REVIEW
A Meta-Analysis Comparing Muscle-Sparing and
Posterolateral Thoracotomy
Mohammed M. Uzzaman, MRCS, MS, J. Daniel Robb, FRCS, MRCP,
Peter C. E. Mhandu, MRCS, MCChB, Habib Khan, MRCS, MBBS,
Kamran Baig, FRCS, MD, Sanjay Chaubey, MRCS, and Donald C. Whitaker, FRCS, MD
Department of Cardiothoracic Surgery, King’s College Hospital, London, United Kingdom
We compared outcomes of posterolateral thoracotomy vs
muscle-sparing thoracotomy after open thoracic operations. Twelve trials were included, comprising 571 patients in the muscle-sparing thoracotomy group and 512
patients in the posterolateral thoracotomy group. There
was significantly improved shoulder internal rotation
(weighted mean difference, –1.28; 95% confidence interval,
–2.45 to –0.11; p [ 0.03) and pain scores on day 7 (weighted
mean difference, –0.76; 95% confidence interval, –1.26
to –0.27; p [ 0.002) but higher seroma rates (odds ratio,
8.26; 95% confidence interval, 2.16 to 31.56; p [ 0.002) in
the muscle-sparing thoracotomy group compared with
the posterolateral thoracotomy group. We advocate
using muscle-sparing thoracotomy, especially on patients
dependant on quicker recovery of shoulder function.
A
Several trials have assessed MST compared with PLT
after thoracic procedures [10–22]. A meta-analysis of
these trials allows a pooled analysis, thereby minimizing
any type II error. The purpose of this meta-analysis was to
assess the clinical outcomes of MST compared with PLT
after open thoracic procedures.
Address correspondence to Dr Uzzaman, Department of Cardiothoracic Surgery, King’s College Hospital, Bessemer Rd, London SE59RS,
UK; e-mail: mohsinuzzaman@yahoo.co.uk.
Ó 2014 by The Society of Thoracic Surgeons
Published by Elsevier Inc
Material and Methods
An electronic search was performed using the MEDLINE,
CINAHL, DARE, ACP, LILACS, SCOPUS, Google
Scholar, and EMBASE databases from 1966 to 2010. The
search terms “Thoracotomy,” “Posterolateral,” “Muscle
sparing,” “Minimal invasive,” “Thoracic,” “Limited,”
“Standard,” “Randomised trial,” “Randomized trial,”
“Trial,” “Prospective,” “Retrospective,” “Anterolateral,”
“Axillary,” “Vertical,” and “Anterior,” and MeSH headings “Thoracotomy” (MeSH), “Thoracic” (MeSH) “Surgery” (MeSH) were used in combination with the Boolean
operators “and” or “or.” The criteria were widened
further by using the “related article” function during the
search. Two authors conducted the electronic searches
independently in March 2012.
This search was supplemented by a hand search of
published abstracts from 1980 to 2010 in meetings of the
Society of Academic and Research Surgery, Surgical
Research Society, Society of Cardiothoracic Surgeons
(SCTS), World Congress of Cardiothoracic Surgery, The
European Society of Cardiothoracic Surgeons, and The
American Association of Thoracic Surgery (AATS).
Finally, the Current Controlled Trials Register, The
Cochrane Database Of Controlled Trials, and Science
Citation Index Expanded were searched.
The reference lists of all articles obtained were also
examined to identify additional relevant studies. Review
0003-4975/$36.00
http://dx.doi.org/10.1016/j.athoracsur.2013.08.014
REVIEW
posterolateral thoracotomy (PLT) in which muscles
are divided is still the standard approach for most
thoracic surgical procedures. It has the advantage of
providing good access and can be easily and quickly
extended if greater access becomes necessary. Although
the choice of thoracotomy incision is guided primarily by
the exposure required to perform a safe procedure,
cosmesis and the potential for improved recovery should
also be considered. PLT requires the division of the latissimus dorsi (LD) and sometimes other chest wall
muscles such as trapezius, the rhomboids, and serratus
anterior (SA). As a result, PLT may be associated with
considerable morbidity, including postoperative pain,
impaired lung function, and compromised function of the
shoulder girdle [1–3].
In an attempt to decrease these shortcomings, less
invasive thoracotomy procedures, such as muscle-sparing
thoracotomy (MST) [4–9] and video-assisted thoracoscopic surgery (VATS), [10] have been introduced.
Various types of MSTs have been proposed during the
last 25 years, including vertical and transverse axillary
thoracotomy [4, 5], auscultatory triangle thoracotomy [6],
limited lateral thoracotomy [7], and muscle-sparing
posterolateral thoracotomy [8]. By using a more limited
muscle dissection than PLT, there is the potential for MST
to reduce postoperative pain, allow better respiratory and
lung function by preserving accessory muscles, and
reduce postoperative complications. However, the difference in the postoperative outcomes between these
techniques remains undetermined.
(Ann Thorac Surg 2014;97:1093–102)
Ó 2014 by The Society of Thoracic Surgeons
1094
REVIEW
UZZAMAN ET AL
META-ANALYSIS OF THORACOTOMY TECHNIQUES
articles were obtained to determine other possible
studies. Trials were included irrespective of the language
in which they were published. We applied the guidelines
for Meta-analysis of Observational Studies in Epidemiology (MOOSE) [23].
Ann Thorac Surg
2014;97:1093–102
the Cochran Q value and quantified using the I2 inconsistency test. In this study, we did not perform metaregression or sensitivity analysis because of the small
number of studies included. All p values were 2-sided,
and a 5% level was considered significant.
Study Eligibility
REVIEW
Eligibility for study inclusion into the meta-analysis and
study quality assessment was performed independently
by 2 of the authors (M.U. and P.M.). Study data were
extracted onto standard forms. Any disagreements were
resolved by the third author (D.R.). Studies were included
if they were trials in which direct comparisons were made
between patients who had PTL vs MST during a thoracic
surgical procedure. A posterolateral thoracotomy procedure must involve division of the latissimus dorsi muscle.
An MST approach does not allow division of the latissimus
dorsi muscle but can allow division of other muscles such
as SA or trapezius. Any unclear or missing information
was obtained by contacting the authors of the individual
trials. For duplicate publications, the smaller data set was
excluded. Any data directly comparing a VAT approach
vs thoracotomy techniques were excluded from the
analysis.
The quality of each study was assessed by use of
Newcastle-Ottawa scale, a 9-point scale that assigns
points on the basis of the process of selection (0 to 4
points), comparability (0 to 2 points), and identification of
the outcomes of study participants (0 to 3 points) [24].
The primary outcome measures were lung function
tests measured by forced expiratory volume in 1 second
(FEV1) and vital capacity (VC), range of shoulder movement (in degrees) at 30 days, and pain score on a linear
visual analog scale at 1, 7, and 30 days after the operation.
Secondary outcomes were duration of operation
(in minutes), incision size (in cm), hospital stay (in days),
and overall documented complication rates, including
seroma. The duration of operation was defined as the
time taken to perform the procedure starting from knifeto-skin through completion of skin closure. Only trials
that reported at least one of the primary or secondary
outcome measures were included in the meta-analysis.
Statistical Analysis
Data from the individual eligible studies were entered
into a spreadsheet for further analysis. StatsDirect 2.5.7
software (StatsDirect, Altrincham, UK) was used to
perform the statistical analysis. Weighted mean differences (WMD) were calculated for the effect size of
continuous variables such as duration of surgery
and postoperative visual analog scale pain scores.
Pooled odds ratios (OR) were calculated for discrete
variables such as complication and seroma rate. A
positive WMD and an OR of less than 1.00 favors PLT,
whereas a negative WMD or an OR of less than 1.00
favors MST.
Random effects models (DerSimion Laird) were used to
calculate the outcomes of binary and continuous data to
control any heterogeneity among the studies. Heterogeneity among the trials was determined by means of
Results
The initial search identified 58 publications (Fig 1). The
meta-analysis included 13 publications describing 12 trials that fulfilled the inclusion criteria [10–22]. Publication
dates ranged from 1991 to 2010. There were 1,083 patients,
of whom 512 had PLT and 571 had MST. There were no
baseline imbalances in age, sex, or body mass index between the two groups. Trial details are reported in Table 1
and Table 2, including details of study design, sample
size, intervention, and follow-up.
Primary Outcome
LUNG FUNCTION TESTS. Six studies reported the percentage
change in FEV1 after 30 days [10, 12, 14, 16, 18, 21]. There
was no statistical heterogeneity between studies (Cochran
Q ¼ 3.13, p ¼ 0.68; I2 ¼ 0%, 95% confidence interval [CI],
0% to 61%). In the random effects model, there was no
significant difference in the FEV1 compared with preoperative levels between the PLT and MST group (pooled
WMD, 2.74; 95% CI, –0.92 to 6.41; p ¼ 0.14; Table 3).
Five studies reported the percentage change in VC after
30 days [12, 14, 15, 18, 21]. There was statistical heterogeneity between studies (Cochran Q ¼ 11.39, p ¼ 0.02; I2 ¼
64.9%, 95% CI, 0% to 84.5%). In the random effects model,
there was no significant difference in the VC compared
with preoperative levels between the PLT and MST group
(pooled WMD, 3.10; 95% CI, –3.05 to 9.24; p ¼ 0.32;
Table 3).
SHOULDER MOVEMENTS. Three studies reported the range of
shoulder movement after 30 days [12, 18, 21].
For shoulder abduction, there was no statistical heterogeneity between studies (Cochran Q ¼ 0.40, p ¼ 0.82;
I2 ¼ 0%, 95% CI, 0% to 72.9%). In the random effects
model, there was no significant difference in shoulder
abduction between the PLT and MST group at 30 days
(pooled WMD, 2.33; 95% CI, –0.69 to 5.35; p ¼ 0.13;
Table 3).
For shoulder flexion, there was statistical heterogeneity
between studies (Cochran Q ¼ 11.95, p ¼ 0.002; I2 ¼ 83.3%,
95% CI, 2.8% to 92.7%). In the random effects model,
there was no significant difference in shoulder flexion
between the PLT and MST group at 30 days (pooled
WMD, 8.18; 95% CI, –1.65 to 18.01; p ¼ 0.10; Table 3).
For internal rotation, there was no statistical heterogeneity between studies (Cochran Q ¼ 2.24, p ¼ 0.32; I2 ¼
10.9%, 95% CI, 0% to 75.7%). In the random effects model,
postoperative internal rotation was significantly better at
30 days in the MST group than in the PLT group (pooled
WMD, –1.28; 95% CI, –2.45 to –0.11; p ¼ 0.03; Fig 2 and
Table 3).
For external rotation of the shoulder, there was statistical heterogeneity between studies (Cochran Q ¼ 14.22,
REVIEW
UZZAMAN ET AL
META-ANALYSIS OF THORACOTOMY TECHNIQUES
1095
Fig 1. Quality of reporting of meta-analyses diagram for the study. (RCT ¼ randomized controlled trial; VATS ¼ video-assisted thoracoscopic
surgery.)
p ¼ 0.00; I2 ¼ 85.9%, 95% CI, 34% to 93.6%). In the random
effects model, there was no significant difference in
postoperative external rotation between the PLT and
MST group (pooled WMD, –3.01; 95% CI, –8.57 to 2.55;
p ¼ 0.29; Table 3).
PAIN SCORES. Six studies reported the pain scores on day 1
after the procedure [10, 14, 15, 18, 21, 22]. There was
statistical heterogeneity between studies (Cochran Q ¼
77.55, p ¼ 0.00; I2 ¼ 93.6%, 95% CI, 89.3% to 95.6%). In
the random effects model, there was no significant
Table 1. General Characteristics of Included Trials
Study
Year
Nosotti [10]
Endoh [11]
2010
2010
Athanassiadi [12]
Ochroch [13]
Akcali [14]
Nomori [15]
Benedetti [17]
Sugi [18]
Landreneau [19]
Ponn [20]
Hazelrigg [21]
Lemmer [22]
2007
2005
2003
1999
1998
1996
1996
1991
1991
1990
NS ¼ not specified;
Trial
Surgeon
Study Period
Follow-Up
Single RCT
Retrospective
Single surgeon
Various surgeons
Jul 2003–2006
Oct 2001–Mar 2007
Single-center RCT
Single-center RCT
Single-center RCT
Prospective
Prospective
Single-center RCT
Retrospective
Prospective
Single-center RCT
Single-center RCT
2 surgeons
3 surgeons
NS
2 surgeons
NS
Single surgeon
Various surgeons
Various surgeons
NS
NS
Jun 2004–Dec 2004
Mar 1998–Jul 1999
1999–2000
Jan 1993–Dec 1997
NS
NS
NS (40 months)
NS
NS
NS
3 years
Hospital stay
(minimum 7 days)
60 days
48 weeks
1 month
6 months
1 month
30 days
1 year
6.4 2.8 months
1 month
24 hours
RCT ¼ randomized controlled trial.
Newcastle-Ottawa
Score
8
6
8
7
7
6
5
5
5
5
7
5
REVIEW
Ann Thorac Surg
2014;97:1093–102
REVIEW
1096
Sample Size, No.
Study
Selection Criteria
Inclusion
Exclusion
Intervention
Experimental
Control
PLT
MST
Nosotti [10]
100
50
50
Stage I/II non-small
cell lung cancer
Pneumonectomy, chest
wall extension,
epidural
Standard way
12-15 cm incision;
auscultatory triangle;
no muscle division
Endoh [11]
73
26
47
Lung cancer
Prolonged pulmonary
leakage, unable to
perform tests due to
complications
15-cm incision;
division of LD
and rib
Athanassiadi [12]
100
50
50
Primary elective
lung operation
LD and lower edge
of SA divided
Ochroch [13]
120
82
38
Segmentectomy,
lobectomy,
bilobectomy
Akcali [14]
60
30
30
Various thoracic
diseases
Thoracic trauma,
thoracic surgery,
pneumonectomies,
chest wall or brachial
plexus involvement,
atypical resections
Pneumonectomy,
chronic pain, opioid
use, neurologic
impairment, CCS
III/IV, severe
pulmonary disease,
unable to have
epidural
Not stated
8-13 cm incision; ALT,
AAT and vertical
approach; division or
partial resection
anterior SA
SA was freed to
inferior aspects
(anterior border
6th rib)
Nomori [15, 16]
84
56
28
Lung cancer
Benign operation
30-35 cm incision,
Lower trapezius and
entire LD and SA
divided
Benedetti [17]
24
13
11
Segmentectomy
Previous thoracotomy
Division of LD þ/– SA/
trapezius
Incision width of LD;
LD divided þ/–
resection posterior
6th rib. SA spared.
7-cm incision anterior
to LD, LD spared;
insertion of SA on
4th and 5th rib
dissected
Division of entire
LD and lower edge
of SA
LD and SA both freed
and retracted
posteriorly/
anteriorly
respectively
(1) 12-cm ALT incision,
dividing pectoralis
major and splitting
SA. (2) 20- to 25-cm
AAT incision,
dividing SA and
pectoralis major
Incision along
auscultatory triangle,
muscle spared
Post-op Analgesia/
Care
Patient controlled
analgesia for 72
hours, ketorolac then
paracetamol and
codeine as required.
No drains
Epidural catheter
30-mg subcutaneous
opioids for 48 hours,
then oral
paracetamol or
codeine. 2 chest
drains
Epidural for 3-5 days,
as required
ketorolac; beyond
this, oral ibuprofen,
codeine and
acetaminophen
Intercostal nerve block,
IV/IM pethidine for
24 hours. 2 soft
drains in the
subcutaneous layer
Epidural analgesia for
8 days;
indomethacin
suppository as
required
Not stated
(Continued)
Ann Thorac Surg
2014;97:1093–102
Total
REVIEW
UZZAMAN ET AL
META-ANALYSIS OF THORACOTOMY TECHNIQUES
Table 2. Specific Characteristics of Included Trials
Sample Size, No.
Study
Selection Criteria
Experimental
Control
30
15
15
Stage 1/2 lung cancer
Benign cases
Division of entire
LD and lower edge
of SA
335
148
187
Pneumonectomy,
chest wall resection
Lobectomy for stage I
lung cancer
Division of entire
LD and lower edge
of SA
Ponn [20]
79
62
17
Segmentectomy,
lobectomy,
pneumonectomy
Concomitant chest
wall resection,
diaphragmatic, or
other radical
resections; initially
wedge resections of
small peripheral
lesions
LD and anterior border
of SA/trapezius
divided
Hazelrigg [21]
50
26
24
Various thoracic
disease requiring
thoracotomy
Previous thoracotomy,
vascular procedures,
thoracotomy in
lower intercostal
space
Division of entire
LD and lower edge
of SA
Lemmer [22]
28
13
15
ASA 1 or 2, elective
wedge, lobectomy or
mediastinal
resection
Pneumonectomy;
ASA > 2
Division of trapezius,
LD and SA
Landreneau [19]
Exclusion
PLT
MST
Both LD and SA were
freed and retracted
posteriorly and
anteriorly,
respectively
(1) Axillary approach:
anterior slips of SA
3rd/4th rib divided
(2) lateral approach
10- to 12-cm sparing
all muscles
(1) Axillary approach:
SA divided in line
with fibers. (2)
Limited lateral; LD
and SA both freed
and retracted.
Occasionally,
serratus insertions
freed from ribs
LD and SA both freed
and retracted
posteriorly/
anteriorly
respectively
Both LD and SA were
freed and retracted;
partial division of SA
if needed
Post-op Analgesia/
Care
Rectal diclofenac as
required. 2 soft
drains in the
subcutaneous layer
Intercostal nerve block,
epidural catheter for
4 days; after, patientcontrolled analgesia
and oral
medications. 28F
chest drain and
Jackson-Pratt drain
in subcutaneous
flaps.
Epidural catheter
for 1-4 days.
Jackson-Pratt drain
to evacuate
subcutaneous flap
Single intercostal
nerve block
(Marcaine) above
and below
intercostal space.
Jackson-Pratt drain
for at least 4 days
Morphine in 2-mg
increments for 24
hours
þ/– ¼ with or without;
AAT ¼ anteroaxillary thoracotomy;
ALT ¼ anterior limited thoracotomy;
ASA ¼ American Society of Anesthesiologists;
CSS ¼ Charlson Comorbidity Score;
intramuscular;
IV ¼ intravenous;
LD ¼ latissimus dorsi;
MST ¼ muscle-sparing thoracotomy;
PLT ¼ posterolateral thoracotomy;
SA ¼ serratus anterior.
IM ¼
REVIEW
UZZAMAN ET AL
META-ANALYSIS OF THORACOTOMY TECHNIQUES
Total
Sugi [18]
Inclusion
Intervention
Ann Thorac Surg
2014;97:1093–102
Table 2. Continued
1097
REVIEW
1098
REVIEW
UZZAMAN ET AL
META-ANALYSIS OF THORACOTOMY TECHNIQUES
Ann Thorac Surg
2014;97:1093–102
Table 3. Summary of Results of Meta-Analysis for Individual Primary and Secondary Outcomes
Variable
Primary outcome
FEV1 % (30 days)
VC % (30 days)
Abduction
Flexion
Internal rotation
External rotation
Pain day 1
Pain day 7
Pain day 30
Secondary outcomes
Incision size
Duration of
operation
Hospital stay
Total complication
Seroma
CI ¼ confidence interval;
posterolateral thoracotomy;
Patients, No.
Included Studies
[Ref]
PLT
MST
WMD
Pooled OR (95% CI)
p Value
6
5
3
3
3
3
6
4
4
194
147
89
89
89
89
162
117
126
196
177
91
91
91
91
190
147
128
þ2.74
þ3.10
þ2.33
þ8.18
–1.28
–3.01
–0.65
–0.76
–0.49
. (–0.92 to þ6.41)
. (–3.05 to þ9.24)
. (–0.69 to þ5.35)
. (1.65 to þ18.01)
. (–2.45 to –0.11)
. (–8.57 to þ2.55)
. (–1.51 to þ0.21)
. (–1.26 to –0.27)
. (–1.08 to þ0.10)
0.14
0.32
0.13
0.10
0.03
0.29
0.14
<0.01
0.10
2
9
61
460
63
470
–2.91
–3.67
. (–7.19 to þ1.36)
. (–16.19 to þ8.85)
0.18
0.57
6
9
9
331
416
416
314
450
450
–1.02
.
.
. (–2.18 to þ0.15)
þ0.94 (þ0.52 to þ1.71)
þ8.26 (þ2.16 to þ31.56)
0.09
0.85
0.002
FEV1 ¼ forced expiratory volume in 1 second;
MST ¼ muscle-sparing thoracotomy;
WMD ¼ weighted mean difference;
VC ¼ vital capacity.
REVIEW
difference in postoperative pain on day 1 between the
PLT and MST group (pooled WMD, –0.65; 95% CI, –1.51
to 0.21; p ¼ 0.14; Table 3).
Four studies reported the pain scores on day 7 after
the procedure [10, 15, 18, 21]. There was statistical heterogeneity between studies (Cochran Q ¼ 9.16, p ¼ 0.03;
I2 ¼ 67.3%, 95% CI, 0% to 86.6%). In the random effects
model, postoperative pain scores were significantly lower
on day 7 in the MST group than in the PLT group (pooled
OR ¼ odds ratio;
PLT ¼
WMD, –0.76; 95% CI, –1.26 to –0.27; p < 0.01; Fig 3 and
Table 3).
Four studies reported the pain scores on day 30 after
the procedure [10, 12, 17, 18]. There was statistical heterogeneity between studies (Cochran Q ¼ 19.38, p < 0.01;
I2 ¼ 84.5%, 95% CI, 48.3% to 92.2%). In the random effects
model, there was no significant difference in postoperative pain scores on day 30 between the PLT and
MST group (pooled WMD, –0.49; 95% CI, –1.08 to 0.10;
p ¼ 0.10; Table 3).
Secondary Outcome
Fig 2. Forest plot for weighted mean difference for the effects on
internal rotation between posterolateral thoracotomy (PLT) and
muscle-sparing thoracotomy (MST) groups 30 days after the
operation. The solid squares denote mean difference, the horizontal
lines represent the 95% confidence intervals (CI), and the diamond
denotes the weighted mean differences. (DL ¼ DerSimion Laird.)
INCISION SIZE. Two trials reported the incision size [10, 17].
There was significant statistical heterogeneity between
the two studies (Cochran Q ¼ 6.30, p ¼ 0.01; I2 ¼ *%, 95%
CI ¼ *% to *%). In the random effect model, there was no
significant difference in the incision size between PLT
and MST group (WMD, –2.91, 95% CI, –7.19 to 1.36; p ¼
0.18; Table 3).
DURATION OF OPERATION. Nine trials reported on duration of
the operation between the two groups [10–15, 18, 19, 22].
There was significant statistical heterogeneity between
the nine studies (Cochran Q ¼ 76.91, p ¼ 0.00; I2 ¼ 89.6%,
95% CI, 82.7% to 92.9%). In the random effect model,
duration of the operation did not differ significantly between the PLT and the MST group (WMD, –3.67, 95% CI,
–16.19 to 8.85; p ¼ 0.57; Table 3).
HOSPITAL STAY. Six trials reported the hospital duration of
stay between the two groups [10, 11, 18–20, 22]. There was
no significant statistical heterogeneity between the six
studies (Cochran Q ¼ 9.77, p ¼ 0.08; I2 ¼ 48.8%, 95% CI,
0% to 77.9%). In the random effect model, the hospital
duration of stay did not differ significantly between the
Fig 3. Forest plot for weighted mean difference postoperative pain
scores between posterolateral thoracotomy (PLT) and muscle-sparing
thoracotomy (MST) groups on postoperative day 7. The solid squares
denote mean difference, the horizontal lines represent the 95%
confidence intervals (CI), and the diamond denotes the weighted
mean differences. (DL ¼ DerSimion Laird.)
PLT and the MST group (WMD, –1.02; 95% CI, –2.18 to
0.15; p ¼ 0.09; Table 3).
TOTAL COMPLICATIONS. Nine trials reported total complications between the two groups [10, 12, 14, 15, 18–22]. These
included bleeding, infection, atelectasis, arrhythmia, prolonged air leak, seroma and thromboembolic complications. There were 102 complications in the 571 patients
(17.9%) in the MST group. On the other hand, there were 96
complications in the 512 patients (18.8%) in the PLT group.
There was no statistical heterogeneity among the studies
(Cochran Q ¼ 13.62, p ¼ 0.06; I2 ¼ 41.3%, 95% CI, 0% to
75.3%). In the random effect model, the total complications
rate did not differ significantly between the PLT and the
MST groups (pooled OR, 0.94; 95% CI, 0.52 to 1.71; p ¼ 0.85;
Table 3). Pulmonary complications consisted of prolonged
air leak in 24 patients in the PLT group vs 19 in the MST
group, atelectasis in 13 patients in the PLT group vs 10 in
the MST group, pneumonia in 14 patients in the PLT group
vs 7 in the MST group, and finally, pneumothorax in 3
patients in the PLT group vs 1 in the MST group.
SEROMA. Nine trials reported seroma between the two
groups [10, 12, 14, 15, 18–22]. Seroma was diagnosed in 23
of 571 patients (4.0%) in the MST group; however, no
seroma was reported in the PLT group. Four studies
documented no seroma in either group. These studies
appear to be excluded on the Forest plot (upper confidence limit is infinity; Fig 4). However, this was not the
case because the StatsDirect software allows for a
correction term by adding 0.5 to all cells that are 0 (Haldane method). There was no statistical heterogeneity
between studies (Cochran Q ¼ 0.68, p ¼ 0.95; I2 ¼ 0%, 95%
CI, 0% to 64.1%). In the random effect model, the risk of
seroma was significantly higher in the MST group than in
the PLT group (pooled OR, 8.26; 95% CI, 2.16 to 31.56;
p < 0.01; Fig 4; Table 3).
REVIEW
UZZAMAN ET AL
META-ANALYSIS OF THORACOTOMY TECHNIQUES
1099
Fig 4. Forest plot for pooled odds ratio of reported case of seroma after
posterolateral thoracotomy (PLT) and muscle-sparing thoracotomy
(MST). The solid squares denote odds ratios, the horizontal lines
represent the 95% confidence intervals, and the diamond denotes the
pooled odds ratio. (DL ¼ DerSimion Laird.)
Comment
A traditional PLT, which provides excellent exposure of
the lung, pulmonary hilum, and mediastinum, has been
the standard incision for pulmonary procedures for the
past 100 years. However, disadvantages of this approach
include the division of the major muscles of the chest,
resulting in increased potential for blood loss, a moderate
time requirement for opening and closing the incision,
prolonged ipsilateral shoulder and arm dysfunction,
scoliosis, compromised pulmonary function, and chronic
postthoracotomy pain syndromes [1–3, 14].
As a result, there has been a search to identify the ideal
incision in open thoracic operations. This search has
been greatly facilitated with the advent of single-lung
ventilation and the liberal use of stapling instruments
that allow safe and precise resection through a less
extensive thoracic opening [21, 22]. Vertical axillary thoracotomy was first reported by Browne in 1953 [4] but
further developed in adult practice by Baeza and Foster
[5]. Mitchell and colleagues [7] proposed a lateral MST,
whereas Bethencourt and Holmes [8] were able to perform
a MST with a conventional posterior thoracotomy
approach. Horowitz and colleagues [6] described a technique involving an incision through the “auscultatory
triangle.” Common to these incisions is preservation
of the integrity of the LD and preservation of the SA
or division of this muscle in line with the direction of
its fibers. Of the described muscle-sparing incisions,
in only the technique described by Horowitz and
colleagues—through the auscultatory triangle—are no
muscles divided [6].
In other described muscle-sparing incisions, when a
lateral incision is used, slips of SA usually need to be
elevated from the ribs, and the anterolateral and vertical
axillary incisions both require splitting of the SA in line
REVIEW
Ann Thorac Surg
2014;97:1093–102
1100
REVIEW
UZZAMAN ET AL
META-ANALYSIS OF THORACOTOMY TECHNIQUES
REVIEW
with its fibers [4, 5, 7, 8]. More recently, there have been
modifications of these techniques, including anteroaxillary and PLT with LD and SA detachment from its
origin [22, 25, 26]. In general, it has been advocated that
these MST approaches reduce postoperative pain, preserve pulmonary function, and lessen postoperative
complications. Another potential advantage is the availability of extrathoracic musculature, such as LD and SA,
for rotational flap control of postresectional space problems, empyema, or bronchopleural fistulas [14].
This meta-analysis shows that the pain scores are lower
with MST than with PLT. However, the pain scores are
only significantly lower on postoperative day 7. On day 1
and day 30, the reduction in pain was not deemed significant. This meta-analysis also showed that MST
significantly improves internal rotation of the shoulder on
day 30. However, there was no improvement in lung
function (FEV1 or VC) on day 30. The secondary outcomes
analysis showed an increased rate of seromas with the
MST technique compared with PLT, although there was
no difference in the total reported complications between
the groups. There was also no significant difference
between the incision sizes, duration of the operation, and
hospital length of stay.
Various factors contribute to postoperative pain after
thoracic operations, including positioning during the
procedure, crush injuries or muscle ischemia, injuries to
intercostal nerves and vessels, especially during retraction, and rib fractures [27]. Pain also results from trauma
to paraspinal ligaments and to the muscles and small
joints of the back [28]. Furthermore, anesthesia, analgesic
requirements, and patient-related factors such as age,
sex, complications, or sociocultural aspects also all affect
pain [29].
This meta-analysis showed that there was no significant improvement with the MST technique in reducing
pain scores on day 1. One possible factor underlying our
finding may be due to the contribution to postthoracotomy pain from incision of the skin and pleura
and retraction of muscles and intercostal nerves. Both
incisions elicit profound pain of both somatic and visceral
origin. However, the difference in pain from muscle division alone, which is higher in the PLT group, may not
be severe enough to impart a difference in outcome when
measured against the degree of rib spreading, which is
greater in MST to compensate for the diminished field of
view. Furthermore, the incision made during certain MST
approaches (eg, vertical and axillary) crosses at least 2
dermatomes, whereas the incision made during PLT runs
along only 1 dermatome, and this may play a role in the
early pain experienced by the patients [13]. Another
explanation for why there appears to be no major difference in pain scores on day 1 may be that aggressive
postoperative epidural analgesia, used in several of the
studies, specifically targets the tissue disruption that
might lead to greater pain in the PLT group [13]. Factors
such as rib retraction, presence of chest drains, and
epidural analgesia are less of an issue by day 7, which
possibly accounts for the significant difference in pain
scores observed between the groups at this stage.
Ann Thorac Surg
2014;97:1093–102
Postthoracotomy pain can present for several weeks
and months and represents a serious problem for some
patients. Chronic pain, in particular, may have a major
effect on the quality of life and is associated with significant health care costs. Neurophysiologic assessments
have shown that intercostal nerve function is more
commonly impaired 1 month after the operation with the
PLT incision than with the MST incision [17]. However,
our meta-analyses have shown no significant difference
in postthoracotomy pain scores between the groups by
day 30. Our failure to appreciate a small difference with
respect to incision type may likely be the result of the
relatively low rate of postthoracotomy pain on day 30,
rendering it difficult to appreciate small differences
between the groups.
This meta-analysis showed no significant difference in
the change in postoperative FEV1 and VC between the
two groups after 30 days. PLT and MST both involve
dividing the intercostal muscles within a given interspace,
impaired lung compliance, and loss of volume due to
lung resection. As a result, both groups will have a degree
of postoperative respiratory dysfunction. However, the
main difference between the two groups lies in the fate of
the LD and SA muscles, both of which are weak accessory
muscles of respiration. As a result, the lack of a significant
difference in lung dysfunction between the groups on day
30 is not surprising. Ponn and colleagues [20] showed
limited muscle sparing incisions may result in better
long-term pulmonary function, but the differences were
small and of no apparent clinical advantage in the
average patient. Varela and colleagues [30] reported that
preoperative FEV1, postoperative pain scores, and
epidural analgesia were independently and reliably
associated with the postoperative FEV1 ratio. By day 30,
the difference in pain scores was insignificant in our
meta-analysis, and this is a possible reason why there is
no difference in the degree of respiratory dysfunction.
There may have been a discernible difference in lung
function tests between the two groups on day 7 when
there was a significant difference in pain scores. However, we did not assess the FEV1 and VC status within the
first postoperative week due to the limited number of
studies available.
This meta-analysis shows a significant improvement in
internal rotation with MST compared with PLT 30 days
after the procedure. This is to be expected given the
sparing of the LD muscle, which has attachment to the
humerus. The preservation of LD and SA allows range of
movement and function of the arm to return to normal
more readily. The studies that explored shoulder movement [12, 18, 21] specifically used the posterolateral MST
approach. This approach does not risk damaging structures functionally involved, such as the long thoracic
nerve. Although not significant, Landreneau and colleagues [19] showed that shoulder function returned to
97% of preoperative function after MST compared with
92% in the PLT group. Several studies, using a manual
muscle test, have shown that MST results in better postoperative muscle strength of the shoulder girdle [14, 22].
Other studies that used a dynamometer to test isokinetic
muscle strength also showed preserved strength of the
shoulder girdle muscle after MST compared with PLT
[28, 31, 32]. Preservation of muscle movement and
strength by MST may be useful in manual workers,
young adults, and athletes, particularly if the operation
involves the dominant side. It is noteworthy that this is
exactly the group of people in whom MST is technically
more difficult and exposure more limited due to bulky
musculature [6, 31].
Our evaluation of secondary outcomes showed a
significantly increased incidence of seroma in the MST
group. This is more commonly observed in the posterolateral MST where there is extensive mobilization of the
LD leading to the creation of subcutaneous flaps [21, 22].
Five studies reported seroma in the MST groups [14,
18–21]. Four of these studies used the posterolateral
approach [14, 18, 20, 21], which accounted for the higher
rates of seroma seen overall in the MST approach due to
the creation of flaps. The other study [19] used an anterior
lateral MST approach. Various authors have tried to
reduce the rates of seroma by modifying their technique
such as the anteroaxillary thoracotomy technique [15, 16].
Despite the higher rates of seroma, the overall rates of
complications reported, and therefore, hospital stay, were
comparable between the groups. We have found that the
MST does not increase the overall operative time. The
increased time required to create the subcutaneous flaps
and mobilize the LD and or SA are probably made up at
closure, because it is not necessary to reapproximate the
muscle edges.
Our meta-analysis also showed no differences in the
incision size between the groups. The MST provides
acceptable access to the chest cavity for pulmonary
resections when required. In addition, the MST allows
access to the mediastinum and complete lymphadenectomy. Occasionally, its exposure may be difficult in
heavily muscled individuals. In particular, axillary and
auscultatory triangle thoracotomies are limited in
opening width due to the lack of movable ribs and
sometimes cause rib fractures at the opening site.
However, if further visualization is required, it may be
converted to the standard opening by simply incising
the thoracic musculature. It was interesting to note that
only one study in this meta-analysis reported rib fractures in both MST and PLT groups [10], whereas no
cases in any study required conversion from MST to
PLT [10–22].
This meta-analysis has inevitable limitations. The heterogeneity of some variables in this study is worthy of
comment. The included trials varied in the inclusion
criteria for the type of patients that they studied. For
example, some only studied patients with lung cancers,
whereas others assessed patients undergoing varied
thoracic procedures. Unfortunately, the insufficient raw
data were available to perform subgroup analyses. In
addition, the types of MST procedures performed varied.
We tried to counter these heterogeneities by using the
random effects model.
In addition, each individual trial had a different followup protocol and used different methods to evaluate some
REVIEW
UZZAMAN ET AL
META-ANALYSIS OF THORACOTOMY TECHNIQUES
1101
of the outcome measures. There were also disparities in
the quality of individual trials. Only three of the studies
reported an intention-to-treat analysis, and this could
lead to attrition bias in the other studies. None of the
studies had a clear allocation concealment trail, and
therefore, all were susceptible to selection bias. In addition, the trials included in our analysis spanned 20 years,
a period in which surgical and anesthetic techniques
changed dramatically.
One thing we have learned with VATS is to compare
approaches according to the rib-spreading process. One
limitation of this meta-analysis is that the extent of rib
spreading is not known. One study by Nosotti and colleagues [10] explored the extent of rib spreading between
the groups but the results were not specified. Likewise,
data on concomitant rib shearing or resection (indeed a
possibility for both muscle-sparing and posterolateral
techniques) are missing and may be important for pain
analysis.
Finally, this review has demonstrated that only a small
number of trials have studied this subject. Conclusions
drawn from a small number of studies must be interpreted with caution, and clearly, further research in this
area is warranted.
In conclusion, this meta-analysis demonstrates that the
MST technique is associated with significant improvement in shoulder movement (internal rotation) and
intermediate pain scores (day 7). There is negligible
improvement in lung function tests on day 30. Despite the
significantly higher rates of seroma, the overall complication rates are comparable between the two groups. We
advocate the MST approach in most cases, particularly on
the physically active patients who are dependent on a
quicker recovery of shoulder function.
We acknowledge the statistical analysis contribution from senior
statistician, Peter Browne, from the institution of Medical
Statistics, Queen Mary’s University, London, UK.
References
1. Bernard A, Brondel L, Arnal E, Favre JP. Evaluation of respiratory muscle strength by randomized controlled trial
comparing thoracoscopy, transaxillary thoracotomy, and
posterolateral thoracotomy for lung biopsy. Eur J Cardiothorac Surg 2006;29:596–600.
2. Karmakar MK, Ho AM. Postthoracotomy pain syndrome.
Thorac Surg Clin 2004;14:345–52.
3. Kucukarslan N, Kirilmaz A, Arslan Y, Sanioglu Y, Ozal E,
Tatar H. Muscle sparing thoracotomy in pediatric age: a
comparative study with standard posterolateral thoracotomy. Pediatr Surg Int 2006;22:779–80.
4. Singh SJ, Kapila L. Denis Browne’s thoracotomy revised.
Pediatr Surg Int 2002;18:90–2.
5. Baeza OR, Foster ED. Vertical axillary thoracotomy: a functional and cosmetically appealing incision. Ann Thorac Surg
1976;22:287–8.
6. Horowitz MD, Ancalmo N, Ochsner JL. Thoracotomy
through the auscultatory triangle. Ann Thorac Surg 1989;47:
782–3.
7. Mitchell R, Angell W, Wuerflein R, Dor V. Simplified lateral
chest incision for most thoracotomies other than sternotomy.
Ann Thorac Surg 1976;22:284–6.
REVIEW
Ann Thorac Surg
2014;97:1093–102
1102
REVIEW
UZZAMAN ET AL
META-ANALYSIS OF THORACOTOMY TECHNIQUES
REVIEW
8. Bethencourt DM, Holmes EC. Muscle-sparing posterolateral
thoracotomy. Ann Thorac Surg 1988;45:337–9.
9. Landreneau RJ, Mack MJ, Hazelrigg SR, et al. Videoassisted thoracic surgery: basic technical concepts and
intercostal approach strategies. Ann Thorac Surg 1992;54:
800–7.
10. Nosotti M, Baisi A, Mendogni P, Palleschi A, Tosi D, Rosso L.
Muscle sparing versus posterolateral thoracotomy for pulmonary lobectomy: randomised controlled trial. Interact
Cardiovasc Thorac Surg 2010;11:415–9.
11. Endoh H, Tanaka S, Yajima T, Ito T, Tajima K, Mogi A, et al.
Pulmonary function after pulmonary resection by posterior
thoracotomy, anterior thoracotomy or video-assisted surgery. Eur J Cardiothorac Surg 2010;37:1209–14.
12. Athanassiadi K, Kakaris S, Theakos N, Skottis I. Musclesparing versus posterolateral thoracotomy: a prospective
study. Eur J Cardiothorac Surg 2007;31:496–9.
13. Ochroch EA, Gottschalk A, Augoustides JG, Aukburg SJ,
Kaiser LR, Shrager JB. Pain and physical function are similar
following axillary, muscle-sparing vs posterolateral thoracotomy. Chest 2005;128:2664–70.
14. Akçali Y, Demir H, Tezcan B. The effect of standard
posterolateral versus muscle-sparing thoracotomy on multiple parameters. Ann Thorac Surg 2003;76:1050–4.
15. Nomori H, Horio H, Suemasu K. Anterior limited thoracotomy with intrathoracic illumination for lung cancer: its advantages over anteroaxillary and posterolateral thoracotomy.
Chest 1999;115:874–80.
16. Nomori H, Horio H, Fuyuno G, Kobayashi R. Non-serratussparing antero-axillary thoracotomy with disconnection of
anterior rib cartilage. Improvement in postoperative pulmonary function and pain in comparison to posterolateral
thoracotomy. Chest 1997;111:572–6.
17. Benedetti F, Vighetti S, Ricco C, et al. Neurophysiologic
assessment of nerve impairment in posterolateral and
muscle-sparing thoracotomy. J Thorac Cardiovasc Surg
1998;115:841–7.
18. Sugi K, Nawata S, Kaneda Y, Nawata K, Ueda K, Esato K.
Disadvantages of muscle-sparing thoracotomy in patients
with lung cancer. World J Surg 1996;20:551–5.
19. Landreneau RJ, Pigula F, Luketich JD, et al. Acute and
chronic morbidity differences between muscle-sparing and
standard lateral thoracotomies. J Thorac Cardiovasc Surg
1996;112:1346–50.
Ann Thorac Surg
2014;97:1093–102
20. Ponn RB, Ferneini A, D’Agostino RS, Toole AL, Stern H.
Comparison of late pulmonary function after posterolateral
and muscle-sparing thoracotomy. Ann Thorac Surg 1992;53:
675–9.
21. Hazelrigg SR, Landreneau RJ, Boley TM, et al. The effect of
muscle-sparing versus standard posterolateral thoracotomy
on pulmonary function, muscle strength, and postoperative
pain. J Thorac Cardiovasc Surg 1991;101:394–400.
22. Lemmer JH Jr, Gomez MN, Symreng T, Ross AF, Rossi NP.
Limited lateral thoracotomy. Improved postoperative pulmonary function. Arch Surg 1990;125:873–7.
23. 2000 Guidelines of Meta-analysis Of Observational Studies
In Epidemiology. Available at: www.consort-statement.org/?
o¼1347. Accessed June 5, 2012.
24. The Newcastle-Ottawa Scale for assessing quality of nonrandomised studies in meta-analyses. Available at: http://
www.ohri.ca/programs/clinical_epidemiology/oxford.asp.
Accessed June 5, 2012.
25. Fraund S, B€
oning A, Scheewe J, Cremer JT. Antero-axillary
access for hypoplastic aortic arch repair. Ann Thorac Surg
2002;73:278–88.
26. Loscertales J, Congregado M, Moreno S, JimenezMerchan R. Posterolateral thoracotomy without muscle division: a new approach to complex procedures. Interact
Cardiovasc Thorac Surg 2012;14:2–4.
27. Khelemsky Y, Noto CJ. Preventing post-thoracotomy pain
syndrome. Mt Sinai J Med 2012;79:133–9.
28. Khan IH, McManus KG, McCraith A, McGuigan JA. Muscle
sparing thoracotomy: a biomechanical analysis confirms
preservation of muscle strength but no improvement in
wound discomfort. Eur J Cardiothorac Surg 2000;18:656–61.
29. Wenk M, Schug SA. Perioperative pain management after
thoracotomy. Curr Opin Anaesthesiol 2011;24:8–12.
30. Varela G, Brunelli A, Rocco G, et al. Predicted versus observed
FEV1 in the immediate postoperative period after pulmonary
lobectomy. Eur J Cardiothorac Surg 2006;30:644–8.
31. Kutlu CA, Akin H, Olcmen A, Biliciler U, Kayserilioglu A,
Olcmen M. Shoulder-girdle strength after standard and
lateral muscle-sparing thoracotomy. Thorac Cardiovasc Surg
2001;49:112–4.
32. Ziyade S, Baskent A, Tanju S, Toker A, Dilege S. Isokinetic
muscle strength after thoracotomy: standard vs. musclesparing posterolateral thoracotomy. Thorac Cardiovasc
Surg 2010;58:295–8.