Systematic review

Systematic review and meta-analysis of the use of fibrin sealant

to prevent seroma formation after breast cancer surgery
P. A. Carless and D. A. Henry
Discipline of Clinical Pharmacology, School of Medicine and Public Health, Faculty of Health, University of Newcastle, Newcastle, Australia
Correspondence to: Mr P. A. Carless, Discipline of Clinical Pharmacology, Level 5 Clinical Sciences Building, NM2, Mater Misericordiae Hospital, Edith
Street, Waratah, 2298, New South Wales, Australia (e-mail: Paul.Carless@newcastle.edu.au)

Background: The use of fibrin sealant has been proposed as a means of preventing seroma formation
following breast cancer surgery. Conflicting trial results require the efficacy of fibrin sealant to be
reviewed critically.
Methods: A systematic review of randomized controlled trials was conducted to examine the efficacy
of fibrin sealants in reducing postoperative drainage and seroma formation after breast cancer surgery.
Studies were identified by computer searches of Medline, Embase, the Cochrane Central Register of
Controlled Trials and manufacturer websites (to June 2005), and bibliographic searches of published
articles. Trials were eligible for inclusion if they reported data on postoperative drainage and the number
of patients who developed a seroma.
Results: Eleven trials met the criteria for inclusion. Generally, the trials were small and of poor
methodological quality. Fibrin sealant did not reduce the rate of postoperative seroma (relative risk 1·14,
95 per cent confidence interval (c.i.) 0·88 to 1·46), the volume of drainage (weighted mean difference
− 117·7, 95 per cent c.i. − 259·2 to 23·8 ml), or the length of hospital stay (weighted mean difference
− 0·38, 95 per cent c.i. − 1·58 to 0·83 days).
Conclusions: The current evidence does not support the use of fibrin sealant in breast cancer surgery
to reduce postoperative drainage or seroma formation.

Paper accepted 8 March 2006

Published online in Wiley InterScience (www.bjs.co.uk). DOI: 10.1002/bjs.5432

Introduction external compression dressings10 , using ultrasound cutting

Breast cancer is the second leading cause of death
among women in the USA, with approximately 210 000
new cases of invasive breast cancer and over 40 000
deaths occurring annually1 . Although there are a number
of treatment modalities for breast cancer, surgical
management comprises resection of the primary tumour
and axillary lymph node surgery2 . Seroma formation is
a frequent complication of breast cancer surgery, with
rates ranging from less than 3 per cent to more than
50 per cent3 . Although seromas may be considered more
of a nuisance than a serious complication, they are
associated with wound infection, wound dehiscence and
skin-flap necrosis, which delay recovery and can extend a
patient’s stay in hospital4 – 7 . Various approaches have been
used to prevent seroma formation, including closing the
dead space under the mastectomy skin flap by suturing8 ,
delaying shoulder exercises in the recovery phase9 , using
devices11 , using suction drainage systems12 , applying
bovine thrombin during surgery7 and using fibrin sealant6 .
Fibrin sealants have been used in surgery for over
30 years. Their potential as haemostatic and sealing agents
has expanded across a wide range of surgical settings,
including cardiac, vascular, liver, prostatic, thoracic, cos-
metic, orthopaedic and gynaecological surgery13 . A num-
ber of studies suggesting fibrin sealant might help prevent
seroma formation in animal mastectomy models14 – 16 led
to the use of fibrin sealant in human mastectomy. By
improving haemostasis and tissue adherence, fibrin sealant
could theoretically prevent seroma formation by reducing
postoperative serosanguinous fluid accumulation, the con-
sequence of transecting multiple small blood vessels and
lymphatics while removing the breast and axillary lymph
nodes6 . However, the use of fibrin sealant in breast cancer
surgery remains controversial owing to conflicting trial

Copyright  2006 British Journal of Surgery Society Ltd British Journal of Surgery 2006; 93: 810–819
Published by John Wiley & Sons Ltd
Fibrin sealants and seroma formation
results. The purpose of the present work was to review sys-
tematically the evidence from randomized controlled trials
studying the impact of fibrin sealant on serosanguinous
drainage and seroma formation following breast cancer
surgery.
Methods
The primary objective of this study was to determine
whether fibrin sealant decreased the rate of seroma
formation and the volume of drainage output following
breast cancer surgery. Secondary analyses included the
number of drainage days, the frequency of wound infection
and length of stay in hospital.
Literature search and study selection
This review used the standard methodology for systematic
reviews established by the Cochrane Collaboration17 .
Studies were identified by computer searches of Medline
(1966–June 2005), Embase (1980–June 2005) and the
Cochrane Central Register of Controlled Trials (Cochrane
Library, Issue 3, 2005). These databases were initially
searched with an unrestricted search strategy, using
exploded MeSH (medical subject heading) terms (‘fibrin
tissue adhesive’ and ‘fibrin glue’) and specific text-
word terms for fibrin sealant (‘fibrin glue$’, ‘fibrin
sealant$’, ‘fibrin seal$’, ‘biological glue$ or biological
seal$’, ‘beriplast’, ‘bolheal’, ‘tissucol’, ‘tisseel’, ‘quixil’,
‘biocol’ and ‘crosseal’). The dollar sign was used to
retrieve all possible suffix variations of the root word or
phrase.
To improve the specificity of these searches, the above
terms were combined with breast-cancer-specific MeSH
and text-word terms and then restricted using a search
filter to identify randomized controlled trials18 . In addition
to these database searches, manufacturer websites were
searched and experts in the field were contacted to
identify relevant reports or projects. The reference lists
of eligible trials, review articles and reports were searched
for potentially relevant studies.
To be included, studies had to be randomized controlled
trials, to report the number of patients developing seroma
and the volume of drainage output following breast cancer
surgery, to be published in the English language and to
investigate the use of fibrin sealant delivered directly to
the wound surface in either liquid or aerosol form. Those
assessing bandages or pads impregnated with lyophilized
fibrin sealant components were excluded, as were any
duplicate publications.
Copyright  2006 British Journal of Surgery Society Ltd
Published by John Wiley & Sons Ltd
811
Data collection
Data were extracted from the studies using a data extraction
form and entered into Review Manager (RevMan 4·2·7,
Cochrane Collaboration, Oxford, UK). The number of
patients developing postoperative seroma and the volume
of drainage output after breast cancer surgery were
recorded. Other outcome measures were the number of
days drains remained in situ, frequency of wound infection
and length of hospital stay.
Studies that met the inclusion criteria were assessed for
methodological quality by two independent raters using
criteria proposed by Schulz et al.19 that specified four items
of assessment: double-blinding, allocation concealment,
participant withdrawal and methods used to achieve
randomization (method of generating random sequence).
Disagreements were resolved by consensus.
Data analysis
Dichotomous data (such as number of patients with
seroma) and continuous data (such as volume of drainage
and days with drains in situ) were analysed using RevMan
Analyses (in Review Manager 4·2·7). Data for continuous
outcomes were not included in the meta-analysis if standard
deviations or standard errors of mean were not reported
or could not be calculated. Outcomes were expressed
as pooled relative risks (RRs), risk differences (RDs)
or, for continuous variables, weighted mean differences
(WMDs) using a random effects model20 . The Q statistic
was used to assess heterogeneity of treatment effects. A
P value of 0·1 or less was used to define statistically
significant heterogeneity. Predefined subgroup analyses
were performed to determine whether effect sizes varied
according to factors such as the type of surgery and the
type and volume of fibrin sealant used.
Results
The literature search identified 11 trials that met the
inclusion criteria21 – 31 . These trials, spanning an interval
of 12 years, formed the basis of this systematic review.
Characteristics of studies
The 11 trials recruited a total of 632 patients, of whom 328
were randomized to fibrin sealant treatment (Table 1). The
trials were generally small, with a low number of patients
in each trial arm (range ten to 58 patients). Modified
radical mastectomy was the most frequently performed
type of surgery (ten of 11 trials). Four trials studied
fibrin sealant exclusively in patients having modified radical
mastectomy22,24,25,29 . The remaining trials also included
www.bjs.co.uk British Journal of Surgery 2006; 93: 810–819
 812
Table 1 Characteristics of included randomized controlled trials of fibrin sealant

No. of participants
Type of Age of participants
Reference Year Country surgery (years) FS Control Summary of results (FS versus control)

Uden et al.24 1993 Sweden MRM FS: median (range) 36 32 Seroma: 63·9 versus 53·1%
73 (42–89) Median (range) drainage days: 3 (1–5) versus 3 (1–5) days (P > 0·050)
Control: median (range) Median (range) anterior drainage: 50 (0–250) versus 60 (0–280) ml
70 (40–84) (P > 0·05) Median (range) posterior drainage: 520 (67–3040) versus 545 (140–2615) ml

Published by John Wiley & Sons Ltd

Median length of hospital stay: 5·7 versus 4·3 days (P > 0·050)
Vaxman et al.23 1995 France MRM FS: 55·6(12) 20 20 Seroma: 20 versus 5%
Lumpectomy Control: 56·2(10) (P = 0·900) Axilla drainage days: 5·85(2·7) versus 4·89(1·9) days (P = 0·200)
Mastectomy drainage days: 2·3(1·0) versus 2·95(1·2) days (P = 0·080)
Total fluid drainage: 479·8(190·3) versus 425·9(212·6) ml (P = 0·400)
Length of hospital stay: 10·8(3·5) versus 9·4(3·5) days (P = 0·230)

Copyright  2006 British Journal of Surgery Society Ltd

Moore et al.22 1997 USA MRM FS: 56·5(13·503) 11 10 Drainage days: 3·9(1·7) versus 6·9(1·19) days (P = 0·000 091)
Control: 62·9(14·9) (P = 0·310) Day 9 cumulative drainage: 346·1(245·3) versus 745·7(267·3) ml (P = 0·001)
Gilly et al.21 1998 France MRM FS: 60·6(10·8) 50 58 Seroma: 2 versus 1·7%
Segmental Control: 62·5(11·5) (P = 0·456) Day 6 cumulative drainage: 214·4(105·9) versus 407·8(240·1) ml (P = 0·001)
mastectomy Length of hospital stay: 8·0(1·6) versus 10·1(2·1) days (P = 0·006)
Dinsmore et al.25 2000 USA MRM FS: 62·3(7·86) 14 13 Seroma: 43 versus 23%
Control: 64·5(10·1) (P = 0·54) Total cumulative drainage: 1308 versus 754 ml (P = 0·001)
Drainage days: 12·3 versus 16·5 days (P = 0·250)
Infected seroma: 21·4 versus 0%
Moore et al.26 2001 USA MRM FS 4 ml: 58(16) 19 Seroma: 15·8 (4 ml), 21·1 (8 ml), 28·6 (16 ml) versus 28·6%
Lumpectomy FS 8 ml 51(9) 19 Wound infection: 5·3 (4 ml), 10·5 (8 ml), 10·0 (16 ml) versus 0%
FS 16 ml 59(16) 20 Haematoma: 5·3 (4 ml), 0 (8 ml), 14·3 (16 ml) versus 0%
Control: 56(14) 21 Skin necrosis: 0 (4 ml), 0 (8 ml), 9·5 (16 ml) versus 9·5%
Langer et al.30 2003 USA MRM FS: mean (range) 60·8 (34–88) Seroma: 3·8 versus 3·4%

www.bjs.co.uk
TM Mean (range) total drainage: 411 (30–1350) versus 504 (72–910) ml (P = 0·150)
ALND Control: mean (range) 56·3 (37–82) Mean (range) drainage days: 7·0 (1–14) versus 8·3 (4–14) (P = 0·050)
Wound infection: 0 versus 3·4%
Ulusoy et al.29 2003 Turkey MRM FS: 51·4 (12·2) 27 27 Seroma: 18·5 versus 11·1%
Control: 50·9 (10·9) Drainage days: 9·88(3·07) versus 9·66(3·01) (P > 0·050)
Total cumulative drainage: 738·5(471) versus 886·4(476·7) ml (P > 0·050)
Jain et al.31 2004 UK Mastectomy FS + control: 62·3(12·3) 29 29 Seroma: 34·5 versus 41·4%
Lumpectomy Median (IQR) frequency of aspiration: 1 (1–1) versus 2 (1–3)
Mustonen et al.28 2004 Finland MRM + axillary FS: 67·5(13·6) 19 21 Seroma: 36·8 versus 47·6%
evaluation Control: 66·1(12·0) (P > 0·050) Total cumulative drainage: 181·3(141) versus 168·9(140·9) ml
Length of hospital stay: 3·9(1·0) versus 3·7(1·0) days (P > 0·050)
Johnson et al.27 2005 USA MRM FS: 58·6(11·3) 38 44 Seroma: 36·8 versus 22·7%
Lumpectomy Control: 59·5(12·8) (P = 0·753) Wound infection: 7·9 versus 6·8%
TM Flap necrosis: 2·6 versus 2·3%
ALND

Values are mean(s.d.) unless otherwise stated. FS, fibrin sealant; MRM, modified radical mastectomy; TM, total mastectomy; ALND, axillary lymph node dissection; IQR, interquartile range.

British Journal of Surgery 2006; 93: 810–819

P. A. Carless and D. A. Henry
Fibrin sealants and seroma formation 813

Table 2 Details of fibrin sealants used in randomized trials

Type of fibrin Quantity Method of

Reference sealant Fibrin sealant components* applied application Site of application

24 
Uden et al. Tisseel † Fibrinogen: 75–115 mg/ml 2 ml Aerosol spray Cut surfaces
Fibronectin: 2–9 mg/ml
Factor XIII: 10–50 units/ml
Plasminogen: 40–120 ug/ml
Aprotinin: 3000 KIU/ml
Thrombin: 500 IU/ml
Ca Cl: 36–44 umol/ml
Vaxman et al.23 Tisseel ‡ As above 5 ml Spray Both the inner and outer walls of the
armpit. Not applied to mastectomy
or lumpectomy sites
Langer et al.30 Tisseel § As above 2 ml for TM Dual syringe Chest wall and skin flaps
or ALND, (Duploject )
5 ml for using gas
MRM pressured spray
device
(Tissomat )
Ulusoy et al.29 Tisseel ‡ As above 4 ml Spray Mastectomy and axillary areas
Jain et al.31 Tisseel ** As above 2 ml Dual syringe All dissected areas of chest wall, skin
(Duploject ) flaps and axilla
using gas
pressured spray
device
(Tissomat )
Mustonen et al.25 Tisseel †† As above 2 ml Spray Mastectomy and axillary sites
Gilly et al.21 Tissucol # As above 2 ml Syringe Dissection sites
Johnson et al.27 Hemaseel APR‡‡ As above 2–4 ml Aerosol spray Sprayed directly into wound cavity or
(HemaMyst ) site of axillary dissection
Moore et al.22 Autologous Fibrinogen: 40 ± 5 mg/ml 20 ml Spray Dried axillary tissue surfaces, between
fibrinogen + Fibrinogen vol.: 19 ± 2·6 ml loose suture loops
bovine thrombin§§ Thrombin: 1000 units/ml
Thrombin vol: 20 ml
Dinsmore et al.25 Autologous Fibrinogen: 25·47 mg/ml 15 ml Spray Chest wall and axilla: flaps then
fibrinogen + bovine Thrombin: 1000 units/ml lowered into place
thrombin Ca Cl: 10 ml
(Thrombinar )##
Moore et al.26 VI Guard*** Fibrinogen: 75 mg/ml 4–24 ml Spray Axillary dissection site + extra fibrin
Thrombin: 200 units/ml sealant (8 ml) applied to skin flap
site of MRM patients

*Sourced from either trial reports or manufacturer internet websites; †Immuno, Sweden AB Ltd; ‡Manufacturer details not reported; §Baxter Healthcare,
Glendale, California, USA; #Immuno AG, Vienna, Austria; **Baxter Healthcare, Newbury, UK; ††Baxter, no further details reported; ‡‡Haemacure,
Sarasota, Florida, USA; §§University of Virginia Blood Bank, USA (fibrinogen) and Johnson & Johnson, New Brunswick, New Jersey, USA (thrombin);
##Jones Medical, St Louis, Missouri, USA; ***VI Technologies/Vitex, Watertown, Massachusetts, USA. Ca Cl, calcium chloride; TM, total mastectomy;
ALND, axillary lymph node dissection; MRM, modified radical mastectomy; Vol., volume.

patients undergoing various other procedures, such as
lumpectomy, axillary lymph node dissection, segmental
mastectomy and total mastectomy. The mean or median
age of the participants ranged from 50·9 to 73·0 years.
Eight of the 11 trials (73 per cent) studied the fibrin
sealant Tisseel (Baxter) or its market variants Tissucol
(Immuno AG) and Hemaseel APR (Haemacure)(Table 2).
The volume of fibrin sealant applied to the surgical site
varied considerably from 2 to 24 ml. In ten trials the fibrin
sealant was applied using a spray device. The number and
type of drain, the use of compression dressings, and the
extent of postoperative shoulder mobilization also varied
between trials (Table 3).
Meta-analyses
Seroma formation
Ten trials reported data for seroma formation (Fig. 1).
These trials evaluated a total of 612 patients, of whom 318
were randomized to fibrin sealant. In patients who received
fibrin sealant, the pooled RR of developing a seroma
following breast cancer surgery compared with control

Copyright  2006 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of Surgery 2006; 93: 810–819
Published by John Wiley & Sons Ltd
814 P. A. Carless and D. A. Henry

Table 3 Potential confounding variables

Reference Compression dressings Drains Shoulder mobilization

Uden et al.24 Drains × 2 (Exudrain * at 21·5 kPa): Anterior

Vaxman et al.23 Pressure dressing was applied
at the end of surgery
Moore et al.22
Gilly et al.21 Pressure bandage was
applied for 2 days
Dinsmore et al.25
Moore et al.26
Langer et al.30
Ulusoy et al.29 An external non-compression
dressing was applied to all
patients
Jain et al.31 A compression dressing was
applied for 24 h after
surgery
Mustonen et al.28 An elastic compression
bandage around the
thoracic wall was applied for
all patients
Johnson et al.27 Only dry light dressings were
used
drain placed beneath the lower thoracic
skin flap, removed on the 1st day after
surgery. Posterior drain placed in the axilla:
removed when discharge < 100 ml/24 h
(usually 3–5 days)
Drains × 2 in each operative area: Active-assisted shoulder mobilization
mastectomy and axillary drains each drain consisted of 90◦ abduction, ante- and
suction was removed when daily volume retropulsion 5 times daily from day 1 after
was < 10 ml. Suction to gravity drainage surgery
from day 7 after surgery
Drains × 1: axillary drain (19-French round Patients instructed not to lift more than 10 lbs
Jackson-Pratt) removed once cumulative or to lift arm above shoulder level
drainage was < 40 ml/24 h
Drains × 1: axillary vacuum drain (at 21·5 Passive physiotherapy was performed 2–6
kPa), removed 6 days after surgery days after surgery
Drains × 1: axillary (10 mm Jackson-Pratt),
removed when output < 30 ml/24 h
Drains 1–2: axillary (lumpectomy) and skin Patients instructed to minimize arm
flap drains (MRM): removed when drain movement and not to raise arm above 90◦
output < 40 ml/24 h
Drains 1-2: TM/ALND received 1 drain Patients instructed to minimize arm
(10-French closed suction) and MRM movement until all drains were removed
received 2 drains, removed when drainage
< 30 ml for 2 consecutive days. All drains
removed by day 14 after surgery regardless
of drain output
Drains × 1: axillary suction drain (10 CH),
removed when drain output < 20 ml for
preceding 24 h
No drains used
Drains × 2 (HandyVac † CH18): axillary and Shoulder exercises were started in the 1st day
mastectomy drains were removed when after surgery
drain output < 50 ml/24 h
Drains × 1–2 (control group only): Arm exercises began on the 1st day after
Jackson-Pratt drains were placed in the surgery. Patients instructed to avoid lifting
axilla. No drains were used in the fibrin more than 10 lbs or lifting arm above
sealant group. Drains were removed when shoulder level for 2 days
drain output was < 30 ml/24 h

*Astra Meditec, Mölndal, Sweden; †Unomedical, Espoo, Finland. MRM, modified radical mastectomy; TM, total mastectomy; ALND, axillary lymph
node dissection.

was 1·14 (95 per cent confidence interval (c.i.) 0·88 to
1·46). The risk difference between fibrin sealant treatment
and control was 0·02 (95 per cent c.i. −0·02 to 0·06).
Heterogeneity of treatment effects was not significant
(P = 0·750). A fixed effects analysis provided similar results
(RR 1·17, 95 per cent c.i. 0·90 to 1·51).
Volume of drainage
Five trials reported data for the volume of drainage (Fig. 2).
These trials evaluated a total of 263 patients, of whom 127
were randomized to fibrin sealant. Although there was a
trend toward decreased drainage in patients receiving fibrin
sealant, the result was not significant (WMD − 117·7 ml,
95 per cent c.i. − 259·2 to 23·8 ml). Heterogeneity of
treatment effects was significant (P < 0·001).
Number of days of drainage
Four trials reported data for the number days that drains
remained in situ (Fig. 3). These trials included a total of 155
patients, of whom 77 were randomized to fibrin sealant.
Fibrin sealant treatment did not appear to impact on the
number of days drains were left in situ (WMD − 0·63 days,

Copyright  2006 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of Surgery 2006; 93: 810–819
Published by John Wiley & Sons Ltd
Fibrin sealants and seroma formation 815

Study Fibrin sealant Control RR (random) Weight (%) RR (random)

Mustonen et al. 200428 7 of 19 10 of 21 11·78 0·77 (0·37, 1·62)

Moore et al. 200126 13 of 59 6 of 21 9·38 0·77 (0·34, 1·77)

Jain et al. 200431 10 of 29 12 of 29 14·69 0·83 (0·43, 1·62)

Langer et al. 200330 1 of 26 1 of 29 0·87 1·12 (0·07, 16·95)

Gilly et al. 199821 1 of 50 1 of 58 0·86 1·16 (0·07, 18·07)

Uden et al. 199324 23 of 36 17 of 32 38·81 1·20 (0·80, 1·81)

Johnson et al. 200527 14 of 38 10 of 44 13·72 1·62 (0·82, 3·22)

Ulusoy et al. 200329 5 of 27 3 of 27 3·66 1·67 (0·44, 6·29)

25 6 of 14 3 of 13 4·78 1·86 (0·58, 5·94)
Dinsmore et al. 2000

Vaxman et al. 199523 4 of 20 1 of 20 1·46 4·00 (0·49, 32·72)

Total 84 of 318 64 of 294 100·00 1·14 (0·88, 1·46)

0·01 0·1 1 10 100

Favours Favours
FS control

Fig. 1 Number of patients developing postoperative seroma. Values in parentheses and whiskers represent 95 per cent confidence
intervals. Test for heterogeneity: χ2 = 8·25, 9 d.f., P = 0·710, I 2 = 0 per cent. Test for overall effect: Z = 0·98, P = 0·330. RR, relative
risk; FS, fibrin sealant

Study n Fibrin sealant* n Control* WMD (random)† Weight (%) WMD (random)†

Moore et al. 199722 11 346·10 (245·30) 10 745·70 (267·30) 15·96 −399·60 (−619·74, −179·46)
Gilly et al. 199821 50 214·40 (105·90) 58 407·80 (240·10) 24·53 −193·40 (−261·81, −124·99)

Ulusoy et al. 200329 27 738·48 (471·03) 27 886·44 (476·75) 14·21 −147·96 (−400·75, 104·83)

Mustonen et al. 200428 19 181·25 (141·01) 21 168·95 (140·95) 23·67 12·30 (−75·19, 99·79)
Vaxman et al. 199523 20 479·80 (190·30) 20 425·90 (212·60) 21·63 53·90 (−71·15, 178·95)

Total 127 136 100·00 −117·69 (−259·16, 23·77)

−1000 −500 0 500 1000

Favours Favours
FS control

Fig. 2Volume of postoperative drainage. * Values are mean(s.d.). † Values in parentheses and whiskers represent 95 per cent confidence
intervals. Test for heterogeneity: χ2 = 26·51, 4 d.f., P < 0·0001, I2 = 84·9 per cent. Test for overall effect: Z = 1·83, P = 0·100.
WMD, weighted mean difference; FS, fibrin sealant

95 per cent c.i. − 2·03 to 0·77 days). Heterogeneity of
treatment effects was also significant (P < 0·001).
Wound infection
Five trials reported data for wound infection (Fig. 4).
These included a total of 324 patients, of whom 177
were randomized to fibrin sealant. The use of fibrin sealant
did not appear to affect the rates of wound infection. The
pooled RR of developing a wound infection in patients
treated with fibrin sealant was 0·95 (95 per cent c.i. 0·30
to 3·01). Heterogeneity of treatment effects was not
significant (P = 0·500). A fixed effects analysis provided
similar results (RR 0·99, 95 per cent c.i. 0·35 to 2·81).
Duration of hospital stay
Four trials reported data for the duration of hospital stay
(Fig. 5). These included a total of 209 patients, of whom
100 were randomized to fibrin sealant. Fibrin sealant use

Copyright  2006 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of Surgery 2006; 93: 810–819
Published by John Wiley & Sons Ltd
816 P. A. Carless and D. A. Henry

Study n Fibrin sealant* n Control* WMD (random)† Weight (%) WMD (random)†

Moore et al. 199722 11 3·90 (1·70) 10 6·90 (1·19) 24·42 −3·00 (−4·25, −1·75)

Mustonen et al. 200428 19 2·75 (1·02) 21 2·70 (0·81) 28·79 0·50 (−0·52, 0·62)
Vaxman et al. 199523 20 4·08 (2·04) 20 3·92 (1·59) 25·26 0·16 (−0·97, 1·29)
Ulusoy et al. 200329 27 9·88 (3·07) 27 9·66 (3·01) 21·53 0·22 (−1·40, 1·84)
Total 77 78 100·00 −0·63 (−2·03, 0·77)

−10 −5 0 5 10
Favours Favours
FS control

Number of days drainage was left in situ. * Values are mean(s.d.). † Values in parentheses and whiskers represent 95 per cent
Fig. 3
confidence intervals. Test for heterogeneity: χ2 = 20·47, 3 d.f., P = 0·0001, I 2 = 86·3 per cent. Test for overall effect: Z = 0·88,
P = 0·380. WMD, weighted mean difference; FS, fibrin sealant

Study Fibrin sealant Control RR (random) Weight (%) RR (random)

Uden et al. 199324 0 of 36 0 of 32

Mustonen et al. 200428 0 of 19 2 of 21 14·91 0·22 (0·01, 4·31)

Langer et al. 200330 0 of 26 1 of 29 13·24 0·37 (0·02, 8·71)

Johnson et al. 200527 3 of 38 3 of 44 55·64 1·16 (0·25, 5·40)

Moore et al. 200126 5 of 58 0 of 21 16·21 4·10 (0·24, 71·15)

Total 8 of 177 6 of 147 100·00 0·95 (0·30, 3·01)

0·01 0·1 1 10 100

Favours Favours
FS control

Fig. 4Number of patients developing a wound infection. Value in parentheses and whiskers represent 95 per cent confidence intervals.
Test for heterogeneity: χ2 = 2·35, 3 d.f., P = 0·500. I 2 = 0 per cent. Test for overall effect: Z = 0·08, P = 0·94. RR, relative risk; FS,
fibrin sealant

did not appear to have a significant impact on the on the was judged to be inadequate for all 11 trials. The small
duration of hospital stay (WMD − 0·38 days, 95 per cent number of trials precluded any formal assessment of the
c.i. − 1·58 to 0·83 days). Heterogeneity of treatment effects relationship between trial quality and effect size.
was significant (P < 0·001).

Methodological quality of trials
For the four items of the Schulz criteria, the inter-rater
agreement for the assessment of methodological quality
was generally good, with kappa (κ) scores ranging from
0·69 to 1·0. Overall, the methodological quality of the
11 trials was poor. None had full blinding of outcome
assessment, although partial blinding (single blinding) was
reported in four. Only one reported the method used
for randomization. Follow-up of patients appeared to
have been complete in most studies (six of 11), with two
reporting only a small number of exclusions. Three failed
to report on patient withdrawal. Allocation concealment
Discussion
Overall, the meta-analysis indicated that fibrin sealant
use in breast cancer surgery was not effective in
preventing seroma formation or decreasing the volume of
postoperative drainage. In fact, there was a non-significant
trend towards an increased risk of seroma formation
in patients treated with fibrin sealant (RR 1·14). It is
important to note that the baseline rates of seroma
varied considerably between trials, ranging from as low
as 1·7 per cent to as high as 53·1 per cent. Such variation
may be anticipated given the wide scope of clinical
heterogeneity in trials involving breast cancer surgery,

Copyright  2006 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of Surgery 2006; 93: 810–819
Published by John Wiley & Sons Ltd
Fibrin sealants and seroma formation
Study n Fibrin sealant* n Control*
Gilly et al. 199821 50 8·00 (1·60) 58 10·10 (2·10)
Moore et al. 199722 11 1·18 (0·60) 10 1·41 (0·69)
Mustonen et al. 200428 19 3·90 (1·00) 21 3·70 (1·00)
Vaxman et al. 199523 20 10·80 (3·50) 20 9·40 (3·50)
Total 100 109
Number of days patients spent in hospital. * Values are mean(s.d.). † Values in parentheses and whiskers represent 95 per cent
Fig. 5
confidence intervals. Test for heterogeneity: χ2 = 28·85, 3 d.f., P < 0·00 001, I 2 = 89·6 per cent. Test for overall effect: Z = 0·61,
P = 0·540. WMD, weighted mean difference; FS, fibrin sealant
with different surgical techniques, surgeon skill mixes
and postoperative management strategies. However, the
observed variation may also have been due to differences in
the methods used to detect seromas or diagnostic criteria.
Predefined subgroup analyses to determine whether effect
sizes varied according to surgical procedure and the type of
fibrin sealant used were uninformative owing to the small
number of trials.
Fibrin sealant had an equivocal effect on drainage output,
with a non-significant trend towards reduced drainage
in treated subjects (WMD − 117·7 ml, 95 per cent c.i.
− 259·2 to 23·8 ml). Two of the five trials that reported
data for this outcome22,22 found that fibrin sealant was
significantly more effective than control in reducing the
total volume of drainage. The dose of fibrin sealant did not
appear to have any bearing on the result. Of the two positive
trials, one used a high dose of fibrin sealant (approximately
760 mg total fibrinogen content)22 while the other used
a relatively low dose (150–230 mg total fibrinogen
content)21 . In the case of the three negative trials, doses
ranged from 150–230 mg to 375–575 mg23,28,29 .
Of the four trials that investigated seroma formation as
the primary study outcome, none was powered adequately
(1–β = 0·1187 to 0·2669). Although three of these trials
had based their sample sizes on detecting a difference in
seroma rates of between 20 and 25 per cent, the actual
differences in seroma rates for these trials ranged from 0
to 11 per cent24,28,31 . It is disappointing that one trial that
had planned a sample size of 250 could not achieve this
owing to the withdrawal of two principal surgeons, and
instead included just 82 subjects27 .
Overall, the methodological quality of the trials was
poor, with only one trial reporting the method used for
Copyright  2006 British Journal of Surgery Society Ltd
Published by John Wiley & Sons Ltd
817
WMD (random)† Weight (%) WMD (random)†
27·70 −2·10 (−2·80, −1·40)
28·69 −0·23 (−0·79, 0·33)
28·26 0·20 (−0·42, 0·82)
15·35 1·40 (−0·77, 3·57)
100·00 −0·38 (−1·58, 0·83)
−4 −2 0 2 4
Favours Favours
FS Control
randomization. Full blinding of outcomes assessment did
not occur in any trial, and only three used an adequate
method for concealing treatment allocation. Of the four
trials that were judged to be single-blind, two reported
blinding at the patient level22,31 and two at the outcome
assessors’ level21,26 . The lack of blinding and allocation
concealment is problematic when the outcome to be
measured relied on the interpretation of clinical findings,
such as the presence of seroma or wound infection, or
on a clinical judgement, such as aspiration of seroma or
discharge from hospital, as opposed to the measuring of
serosanguinous or lymphatic fluid collected in a closed
drainage system. Given the lack of detail in the reports,
it is difficult to quantify the impact of bias on effect. It
is interesting to note that a non-randomized, prospective
study by Medl et al.32 (n = 80) showed that fibrin sealant
treatment in breast cancer surgery did not affect the rates
of seroma (P = 0·606), days of drainage (P = 0·191) or
volume of drainage (P = 0·772) compared with control.
An additional consideration is the possibility of
publication bias, where positive studies are more likely
to be published than negative studies33 . A statistical test for
funnel plot asymmetry was performed using the method
of analysis proposed by Egger et al.34 . This indicated that
there was little evidence of publication bias (P = 0·431). It
is important to recognize that the power of this method
to detect bias is low when the number of trials analysed is
small. However, given that most trials reported either non-
significant or negative findings, publication bias is unlikely
to have affected the findings of this study.
Another factor to consider is language bias. Two foreign
language, randomized controlled trials, identified in the
unrestricted literature searches35,36 showed conflicting
www.bjs.co.uk British Journal of Surgery 2006; 93: 810–819
818
results. The trial by Gioffre-Florio et al.35 , involving
segmental mastectomy (n = 12) and modified radical
mastectomy (n = 12), found that the use of fibrin sealant
significantly reduced drainage volumes over 5 to 6 days
compared with control (169·5, s.d. 9·0, ml versus 219·3,
s.d. 15·2, ml; P < 0·001). In contrast, Nielsen et al.36
found that fibrin sealant was unable to reduce the volume
of drainage in patients undergoing mastectomy with the
2 ml dose (n = 10) having no effect compared with control
(n = 20) and the 4 ml dose (n = 8) causing prolonged
drainage.
The cost of fibrin sealant also warrants discussion. In
the trial by Johnson et al.27 , patients were charged US$440
for fibrin sealant treatment (2 ml kit, Hemaseel APR).
Moore et al.26 estimated the cost of fibrin sealant used
in their trial to be between US$180 and US$2160 per
patient. In contrast, the fibrin sealant Mustonen et al.28
used in their trial cost less than 200 euros (approximately
US$240) per patient. It has been suggested that fibrin
sealant treatment might save costs but this is based on the
premise that using fibrin sealant could completely eliminate
the need for drains26 . However, Johnson et al.27 compared
the costs of fibrin sealant treatment with conventional
drain placement and estimated the incremental cost of
fibrin sealant treatment to be US$366 per patient when
one drain was used and US$293 when two drains were
used. Given the lack of evidence of efficacy, the question
of cost-effectiveness is rather a moot point. Based on the
evidence reviewed here, fibrin sealant treatment is not a
cost-effective alternative to other treatments currently used
to prevent seroma formation.
There is no evidence to support the use of fibrin sealant
in breast cancer surgery to prevent seroma formation
or reduce the volume of postoperative drainage. The
trials are small, underpowered and lack the necessary
methodological rigour required for clinicians to have
confidence in the findings. On the basis of the evidence
reviewed, further research involving large, randomized
clinical trials may be warranted.
Acknowledgements
This study was funded by a grant from the National Health
and Medical Research Council of Australia.
References
1 American Cancer Society. Cancer Facts & Figures 2005.
Atlanta: American Cancer Society 2005; 1–64.
2 Purushotham AD, Upponi S, Klevesath MB, Bobrow L,
Millar K, Myles JP et al. Morbidity after sentinel lymph node
Copyright  2006 British Journal of Surgery Society Ltd
Published by John Wiley & Sons Ltd
P. A. Carless and D. A. Henry
biopsy in primary breast cancer: results from a randomized
controlled trial. J Clin Oncol 2005; 23: 4312–4321.
3 Lumachi F, Brandes AA, Burelli P, Basso SM, Iacobone M,
Ermani M. Seroma prevention following axillary dissection
in patients with breast cancer by using ultrasound scissors: a
prospective clinical study. Eur J Surg Oncol 2004; 30:
526–530.
4 Gonzalez EA, Saltzstein EC, Riedner CS, Nelson BK.
Seroma formation following breast cancer surgery. Breast J
2003; 9: 385–388.
5 Pogson CJ, Adwani A, Ebbs SR. Seroma following breast
cancer surgery. Eur J Surg Oncol 2003; 29: 711–717.
6 Moore MM, Freeman MG. Fibrin sealant in breast surgery.
J Long Term Eff Med Implants 1998; 8: 133–142.
7 Burak WE, Jr., Goodman PS, Young DC, Farrar WB.
Seroma formation following axillary dissection for breast
cancer: risk factors and lack of influence of bovine thrombin.
J Surg Oncol 1997; 64: 27–31.
8 Coveney EC, O’Dwyer PJ, Geraghty JG, O’Higgins NJ.
Effect of closing dead space on seroma formation after
mastectomy – a prospective randomized clinical trial. Eur J
Surg Oncol 1993; 19: 143–146.
9 Shamley DR, Barker K, Simonite V, Beardshaw A. Delayed
versus immediate exercises following surgery for breast
cancer: a systematic review. Breast Cancer Res Treat 2005; 90:
263–271.
10 O’Hea BJ, Ho MN, Petrek JA. External compression
dressing versus standard dressing after axillary
lymphadenectomy. Am J Surg 1999; 177: 450–453.
11 Galatius H, Okholm M, Hoffmann J. Mastectomy using
ultrasonic dissection: effect on seroma formation. Breast
2003; 12: 338–341.
12 Talbot ML, Magarey CJ. Reduced use of drains following
axillary lymphadenectomy for breast cancer. ANZ J Surg
2002; 72: 488–490.
13 Sierra DH. Fibrin sealant adhesive systems: a review of their
chemistry, material properties and clinical applications.
Journal of Biomaterials Applications 1993; 7: 309–352.
14 Lindsey WH, Masterson TM, Spotnitz WD, Wilhelm MC,
Morgan RF. Seroma prevention using fibrin glue in a rat
mastectomy model. Arch Surg 1990; 125: 305–307.
15 Eroglu E, Oral S, Unal E, Kalayci M, Oksuz O, Tilmaz M.
Reducing seroma formation with fibrin glue in an animal
mastectomy model. Eur J Surg Oncol 1996; 22: 137–139.
16 Harada RN, Pressler VM, McNamara JJ. Fibrin glue reduces
seroma formation in the rat after mastectomy. Surg Gynecol
Obstet 1992; 175: 450–454.
17 Higgins JPT, Green S (eds). Cochrane Handbook for Systematic
Reviews of Interventions 4·2·5 [updated May 2005]. In The
Cochrane Library, Issue 3, 2005 John Wiley & Sons:
Chichester. Online at http://www.cochrane.org/resources/
handbook/handbook pdf.
18 Dickersin K, Scherer R, Lefebvre C. Identifying relevant
studies for systematic reviews. Br Med J 1994; 309:
1286–1291.
www.bjs.co.uk British Journal of Surgery 2006; 93: 810–819
Fibrin sealants and seroma formation
19 Schulz KF, Chalmers I, Hayes RJ, Altman DJ. Empirical
evidence of bias. Dimensions of methodological quality
associated with estimates of treatment effects in controlled
trials. JAMA 1995; 273: 408–412.
20 DerSimonian R, Laird N. Meta-analysis in clinical trials.
Control Clin Trials 1986; 7: 177–188.
21 Gilly FN, Francois Y, Sayag-Beaujard AC, Glehen O,
Brachet A, Vignal J. Prevention of lymphorrhea by means of
fibrin glue after axillary lymphadenectomy in breast cancer:
prospective randomized trial. Eur Surg Res 1998; 30:
439–443.
22 Moore MM, Nguyen DH, Spotnitz WD. Fibrin sealant
reduces serous drainage and allows for earlier drain removal
after axillary dissection: a randomized prospective trial. Am
Surg 1997; 63: 97–102.
23 Vaxman F, Kolbe A, Stricher F, Zund D, Volkmar P, Gros D
et al. Does fibrin glue improve drainage after axillary lymph
node dissection? Prospective and randomized study in
humans. Eur Surg Res 1995; 27: 346–352.
24 Uden P, Aspegren K, Balldin G, Garne JP, Larsson SA.
Fibrin adhesive in radical mastectomy. Eur J Surg 1993; 159:
263–265.
25 Dinsmore RC, Harris JA, Gustafson RJ. Effect of fibrin
glue on lymphatic drainage after modified radical
mastectomy: a prospective randomized trial. Am Surg 2000;
66: 982–985.
26 Moore M, Burak WE JR, Nelson E, Kearney T, Simmons R,
Mayers L et al. Fibrin sealant reduces the duration and
amount of fluid drainage after axillary dissection: a
randomized prospective clinical trial. J Am Coll Surg 2001;
192: 591–599.
27 Johnson L, Cusick TE, Helmer SD, Osland JS. Influence of
fibrin glue on seroma formation after breast surgery. Am J
Surg 2005; 189: 319–323.
Copyright  2006 British Journal of Surgery Society Ltd
Published by John Wiley & Sons Ltd
819
28 Mustonen PK, Harma MA, Eskelinen MJ. The effect of
fibrin sealant combined with fibrinolysis inhibitor on
reducing the amount of lymphatic leakage after axillary
evacuation in breast cancer. A prospective randomized
clinical trial. Scand J Surg 2004; 93: 209–212.
29 Ulusoy AN, Polat C, Alvur M, Kandemir B, Bulut F. Effect
of fibrin glue on lymphatic drainage and on drain removal
time after modified radical mastectomy: a prospective
randomized study. Breast J 2003; 9: 393–396.
30 Langer S, Guenther JM, DiFronzo LA. Does fibrin sealant
reduce drain output and allow earlier removal of drainage
catheters in women undergoing operation for breast cancer?
Am Surg 2003; 69: 77–81.
31 Jain PK, Sowdi R, Anderson AD, MacFie J. Randomized
clinical trial investigating the use of drains and fibrin sealant
following surgery for breast cancer. Br J Surg 2004; 91:
54–60.
32 Medl M, Mayerhofer K, Peters-Engl C, Mahrhofer P,
Huber S, Buxbaum P et al. The application of fibrin glue
after axillary lymphadenectomy in the surgical treatment of
human breast cancer. Anticancer Res 1995; 15: 2843–2845.
33 Sterne JA, Egger M, Smith GD. Systematic reviews in health
care: Investigating and dealing with publication and other
biases in meta-analysis. BMJ 2001; 323: 101–105.
34 Egger M, Davey SG, Schneider M, Minder C. Bias in
meta-analysis detected by a simple, graphical test. BMJ 1997;
315: 629–634.
35 Gioffre FM, Mezzasalma F, Manganaro T, Pakravanan H,
Cogliandolo A. [The use of fibrin glue in the surgery of
breast carcinoma.] Gi Chir 1993; 14: 239–241.
36 Nielsen BP, Willumsen H, Rasmussen OV. [The secretion
inhibition effect of fibrin glue in mastectomy scars. A
prospective, randomized, clinically controlled study.] Ugeskr
Laeger 1985; 147: 1995–1996.
www.bjs.co.uk British Journal of Surgery 2006; 93: 810–819