f e r t i l i t y

f a c t o r p r e s e n t

P r i o r I V F o r

I C S

I t r e a t m e n t

n

A g e

( y e a r s )

D u r a t i o n o f

i n f e r t i l i t y

( y e a r s )

B M

I

F I

I n t e r v e n t i o n

N o t e s

M +

I V F g r o u p

F e d o r c s a k e t a l .

( 2 0 0 3 ) , N o r w a y

R C T

R a n d o m n u

m b e r s

t a b l e ( y e s )

P r e g n a n c y

O v a r i a n

r e s p o n s e O H S S

N A

N A

1 7 a

3 1 . 0

N A

3 1 . 5

2 6 . 5

M 5

0 0 m g t h r e e t i m e s

d a i l y s t a r t i n g 3 w e e k s

b e f o r e d o w n - r e g u l a t i o n

w i t h G n R H a b e g a n a n d

c o n t i n u e d t h r o u g h o u t

o v a r i a n s t i m u l a t i o n f o r

o n e c y c l e o f I V F o r I C S I .

M s t o

p p e d o n d a y o f

H C G t r i g g e r

D o w n - r e g

u l a t i o n

w i t h G n R

H a

f o l l o w e d

b y o v a r i a n

s t i m u l a t i o

n

f o r o n e c y

c l e o f

I V F o r I C

S I

A l l w o m e n w e r e

i n s u l i n r e s i s t a n t .

T h i s t r i a l w a s a

c r o s s o v e r t r i a l a n d

o n l y p r e - c r o s s o v e r

d a t a w e r e

c o n s i d e r e d .

G n = r F S H .

E m b r y o t r a n s f e r

d a y 3 .

I n t e n t i o n - t o -

t r e a t a n a l y s i s

V i s n o v a e t a l .

( 2 0 0 3 ) , C z e c h

R e p u b l i c

R C T

N A

P r e g n a n c y

O v a r i a n


N A

Y e s ( n

= N A )

1 3 7

2 8 . 7

N A

2 3 . 5

N A

M 5

0 0 m g t w i c e d a i l y

s t a r t i n g w i t h F S H

i n j e c t i o n s d u r i n g l o n g

d o w n - r e g u l a t i o n p r o t o c o l

w i t h G n R H a a n d



o n e c y c l e o f I V F .

M

s t o p p e d a t d i a g n o s i s o f

p r e g n a n c y

D o w n - r e g

u l a t i o n

w i t h G n R

H a

f o l l o w e d b y o v a r i a n

s t i m u l a t i o

n f o r o n e

c y c l e o f I V F

G n = r F S H o r

h i g h l y p u r i f i e d

u F S H .

I V F o r I C S I

N A .

D a y o f

e m b r y o t r a n s f e r

N A .

P e r p r o t o c o l

a n a l y s i s

K j o t r o d e t a l .

( 2 0 0 4 ) , N o r w a y

D B P C

N A

P r e g n a n c y L i v e

b i r t h O v a r i a n


T u b a l ( n

= 1 2 ) ,

e n d o m e t r i o s i s

( n =

3 ) , m a l e

f a c t o r ( n

= 2 2 )

N A

7 3 a

2 9 . 6

4 . 1

2 9 . 2

N A

M 1

g t w i c e d a i l y f o r 1 6

w e e k s b e f o r e d o w n -

r e g u l a t i o n w i t h G n R H a

a n d c o n t i n u e d t h r o u g h o u t



M s t o


H C G t r i g g e r

P t w i c e d a i l y f o r 1 6


r e g u l a t i o n

w i t h

G n R H a a n d

c o n t i n u e d

t h r o u g h o u t

o v a r i a n s t

i m u l a t i o n

f o r o n e c y

c l e o f I V F

o r I C S I . P

s t o p p e d o n

d a y o f H C

G t r i g g e r

G n = r F S H .


d a y 3 .


a n a l y s i s




1392

B M I , m e a n b o d y m a s s i n d e x ( k g / m 2 ) ; D B

, d o u b l e b l i n d ; F I , m e a n f a s t i n g i n s u l i n ( m U / l ) ;

G n , g o n a d o t r o p h i n ; M , m e t f o r m i n ; N A , n o t a v a i l a b l e ; O H S S , o v a r i a n h y p e r s t i m u l a t i o n s y n d r o m

e ; P , p l a c e b o ; P C , p l a c e b o

c o n t r o l l e d ; P C O S , p o l y c y s t i c o v a r y s y n d r o m e ; R C T , r a n d o m i z e d c o n t r o l l e d t r i a l ; r F S H , r e c o m b i n a n t F S H ; u F S H , u r i n a r y F S H .

a D i a g n o s i s o f P C O S m e e t s T h e R o t t e r d a m

C o n s e n s u s G r o u p 2 0 0 3 c r i t e r i a ( F a u s e r e t a l . , 2 0 0 4 ) .

T a b l e I I .

C o n t i n u e d

R e f e r e n c e a n d

l o c a t i o n

R C T t y p e

M e t h o d o f

r a n d o m i z a t i o n

( a l l o c a t i o n

c o n c e a l m e n

t )

O u t c o m e s

r e p o r t e d

O t h e r i n f e r t i l i t y

f a c t o r p r e s e n t

P r i o r I V F o r

I C S

I t r e a t m e n t

n

A g e

( y e a r s )

D u r a t i o n o f

i n f e r t i l i t y

( y e a r s )

B M

I

F I

I n t e r v e n t i o n

N o t e s

M +

I V F g r o u p

T a n g e t a l .

( 2 0 0 5 ) , U n i t e d

K i n g d o m

D B P C

R a n d o m n u

m b e r s

t a b l e ( y e s )

P r e g n a n c y L i v e

b i r t h O v a r i a n


T u b a l ( n

= 1 1 ) ,

e n d o m e t r i o s i s

( n =

5 ) , m a l e

( n =

5 0 )

Y e s ( n

= 5 8 )

1 0 1 a

3 1 . 2

4 . 2

5

2 7 . 3

7 . 1

M 8


s t a r t i n g a t d o w n -

r e g u l a t i o n w i t h G n R H a

a n d c o n t i n u e d t h r o u g h o u t



M s t o

p p e d o n d a y o f e g g

c o l l e c t i o n

P t w i c e d a i l y s t a r t i n g

a t d o w n - r e g u l a t i o n

w i t h G n R

H a a n d

c o n t i n u e d

t h r o u g h o u t

o v a r i a n s t

i m u l a t i o n

f o r o n e c y

c l e o f I V F

o r I C S I . P

s t o p p e d o n

d a y o f e g g c o l l e c t i o n

P u b l i s h e d i n

a b s t r a c t f o r m a t

o n l y .

G n = r F S H .


d a y 2 .

I n t e n t i o n - t o -

t r e a t a n a l y s i s .

A d d i t i o n a l

i n f o r m a t i o n t o t h a t

c o n t a i n e d i n t h e

a b s t r a c t a c q u i r e d

t h r o u g h p e r s o n a l

c o m m u n i c a t i o n

O n a l a n e t a l .

( 2 0 0 5 ) , T u r k e y

D B P C

C o m p u t e r

g e n e r a t e d ( y e s )

P r e g n a n c y

O v a r i a n


N o

N o

1 1 0 a

2 9 . 5

( a l l < 4 0 )

7 . 9

2 4 . 8

1 5 . 2

M 8


( B M I < 2 8 k g / m 2 ) o r

t h r e e t i m e s d a i l y ( B M I ≥

2 8 k g / m 2 ) f o r 8 w e e k s

b e f o r e d o w n - r e g u l a t i o n

w i t h G n R H a a n d



o n e c y c l e o f I C S I . M

s t o p p e d o n d a y o f

p r e g n a n c y t e s t

P t w i c e d a i l y ( B M I

< 2 8 k g / m

2 ) o r t h r e e

t i m e s d a i l

y ( B M I ≥

2 8 k g / m 2 ) f o r 8


r e g u l a t i o n

w i t h

G n R H a a n d

c o n t i n u e d

t h r o u g h o u t

o v a r i a n s t

i m u l a t i o n

f o r o n e c y

c l e o f I C S I .

M s t o


p r e g n a n c y t e s t

G n = r F S H .

I C S I

o n l y .

A s s i s t e d

h a t c h i n g

p e r f o r m e d o n

s e l e c t e d p a t i e n t s .


d a y 3 .


a n a l y s i s .

A d d i t i o n a l

i n f o r m a t i o n t o t h a t

c o n t a i n e d i n t h e

p u b l i c a t i o n

a c q u i r e d t h r o u g h

p e r s o n a l

c o m m u n i c a t i o n




139

Total FSH dose. The co-administration of metformin with

gonadotrophin OI significantly decreases the total dose

of FSH used during the treatment cycle (WMD = –290.42

IU, P = 0.0004, 95% CI = –450.34 to –130.51) (Figure 10).

Serum E 2 level on the day of HCG trigger. Metformin did not

reduce the serum E2 level on the day of HCG trigger (WMD

= –3.53 nmol/l, P = 0.23, 95% CI = –9.24 to 2.18) (Figure 11).

Statistical heterogeneity (variability in the treatment effec

being evaluated in the different trials) was present (P < 0.0000

c 2 = 29.12, I 2 = 93.1%).

Number of oocytes collected at IVF. Metformin has no effe

on the number of oocytes collected at IVF (WMD = 0.4

P = 0.54, 95% CI = –0.98 to 1.86) (Figure 12).

Figure 2. Comparison of metformin versus placebo or no treatment in gonadotrophin ovulation induction (OI) with outcome of ovulation rate.

Review: Metformin and Gonadotropins in PCOS

Comparison: 01 Metformin versus placebo or no treatment (Gonadotropin OI)

Outcome: 01 Ovulation rate

Study Metformin Control OR (fixed) Weight OR (fixed)

or sub-category n/N n/N 95% CI % 95% CI

Yarali 2002 9/10 11/15 100.00 3.27 [0.31, 34.72]

Total (95% CI) 10 15 100.00 3.27 [0.31, 34.72]

Total events: 9 (Metformin), 11 (Control)

Test for heterogeneity: not applicable

Test for overall effect: Z = 0.98 (P = 0.33)

0.01 0.1 1 10 100

Favours control Favoursmetformin

Figure 3. Comparison of metformin versus placebo or no treatment in gonadotrophin ovulation induction (OI) with outcome of pregnancy rate


Comparison: 01 Metformin versus placebo or no treatment (Gonadotropin OI)

Outcome: 02 Pregnancy rate



De Leo 1999 3/10 1/10 25.36 3.86 [0.33, 45.57]

Yarali 2002 3/10 1/15 20.29 6.00 [0.52, 68.72]

Tasdemir 2004 4/16 2/16 54.35 2.33 [0.36, 15.05]

Total (95% CI) 36 41 100.00 3.46 [0.98, 12.20]


Test for heterogeneity: Chi² = 0.37, df = 2 (P = 0.83), I² = 0%


0.1 0.2 0.5 1 2 5 10


Figure 4. Comparison of metformin versus placebo or no treatment in gonadotrophin ovulation induction (OI) with outcome of length of ovariastimulation.

Review: Metformin and Gonadotropinsin PCOS

Comparison: 01 Metformin versusplacebo or no treatment (Gonadotropin OI)

Outcome: 10 Length of ovarian stimulation (days)

Study Metformin Control WMD (fixed) Weight WMD (fixed)

or sub-category N Mean (SD) N Mean (SD) 95% CI % 95%CI

Yarali 2002 10 17.10(7.80) 15 19.70(10.80) 9.19 -2.60 [-9.90, 4.70]

Tasdemir 2004 16 11.20(3.30) 16 15.50(3.40) 90.81 -4.30 [-6.62, -1.98]

Total (95% CI) 26 31 100.00 -4.14 [-6.36, -1.93]



-10 -5 0 5 10

Favoursmetformin Favourscontrol

Figure 5. Comparison of metformin versus placebo or no treatment in gonadotrophin ovulation induction (OI) with outcome of total FSH dose used.



Outcome: 11 Total FSH dose (IU)



Yarali 2002 10 1607.80(887.10) 15 2062.50(1397.10) 0.84 -454.70 [-1350.34, 440.94]

Tasdemir 2004 16 1050.30(75.10) 16 1475.10(150.40) 99.16 -424.80 [-507.17, -342.43]

Total (95% CI) 26 31 100.00 -425.05 [-507.08, -343.03]


Test for overall ef fect: Z = 10.16 (P < 0.00001)

-1000 -500 0 500 1000





1394

OHSS. Figure 13 demonstrates the forest plot of RCTs

reporting the development of OHSS as an outcome measure.

The meta-analysis of the four RCTs shows that metformin sig-

nificantly reduces the risk of OHSS at IVF (5.6 versus 21.0%;

OR = 0.21, P < 0.00001, 95% CI = 0.11–0.41).

Discussion

This is the first systematic review and meta-analysis of the use of

metformin co-administration in PCOS patients undergoing gona-

dotrophin OI or IVF. This systematic review demonstrates

that the co-administration of metformin to gonadotrophin OI

Figure 6. Comparison of metformin versus placebo or no treatment in gonadotrophin ovulation induction (OI) with outcome of maximum serumestradiol level during stimulation.



Outcome: 16 Serum estradiol level (nmol/L)



Yarali 2002 10 1.87(0.90) 15 1.99(2.28) 4.07 -0.12 [-1.40, 1.16]

Tasdemir 2004 16 1.47(0.28) 16 2.76(0.46) 95.93 -1.29 [-1.55, -1.03]

Total (95% CI) 26 31 100.00 -1.24 [-1.50, -0.98]

Test for heterogeneity: Chi² = 3.07, df = 1 (P = 0.08), I² = 67.4%

Test for overall ef fect: Z = 9.42 (P < 0.00001)

-4 -2 0 2 4


Figure 7. Comparison of metformin versus placebo or no treatment in IVF with outcome of pregnancy rate.


Comparison: 02 Metformin versus placebo or no treatment (IVF)

Outcome: 02 Pregnancy rate



Fedorcsak 2003 2/9 2/8 4.50 0.86 [0.09, 8.07]

Visnova 2003 17/62 10/51 21.74 1.55 [0.64, 3.77]

Kjotrod 2004 15/29 14/28 18.77 1.07 [0.38, 3.03]

Onalan 2005 16/53 22/55 41.15 0.65 [0.29, 1.44]

Tang 2005 20/52 8/49 13.84 3.20 [1.25, 8.21]

otal (95% CI) 205 191 100.00 1.29 [0.84, 1.98]

otal events: 70 (Metformin), 56 (Control)

est for heterogeneity: Chi² = 6.86, df = 4 (P = 0.14), I² = 41.7%

est for overall effect: Z = 1.15 (P = 0.25)

0.2 0.5 1 2 5


Figure 8. Comparison of metformin versus placebo or no treatment in IVF with outcome of live birth rate.



Outcome: 03 Live birth rate



Kjotrod 2004 12/29 11/28 61.21 1.09 [0.38, 3.15]

Tang 2005 17/52 6/49 38.79 3.48 [1.24, 9.77]

otal (95% CI) 81 77 100.00 2.02 [0.98, 4.14]

otal events: 29 (Metformin), 17 (Control)est for heterogeneity: Chi² = 2.37, df = 1 (P = 0.12), I² = 57.8%

est for overall effect: Z = 1.91 (P = 0.06)

0.2 0.5 1 2 5


Figure 9. Comparison of metformin versus placebo or no treatment in IVF with outcome of length of ovarian stimulation.


Comparison: 02 Metformin versusplacebo or no treatment (IVF)

Outcome: 04 Length of ovarian stimulation (days)



Fedorcsak2003 9 18.40(7.40) 8 13.70(4.10) 0.50 4.70 [-0.91, 10.31]

Visnova 2003 71 10.85(2.32) 66 11.35(1.96) 30.84 -0.50 [-1.22, 0.22]

Kjotrod 2004 30 14.40(3.48) 32 14.20(4.30) 4.21 0.20 [-1.74, 2.14]

Onalan 2005 53 9.40(1.45) 55 9.35(1.16) 64.44 0.05 [-0.45, 0.55]

Total (95% CI) 163 161 100.00 -0.09 [-0.49, 0.31]Test for heterogeneity: Chi² = 4.45, df = 3 (P = 0.22), I² = 32.6%


-1 -0.5 0 0.5 1





139

and IVF does not improve ovulation, pregnancy or live birth

rates in PCOS women. However, the number of subjects in both

the individual trials and combined meta-analysis was small, as

evidenced by the wide confidence limits, thus limiting the

power of the meta-analysis to exclude a treatment benefit.

Metformin has a consistent effect on ovarian response du

ing gonadotrophin OI, with the length of ovarian stimulatio

total dose of FSH used and serum E2 level on the day of HC

trigger all reduced with the use of metformin. Metformin h

a variable effect on ovarian response during IVF treatmen

Figure 10. Comparison of metformin versus placebo or no treatment in IVF with outcome of total FSH dose used.



Outcome: 05 Total FSH dose (IU)



Fedorcsak2003 9 3641.70(1964.40) 8 2542.20(1299.10) 1.04 1099.50 [-468.13, 2667.13]

isnova 2003 71 1850.00(692.10) 66 2195.00(405.90) 72.02 -345.00 [-533.43, -156.57]

Kjotrod 2004 30 1883.00(732.00) 32 2039.00(1414.00) 8.29 -156.00 [-711.54, 399.54]

Onalan 2005 53 2067.15(1014.79) 55 2284.09(945.75) 18.65 -216.94 [-587.23, 153.35]

Total (95% CI) 163 161 100.00 -290.42 [-450.34, -130.51]

Test for heterogeneity: Chi² = 3.72, df = 3 (P = 0.29), I² = 19.3%


-1000 -500 0 500 1000


Figure 11. Comparison of metformin versus placebo or no treatment in IVF with outcome of maximum serum estradiol level during stimulatio



Outcome: 12 Serum estradiol level (nmol/L)

Study Metformin Control WMD (random) Weight WMD (random)


isnova 2003 71 12.56(6.21) 66 22.69(11.59) 32.36 -10.13 [-13.28, -6.98]

Kjotrod 2004 31 6.80(3.96) 30 7.60(5.36) 33.79 -0.80 [-3.17, 1.57]

Onalan 2005 53 14.56(7.04) 55 14.51(5.19) 33.85 0.05 [-2.29, 2.39]

Total (95% CI) 155 151 100.00 -3.53 [-9.24, 2.18]

Test for heterogeneity: Chi² = 29.12, df = 2 (P < 0.00001), I² = 93.1%


-10 -5 0 5 10


Figure 12. Comparison of metformin versus placebo or no treatment in IVF with outcome of number of oocytes retrieved.



Outcome: 08 Number of oocytescollected



Fedorcsak2003 9 7.90(6.00) 8 7.40(5.90) 6.31 0.50 [-5.16, 6.16]

isnova 2003 71 14.79(8.01) 66 16.22(8.59) 26.08 -1.43 [-4.22, 1.36]

Kjotrod 2004 30 13.90(7.50) 31 13.10(6.50) 16.28 0.80 [-2.73, 4.33]

Onalan 2005 53 19.51(9.03) 55 18.07(5.33) 25.65 1.44 [-1.37, 4.25]

Tang 2005 52 17.30(7.40) 49 16.20(7.00) 25.68 1.10 [-1.71, 3.91]

Total (95% CI) 215 209 100.00 0.44 [-0.98, 1.86]



-4 -2 0 2 4


Figure 13. Comparison of metformin versus placebo or no treatment in IVF with outcome of ovarian hyperstimulation syndrome.



Outcome: 01 Ovarian Hyperstimulation Syndrome (OHSS) rate



Fedorcsak 2003 0/9 0/8 Not estimable

Visnova 2003 6/71 26/66 58.62 0.14 [0.05, 0.38]

Kjotrod 2004 1/31 4/32 9.05 0.23 [0.02, 2.22]

Onalan 2005 3/53 4/55 8.80 0.77 [0.16, 3.59]

Tang 2005 2/52 10/49 23.53 0.16 [0.03, 0.75]

Total (95% CI) 216 210 100.00 0.21 [0.11, 0.41]


Test for heterogeneity: Chi² = 3.45, df = 3 (P = 0.33), I² = 13.1%Test for overall effect: Z = 4.55 (P < 0.00001)

0.01 0.1 1 10 100

Favours metformin Favourscontrol




1396

Metformin reduced the total FSH dose used but had no effect

on the length of ovarian stimulation, serum E2 level on the

day of HCG trigger or number of oocytes collected. The risk

of OHSS in PCOS women undergoing IVF was reduced with

metformin.

A recent international consensus workshop group recom-

mended that all clinicians and investigators now use the inter-

nationally agreed definition of PCOS to ensure uniformity in

routine clinical management and research studies (Fauser

et al., 2004). We therefore believed it important to note

whether the definition of PCOS in each of the individual RCTs

of this systematic review met the new Rotterdam Consensus

Group 2003 criteria (Tables I and II). Subjects in only one of

the eight RCTs in this review did not meet the new consensus

criteria (Visnova et al., 2003).

Metformin co-administration with gonadotrophin OI

The primary outcome measure in all the three studies assessing

the effect of metformin in gonadotrophin OI was ovarian

response rather than ovulation or pregnancy rate (Table I; Fig-

ures 2–6). Each of the trials was small (n < 50) and thus under-

powered to assess pregnancy rates. All the trials includedPCOS patients who met the Rotterdam consensus criteria and

who were resistant to clomiphene citrate. The patients as a

group tended to be young and overweight in the individual

studies. Two of the three RCTs excluded both tubal uterine and

male factor causes of infertility (Yarali et al., 2002; Tasdemir

et al., 2004).

The trials used different gonadotrophin OI protocols, and

no mention was made of coital timing in any of the three trials.

The length of pretreatment with metformin ranged between

1 and 2 months, and metformin was terminated either on the

day of HCG trigger or on the day of pregnancy test. The study

by Yarali et al. (2002) measured biochemical pregnancies only

(positive serum pregnancy test 13–15 days after HCG triggerinjection), whilst the other two papers (De Leo et al., 1999;

Tasdemir et al., 2004) did not define the diagnosis of preg-

nancy. No patients in any of the three RCTs had to discontinue

their metformin as a result of side effects (De Leo et al., 1999;

Yarali et al., 2002; Tasdemir et al., 2004).

Only one of the three trials assessed endocrine and/or meta-

bolic parameters and found that there was no change in insulin

sensitivity indices following a 6-week pretreatment period with

metformin or placebo (Yarali et al., 2002). The only endocrine

change noted during this same period was a decline in serum-

free testosterone concentration with metformin.

Metformin appears to have a favourable effect on ovarian

response parameters at gonadotrophin OI in terms of reducing

the length of ovarian stimulation by approximately 4 days,

total dose of FSH used by around 425 IU and serum E2 on the

day of HCG trigger by about 1.24 nmol/l (1240 pmol/l or 338

pg/ml) (Figures 4–6). These findings translate into a saving of

both time and pharmacological expense for the patient being

treated with metformin during gonadotrophin OI, even for the

patient who undergoes up to 8 weeks of pretreatment with

metformin (MIMS Australia Online Prescribing Section,

2005).

The meta-analysis demonstrated no improvement in preg-

nancy rates with metformin use in gonadotrophin OI (Figure 3).

However, all the three small (n < 50) individual trials and the

meta-analysis showed a trend towards an improvement in cycle

pregnancy rates with metformin co-administration. A very

likely reason for this lack of statistically significant benefit

with metformin use is a lack of study power (the probability of

finding an effect when it does exist), resulting in type 2 or beta

error, commonly seen in small trials, where no significant dif-

ference in treatment effect can be detected, while in fact a real

difference exists. The combined data, comparing only approxi-

mately 40 cycles in each group, demonstrated a non-significant

3.46 OR improvement in pregnancy with metformin co-admin-

istration. The power of this meta-analysis (Figure 3) to detect a

significant difference (P value <0.05) between the two treat-

ments, given the observed differences, is 73%. In other words,

the meta-analysis has a 73% chance of detecting (or 27% like-

lihood of missing) a true difference between the intervention

and control groups; compared to an acceptable level of 80%

(Kirby et al., 2002).

Further evidence of the lack of study power in the meta-

analysis is seen in the very wide CIs in the individual trials andthe meta-analysis, although the latter’s confidence limits are

considerably narrower as a result of the increased patient num-

bers. The upper limit of the combined data’s 95% CI is 12.20.

If the OR at this upper boundary were true, then this would be

clinically important (i.e. metformin may increase the odds of

pregnancy in gonadotrophin OI by 12-fold), and thus the meta-

analysis has failed to exclude an important treatment effect of

metformin co-administration.

It was not possible to perform a meta-analysis of the rate of

OHSS development as it was not reported in two of the trials

(Yarali et al., 2002; Tasdemir et al., 2004). The third (crosso-

ver) trial by De Leo et al. (1999) did not report OHSS events

before treatment crossover but did report no difference in thenumber of cycles with ‘hyperstimulation after HCG’ with (1/8

= 12.5%) or without (5/19 = 26.3%) metformin co-treatment in

the women who completed both pre- and post-crossover treat-

ment arms.

Metformin co-administration with IVF

None of the five trials assessing the effect of metformin in IVF

treatment assessed pregnancy rate as the main outcome meas-

ure (Fedorcsak et al., 2003; Visnova et al., 2003; Kjøtrød et al.,

2004; Onalan et al., 2005; Tang et al., 2005). The primary end-

points were either not reported (Visnova et al., 2003; Onalan

et al., 2005) or based on ovarian response parameters (Fedorcsak

et al., 2003; Kjøtrød et al., 2004; Tang et al., 2005). Only two

of the trials performed a priori sample size calculations to

assess their primary outcome measures (Kjøtrød et al., 2004;

Tang et al., 2005).

The patients in one of the trials did not meet the Rotterdam

consensus criteria (Fauser et al., 2004) as other causes for

hyperandrogenism which mimic PCOS (such as congenital

adrenal hyperplasia, Cushing’s syndrome or androgen-secreting

tumours) were not excluded (Visnova et al., 2003). The

patients in the five trials were young and were overweight or




139

obese in three of the five trials (Table II). Previous exposure to

clomiphene citrate including whether the patients were clomi-

phene citrate resistant was not stated in any of the five RCTs.

All five trials used the long down-regulation protocol using a

GnRH agonist (Fedorcsak et al., 2003; Visnova et al., 2003;

Kjøtrød et al., 2004; Onalan et al., 2005; Tang et al., 2005).

Four of the trials assessed clinical pregnancies (intrauterine

pregnancy on ultrasound) (Fedorcsak et al., 2003; Kjøtrød et al.,

2004; Onalan et al., 2005; Tang et al., 2005), whilst pregnancy

was not defined in the other trial (Visnova et al., 2003). No

women had to stop their metformin treatment as a result of side

effects in the two trials which reported on such adverse effects

(Fedorcsak et al., 2003; Kjøtrød et al., 2004). Only one of the

trials (Tang et al., 2005) compared the changes in endocrine or

metabolic variables over the course of the study between the

treatment and control groups (personal communication).

Metformin reduced the total FSH dose used by about 290 IU

but had no effect on the length of ovarian stimulation, serum E2

level on the day of HCG trigger or number of oocytes collected

(Figures 9–12). Again, this finding translates into a saving of

pharmacological expense for the patient being treated with

metformin during IVF, even for the patient who undergoes upto 16 weeks of pretreatment with metformin before down-

regulation (MIMS Australia Online Prescribing Section, 2005).

Caution is needed in interpreting the meta-analysis result on

the serum E2 level on the day of HCG trigger as statistical het-

erogeneity was present (P < 0.00001, c 2 = 29.12, I 2 = 93.1%)

(Figure 11). The results of Visnova et al. (2003) were consider-

ably different to the other two trials (Kjøtrød et al., 2004;

Onalan et al., 2005). Statistically significant heterogeneity implies

a low probability that the observed differences in results from

study to study are due to random error (chance) alone, indicat-

ing that differences in clinical parameters (patients, exposures

or outcomes) or study methodology (design or conduct) are

responsible for the varying treatment effect rather than theinterventions being compared (Oxman et al., 1994).

Statistical heterogeneity could possibly be explained by dif-

ferences in the study design/quality, patient population or

interventions between the three trials (Visnova et al., 2003;

Kjøtrød et al., 2004; Onalan et al., 2005) (Table II). However,

these possible causes of the heterogeneity cannot be explored

using subgroup sensitivity analyses because of too few studies

to perform this adequately. Therefore, as the statistical hetero-

geneity is unexplained, the random-effects model of meta-

analysis was used.

The co-administration of metformin does not improve the

pregnancy or live birth rate at IVF (Figures 7 and 8). However,

the meta-analysis showed a trend towards an improvement inpregnancy and live birth rates with metformin co-administration.

Only one of five RCTs and one of two RCTs demonstrated an

improvement in pregnancy rate and live birth rate, respec-

tively, with metformin (Tang et al., 2005). This trial by Tang

et al. did not seem to differ from the other four trials in terms

of patient selection or intervention (Table II).

The most likely reason for the meta-analysis showing no sig-

nificant benefit for metformin co-administration in terms of

pregnancy or live birth rates at IVF is type 2 or beta error as the

power of the meta-analysis in Figures 7 and 8 is only 20 and

56%, respectively, given the sample sizes of each study grou

observed differences seen between the groups and a = 0.0

(5% significance level). Therefore, the meta-analyses in Fi

ures 7 and 8 are inconclusive, and this is further evidenced b

the upper limit of the CIs which is 1.98 for pregnancy rate an

4.14 for live birth rates. If the OR at this upper boundary we

true (i.e. metformin co-treatment in IVF approximately do

bling the odds of pregnancy and quadrupling the odds of li

birth), then the meta-analyses have failed to exclude an impo

ant clinical treatment effect.

Metformin appears to reduce the risk of developing OHSS

IVF (Figure 13). This finding is interesting, particularly wh

the meta-analysis also shows that metformin does not affect t

serum E2 levels on the day of HCG or the number of oocyt

collected (Figures 11 and 12). All five RCTs evaluated the ra

of OHSS but did not define the condition or its severity, wi

the exception of Tang et al. (2005) who defined OHSS

severe requiring hospitalization (personal communication). N

cases of OHSS were seen in the trial by Fedorcsak et a

(2003). None of the authors offered an explanation as to wh

metformin may reduce the risk of OHSS.

The underlying reason for a reduction in the incidence OHSS with metformin is not known definitively. Improveme

in insulin sensitivity with consequent lowering of insulin leve

may be one possible explanation. Hyperinsulinaemia is a ri

factor for OHSS as women with PCOS who are hyperinsu

naemic have a higher level of E2 and incidence of ovari

hyperstimulation in response to ovarian stimulation with FS

compared to those with normoinsulinaemia (Fulghesu et a

1997). Insulin increases vascular endothelial growth fact

(VEGF) expression in vascular smooth muscle cells (Doron

et al., 2004), and VEGF has emerged as one of the factors mo

likely involved in the pathophysiology of OHSS (Anonymou

2003). Therefore, it may be postulated that metformin reduc

the risk of OHSS by decreasing serum insulin levels, leading a consequent reduction in the production of VEGF.

Limitations

There are a number of limitations to this review. The value

any meta-analysis is totally dependent on the quality and la

of bias in its component primary studies. A significant conce

with any meta-analysis is that meta-analytic methods may b

used inappropriately to combine biased and disparate studie

leading to misleading systematic reviews. The caref

assessment of study validity and heterogeneity is essential

minimizing this risk (Hughes, 1996). Therefore, it is impor

ant to restrict inclusion to RCTs, ideally with adequate randomzation, objective or blinded outcome assessment, comple

follow-up information and intention-to-treat analysis (Guya

et al., 1993). Unfortunately, very few of the RCTs included

this review satisfied all these criteria (Tables I and II).

Five of the eight RCTs clearly described the method used f

random allocation (De Leo et al., 1999; Yarali et al., 200

Fedorcsak et al., 2003; Onalan et al., 2005; Tang et al., 2005

and only three of the trials reported on allocation concealme

(Fedorcsak et al., 2003; Onalan et al., 2005; Tang et al., 200

(Tables I and II). Trials with insecure treatment allocatio




1398

concealment report effect sizes that are 30–40% larger than

studies with secure randomization (Schulz et al., 1995; Moher

et al., 1998). Allocation concealment aims to avoid selection

bias by ensuring that both study groups are equal at the start of

the trial (Juni et al., 2001).

Only four of the eight RCTs in this meta-analysis were

blinded to ensure that both the study groups are treated and

assessed equally (Yarali et al., 2002; Kjøtrød et al., 2004;

Onalan et al., 2005; Tang et al., 2005). Blinding of patients and

care providers prevents performance bias (i.e. where additional

treatment interventions are provided preferentially to one

group), whilst the blinding of those assessing outcomes (i.e.

patients, care providers, others etc.) prevents detection bias

(Juni et al., 2001). Lack of double blinding has been shown to

exaggerate treatment effects by 17% (Schulz et al., 1995).

Analysis should be ideally performed according to the inten-

tion-to-treat principle (i.e. ensuring that the patient groups are

equivalent at the end of the trial as they were at the beginning) to

avoid attrition bias by excluding patients who either deviated from

the protocol or were lost to follow-up; as such patients excluded

after treatment allocation or randomization are unlikely to be rep-

resentative of patients remaining in the study (Juni et al., 2001).However, studies addressing attrition bias have demonstrated that

trials with inadequate reporting on patient exclusions did not

affect treatment estimates (Schulz et al., 1995; Juni et al., 2001).

Four of the eight trials in this review specifically stated

whether the participating couples had undergone prior infertil-

ity treatment cycles beforehand (Yarali et al., 2002; Visnova

et al., 2003; Onalan et al., 2005; Tang et al., 2005). A subfer-

tile population that has already undergone infertility treatment

is likely to have lower fecundity as couples with the highest

fecundity might have conceived already. If differences

between treatments are more likely to show in a more fertile

population, this may lead to underestimation of treatment dif-

ferences in a trial (Cohlen et al., 2003).Despite the absence of statistically significant heterogeneity

between the RCTs combined for the meta-analyses of preg-

nancy and ovarian response parameters (apart from Figure 11

discussed above), clinical heterogeneity or diversity existed

because of inherent variability in patient selection (i.e. differ-

ent definitions of clomiphene citrate resistance in the gonado-

trophin OI trials, other causes of infertility in the IVF trials,

prior treatment cycles in gonadotrophin OI or IVF, age, dura-

tion of infertility, BMI and insulin resistance), interventions

including co-interventions (i.e. different gonadotrophin OI pro-

tocols, different FSH injection types and doses for OI or IVF

ovarian stimulation, different metformin dosages, different

pretreatment periods with metformin, different times for stop-ping the metformin, different cycle monitoring and HCG trig-

gering criteria, probable different coital timing in

gonadotrophin OI trials, different luteal phase support for the

IVF studies) and outcomes (i.e. methods of measuring serum

E2, definition of OHSS and pregnancy) between the individual

trials. Such differences observed between the trials are to be

expected because of the variability in the nature of clinical

practice where such studies should be conducted.

Although the inclusion of unpublished studies is controver-

sial (Cook et al., 1993), reliance upon published studies alone

may distort the results of a meta-analysis because positive

studies (statistically significant) are more likely to be published

than negative ones (non-significant), with the attendant risk for

the review to overestimate treatment efficacy (Dickersin et al.,

1987). However, only one of the individual trials in this meta-

analysis was a positive study for the primary outcome mea-

sures of ovulation, pregnancy or live birth (Tang et al., 2005),

thus limiting any potential for publication bias.

A treatment can be evaluated according to several criteria.

Efficacy is usually the most important outcome parameter,

but the number and severity of side effects, the cost associ-

ated with treatment and the treatment effects on quality of

life are also important to assess. This review focused on effi-

cacy, particularly in terms of pregnancy rate. No assessment

of treatment cost or its effect on quality of life was per-

formed in this review as none of the trials assessed these out-

come measures.

Conclusions

This systematic review and meta-analysis, based on limited

available level 1 evidence, demonstrates that metformin co-

treatment does not significantly improve ovulation, pregnancy

or live birth rates in women with PCOS undergoing gonado-

trophin OI or IVF. However, the review is inconclusive in

terms of not being able to exclude an important clinical treat-

ment effect because of the small number of trials and small

sample sizes of the individual trials limiting the power of the

meta-analysis. It is too soon to draw conclusions from the liter-

ature. Further large, well-designed and executed RCTs are

necessary to definitively answer the important clinical question

of whether metformin use in PCOS women improves preg-

nancy and live birth rates in gonadotrophin OI or IVF.

Metformin has a consistent effect on ovarian response dur-

ing gonadotrophin OI, with the length of ovarian stimulation,total dose of FSH used and maximal E2 level all reduced with

the use of metformin. Metformin reduces the total FSH dose

used in IVF but has no effect on the length of ovarian stimula-

tion, serum E2 level on the day of HCG trigger or number of

oocytes collected. The risk of OHSS in PCOS women under-

going IVF was reduced with metformin.

Acknowledgement

The authors thank the Cochrane Collaboration for providing the RevMan4.2 software used for statistical analysis and construction of forest (orCI) plots. A full review on metformin and Assisted Reproductive

Technology is currently in progress by the Cochrane Menstrual Disor-ders and Subfertility Group and will be available on the CochraneLibrary upon completion.

The authors also wish to thank Dr T.Tang and Dr G.Onalan for sup-plying additional information on their trials and Dr Eva Durna at theRoyal Hospital for Women in Sydney, Australia for translating theCzech publication (Visnova et al., 2003) into English.

References

Aboulghar MA and Mansour RT (2003) Ovarian hyperstimulation syndrome:classifications and critical analysis of preventive measures. Hum ReprodUpdate 9,275–289.




139

Adashi EY, Resnick CE, D’Ercole AJ, Svoboda ME and Van Wyk JJ (1985)Insulin-like growth factors as intraovarian regulators of granulosa cellgrowth and function. Endocr Rev 6,400–420.

Anonymous (2003) The Practice Committee of the American Society forReproductive Medicine ovarian hyperstimulation syndrome. Fertil Steril80,1309–1314.

Attia GR, Rainey WE and Carr BR (2001) Metformin directly inhibits andro-gen production in human thecal cells. Fertil Steril 76,517–524.

Balen A (1998) Endocrine methods of ovulation induction. Baillieres ClinObstet Gynaecol 12,521–539.

Balen A (2004) The pathophysiology of polycystic ovary syndrome: trying to

understand PCOS and its endocrinology. Best Pract Res Clin Obstet Gynae-col 18,685–706.

Barbieri RL, Makris A, Randall RW, Daniels G, Kistner RW and Ryan KJ(1986) Insulin stimulates androgen accumulation in incubations of ovarianstroma obtained from women with hyperandrogenism. J Clin EndocrinolMetab 62,904–910.

Buyalos RP and Lee CT (1996) Polycystic ovary syndrome: pathophysiologyand outcome with in vitro fertilization. Fertil Steril 65,1–10.

Chalmers TC, Smith H Jr, Blackburn B, Silverman B, Schroeder B, Reitman Dand Ambroz A (1981) A method for assessing the quality of a randomizedcontrol trial. Control Clin Trials 2,31–49.

Cohlen BJ, Vandekerckhove P, te Velde ER and Habbema JDF (2003) Timedintercourse versus intra-uterine insemination with or without ovarian hyper-stimulation for subfertility in men (Cochrane Review). In The CochraneLibrary, Issue 1. Update Software, Oxford, pp. 00–00.

Cook DJ, Guyatt GH, Ryan G, Clifton J, Buckingham L, Willan A, McIlroy Wand Oxman AD (1993) Should unpublished data be included in meta-analyses?Current convictions and controversies. JAMA 269,2749–2753.

Costello MF and Eden JA (2003) A systematic review of the reproductive systemeffects of metformin in patients with polycystic ovary syndrome. Fertil Steril79,1–13.

Dale PO, Tanbo T, Haug E and Abyholm T (1998) The impact of insulin resistanceon the outcome of ovulation induction with low-dose follicle stimulating hor-mone in women with polycystic ovary syndrome. Hum Reprod 13,567–570.

De Leo V, la Marca A, Ditto A, Morgante G and Cianci A (1999) Effects of metformin on gonadotropin-induced ovulation in women with polycysticovary syndrome. Fertil Steril 72,282–285.

Detsky AS, Naylor CD, O’Rourke K, McGeer AJ and L’Abbe KA (1992)Incorporating variations in the quality of individual randomized trials intometa-analysis. J Clin Epidemiol 45,255–265.

Diamanti-Kandarakis E, Kouli CR, Bergiele AT, Filandra FA, Tsianateli TC,Spina GG, Zapanti ED and Bartzis MI (1999) A survey of the polycysticovary syndrome in the Greek island of Lesbos: hormonal and metabolic pro-

file. J Clin Endocrinol Metab 84,4006–4011.Dickersin K, Chan S, Chalmers TC, Sacks HS and Smith H Jr (1987) Publica-

tion bias and clinical trials. Control Clin Trials 8,343–353.

Doronzo G, Russo I, Mattiello L, Anfossi G, Bosia A and Trovati M (2004) Insulinactivates vascular endothelial growth factor in vascular smooth muscle cells:influence of nitric oxide and of insulin resistance. Eur J Clin Invest 34,664–673.

Dunaif A, Segal KR, Futterweit W and Dobrjansky A (1989) Profound periph-eral insulin resistance, independent of obesity, in polycystic ovary syn-drome. Diabetes 38,1165–1174.

Dunn CJ and Peters DH (1995) Metformin: a review of its pharmacological proper-ties and therapeutic uses in non-insulin dependent diabetes. Drugs 49,721–749.

Fauser B, Tarlatzis B, Chang J, Azziz R, Legro R, Dewailly D, Franks S,Balen AH, Bouchard P, Dahlgren E et al. (2004) The Rotterdam ESHRE/ ASRM-sponsored PCOS consensus workshop group. Revised 2003 consen-sus on diagnostic criteria and long-term health risks related to polycysticovary syndrome (PCOS). Hum Reprod 19,41–47.

Fedorcsak P, Dale PO, Storeng R, Tanbo T and Abyholm T (2001) The impactof obesity and insulin resistance on the outcome of IVF or ICSI in womenwith polycystic ovarian syndrome. Hum Reprod 16,1086–1091.

Fedorcsak P, Dale PO, Storeng R, Abyholm T and Tanbo T (2003) The effectof metformin on ovarian stimulation and in vitro fertilization in insulin-resistant women with polycystic ovary syndrome: an open-label randomizedcross-over trial. Gynecol Endocrinol 17,207–214.

Fulghesu AM, Villa P, Pavone V, Guido M, Apa R, Caruso A, Lanzone A,Rossodivita A and Mancuso S (1997) The impact of insulin secretion on theovarian response to exogenous gonadotropins in polycystic ovary syndrome.J Clin Endocrinol Metab 82,644–648.

Guyatt GH, Sackett DL and Cook DJ (1993) Users’ guides to the medical literature.II. How to use an article about therapy or prevention. A. Are the results of thestudy valid? Evidence-Based Medicine Working Group. JAMA 270,2598–2601.

Homburg R (1996) Polycystic ovary syndrome – from gynaecological curioity to multisystem endocrinopathy. Hum Reprod 11,29–39.

Hughes EG (1996) Systematic literature review and meta-analysis. SemReprod Endocrinol 14,161–169.

Juni P, Altman DG and Egger M (2001) Assessing the quality of controllclinical trials. BMJ 323,42–46.

Juni P, Holenstein F, Sterne J, Bartlett C and Egger M (2002) Direction aimpact of language bias in meta-analyses of controlled trials: empiristudy [comment]. Int J Epidemiol 31,115–123.

Kashyap S, Wells GA and Rosenwaks Z (2004) Insulin-sensitizing agents primary therapy for patients with polycystic ovarian syndrome. Hum Repr

19,2474–2483.Khan KS, Daya S, Collins JA and Walter SD (1996) Empirical evidence

bias in infertility research: overestimation of treatment effect in crossovtrials using pregnancy as the outcome measure. Fertil Steril 65,939–945.

Kirby A, Gebski V and Keech AC (2002) Determining the sample size inclinical trial. Med J Aust 177,256–257.

Kjøtrød SB, von Düring V and Carlsen SM (2004) Metformin treatment befoIVF/ICSI in women with polycystic ovary syndrome; a prospective, ranomized, double blind study. Hum Reprod 19,1315–1322.

Knochenhauer ES, Key TJ, Kahsar-Miller M, Waggoner W, Boots LR aAzziz R (1998) Prevalence of the polycystic ovary syndrome in unselectblack and white women of the southeastern United States: a prospectistudy. J Clin Endocrinol Metab 83,3078–3082.

Lord JM, Flight IH and Norman RJ (2003) Metformin in polycystic ovary sydrome: systematic review and meta-analysis. BMJ 327,951–953.

Lord JM, Flight IHK and Norman RJ (2004) Insulin-sensitising drugs (m

formin, troglitazone, rosiglitazone, pioglitazone,D

-chiro-inositol) for pocystic ovary syndrome (Cochrane Review). In The Cochrane Library, Iss2. John Wiley & Sons, Chichester, UK.

MIMS Australia Online Prescribing Section (2005) Subsection 6 (c) PituitaHormones and subsection 6 (e) Hypoglycaemic Agents (httpmims.hcn.net.au/ifmx-nsapi/mims-data/?MIval=2MIMS_ssearch).

Moher D, Pham B, Jones A, Cook DJ, Jadad AR, Moher M, Tugwell P aKlassen TP (1998) Does quality of reports of randomised trials affect estimaof intervention efficacy reported in meta-analyses? Lancet 352,609–613.

Moher D, Pham B, Klassen TP, Schulz KF, Berlin JA, Jadad AR and Liberati(2000) What contributions do languages other than English make on tresults of meta-analyses?. J Clin Epidemiol 53,964–972.

Nestler JE, Powers LP, Matt DW, Steingold KA, Plymate SR, Rittmaster RClore JN and Blackard WG (1991) A direct effect of hyperinsulinemia serum sex hormone-binding globulin levels in obese women with the pocystic ovary syndrome. J Clin Endocrinol Metab 72,83–89.

Nestler JE, Stovall D, Akhter N, Iuorno MJ and Jakubowicz MJ (2002) Stra

gies for the use of insulin-sensitizing drugs to treat infertility in women wpolycystic ovary syndrome. Fertil Steril 77,209–215.

Onalan G, Pabuccu R, Goktolga U, Ceyhan T, Bagis T and Cincik M (200Metformin treatment in patients with polycystic ovary syndrome undergoiin vitro fertilization: a prospective randomized trial. Fertil Steril 84,798–80

Oxman AD, Cook DJ and Guyatt GH (1994) Users’ guide to the medical litature VI. How to use an overview. JAMA 272,1367–1371.

Schulz KF, Chalmers I, Hayes RJ and Altman DG (1995) Empirical evidenof bias. Dimensions of methodological quality associated with estimatestreatment effects in controlled trials. JAMA 273,408–412.

Tang T, Barth JH and Balen AH (2005) The use of metformin for women wPCOS undergoing IVF treatment. Hum Reprod 20,17–18.

Tasdemir S, Ficicioglu C, Yalti S, Gurbuz B, Basaran T and Yildirim G (200The effect of metformin treatment to ovarian response in cases with PCOArch Gynecol Obstet 2692,121–124.

Thompson SG and Pocock SJ (1991) Can meta-analyses be trusted? Lan

338,1127–1130.Visnova H, Ventruba P, Crha I and Zakova J (2003) Importance of sensitiz

tion of insulin receptors in the prevention of ovarian hyperstimulation sydrome. Ceska Gynekol 683,155–162.

Yarali H and Zeyneloglu HB (2004) Gonadotrophin treatment in patients wpolycystic ovary syndrome. Reprod Biomed Online 8,528–537.

Yarali H, Yildiz BO, Demirol A, Zeyneloglu HB, Yigit N, Bukulmez O aKoray Z (2002) Co-administration of metformin during rFSH treatmentpatients with clomiphene citrate-resistant polycystic ovarian syndromeprospective randomized trial. Hum Reprod 17,289–294.

Submitted on November 9, 2005; resubmitted on December 20, 2005; accepton December 22, 2005

http://mims.hcn.net.au/ifmx-nsapi/mims-data/?MIval=2MIMS_ssearch




Date post:	06-Apr-2018
Category:	Documents
Upload:	lilian-kusuma
View:	219 times
Download:	0 times

Hum. Reprod. 2006 Costello 1387 99

Documents