Ultrasound in Med. & Biol., Vol. 35, No. 9, pp. 1411–1420, 2009
Copyright Ó 2009 World Federation for Ultrasound in Medicine & Biology
Printed in the USA. All rights reserved
0301-5629/09/$–see front matter
doi:10.1016/j.ultrasmedbio.2009.03.018
d
Original Contribution
PREOPERATIVE MEASUREMENT OF THICKNESS OF CUTANEOUS MELANOMA
USING HIGH-RESOLUTION 20 MHZ ULTRASOUND IMAGING: A MONOCENTER
PROSPECTIVE STUDY AND SYSTEMATIC REVIEW OF THE LITERATURE
LAURENT MACHET,*yz VÉRONIQUE BELOT,y MICHAEL NAOURI,y MICHEL BOKA,x YOUSSEF MOURTADA,y
BRUNO GIRAUDEAU,k BORIS LAURE,{ ADELINE PERRINAUD,yz MARIE-CHRISTINE MACHET,x
and LOÏC VAILLANT*yz
* Université Francois Rabelais de Tours, UMR INSERM U930 CNRS ERL 3106, France; y Service de Dermatologie, CHRU,
Tours, France; z Réseau Mélanome Grand Ouest, CIC - BIOTHERAPIE INSERM 503, CHRU, Nantes, France; x Service
d’Anatomie Pathologique, CHRU de Tours, France; k INSERM CIC 202, France; Université François Rabelais Tours, France;
CHRU, Tours, France; and { Service de Chirurgie Maxillo-faciale, CHRU, Tours, France
(Received 10 January 2008, revised 6 March 2009, in final form 19 March 2009)
Abstract—Histologic measurement of the thickness of melanoma is a major prognostic factor and governs the size
of the surgical excision (1cm for melanomas less than 1 mm thick, 2 cm for melanomas thicker than 2 mm and 3 cm
beyond 4 mm). To determine whether high-resolution ultrasound can be used to predict surgical margins and, thus,
to operate on patients in a single procedure avoiding further re-excision, we performed a systematic review of
studies published from January 1987 to June 2007 and a prospective study. The systematic review selected 14
studies comparing histologic and ultrasound measurements and showing correlation coefficients generally greater
than 0.9. Data available from 7 of the 14 studies (total 869 patients) showed predictive values of adequate margins
in at least 72% of lesions using preoperative measurement of ultrasound thickness. The prospective study included
31 patients referred with a primary melanoma from March 2005 to March 2007. Ultrasound measurement of
thickness was possible except for thin melanomas (,0.4 mm) in areas with marked photoaging, in the plantar
zone, and in the case of very thick melanomas exceeding the explored depth (7.6 mm). The average thickness
was 1.96 mm measured by ultrasound (SD: 2.15) and 1.95 mm by histology (SD: 2.62) and the Bland and Altman
graph showed moderate agreement between ultrasound and histology. Limits of agreement were estimated at –1.4
and 11.8, corresponding to relative limits of agreement of –40 to 180%. Ultrasound predicted appropriate
margins (1, 2 or 3 cm wide according to sonometric thickness) in 26 of the 31 subjects (84%, 95% CI 66–95). Preoperative high-resolution ultrasound is a noninvasive examination that can help in choosing appropriate surgical
margins and should reduce the need of partial or excisional biopsy before surgery, and the need for further re-excision. (E-mail: [email protected]) Ó 2009 World Federation for Ultrasound in Medicine & Biology.
Key Words: Skin, Melanoma, Ultrasound, Breslow thickness, Pathology, Surgical margins, Agreement.
et al. 2005; Roberts et al. 2002), a margin of 0.5 cm is
necessary for in situ melanomas except on the face and
a margin of 1 cm is necessary for tumors of 1 mm thickness
or less and for in situ melanomas on the face. Although the
impact on overall survival has not been demonstrated,
wider resections are recommended for melanomas thicker
than 1 mm to decrease the risks of local recurrence (Lens
et al. 2002). Thus, the recommended margins are 1 to
2 cm for tumors between 1.01 and 2 mm, 2 cm for tumors
between 2.01 and 4 mm and 3 cm for tumors thicker than
4 mm (Négrier et al. 2005; Roberts et al. 2002). According
to these recommendations, when a pigmented lesion is suspected of being a melanoma on clinical examination, excisional biopsy is usually recommended with a margin of
INTRODUCTION
The maximum thickness of a melanoma is a major factor
prognostic of metastatic dissemination and governs the
size of surgical margins. The National Institutes of Health
(NIH) consensus recommends a 5 mm margin for all in
situ melanomas but this can be inadequate in a large
percentage of cases, especially lesions occurring on the
head and neck or hands and feet (Cook 2003). According
to both French and British Consensus reports (Négrier
Address correspondence to: Laurent Machet, Service de Dermatologie, CHRU, 37044 Tours Cedex 09, France. E-mail: machet@
univ-tours.fr
1411
1412
Ultrasound in Medicine and Biology
2–3 mm. If the histologic examination confirms the diagnosis of melanoma, surgical removal is then carried out
with margins depending on the maximum thickness
measured with a micrometer on the histology slide (Breslow et al. 1970).
Noninvasive preoperative measurement of the
maximum depth of a melanoma would be valuable,
permitting choice of appropriate surgical margins. The
traditional ultrasound equipment used for medical
imaging operating at 7.5 MHz does not allow sufficiently
precise measurement because the resolution is inadequate
(Ulrich et al. 1999). As several studies have previously
shown that ultrasound examination using 20 MHz allowed preoperative assessment of melanoma thickness
that correlated well with histologic measurement (Hoffmann et al. 1992, 1999; Dummer et al. 1995; Tacke
et al. 1995; Lassau et al. 1999; Serrone et al. 2002), our
main aim was to check whether the use of 20 MHz ultrasound imaging made it possible to measure melanoma
thickness in a sufficiently precise way to choose the appropriate surgical margin. We, therefore, carried out
a prospective study comparing histologic and ultrasound
thicknesses in a consecutive series of subjects referred
with primary cutaneous melanoma not previously
removed surgically and we performed a systematic review
of the literature to analyze the causes of error and to determine the impact of preoperative ultrasound imaging on the
surgical management of melanoma.
METHODS
Subject selection criteria
From March 2005 to March 2007, all the patients
referred to our department for suspected cutaneous melanoma that had not been removed surgically were offered
cutaneous ultrasound imaging before surgical removal.
The study was approved by the local ethics committee.
The inclusion criteria included suspected primary
cutaneous melanoma following clinical and dermatoscopic examination or proven melanoma following partial
biopsy providing the pathology diagnosis of in situ or
invasive cutaneous melanoma also allowing ultrasound
imaging of the remaining lesion. All histologic subtypes
of melanoma were eligible for inclusion, including those
that were ulcerated or regressive. Subjects with lesions
whose histologic examination later showed that it was
not a primary cutaneous melanoma or cutaneous metastasis of melanoma were excluded, as were patients
referred after having melanomas completely removed by
excisional biopsy, mucosal melanoma and patients with
pigmented lesions which were not clinically suggestive
of melanoma.
Volume 35, Number 9, 2009
Ultrasound imaging
Real-time high-resolution 20 MHz ultrasound
imaging equipment (Dermcup 2020; Atys Médica, Soucieu en Jarrest, France) was used with an axial resolution
of 80 mm and lateral resolution of 250 mm and maximum
measurable depth of 7.6 mm (Berson et al. 1999; Machet
et al. 2006).
A standard echographic gel was used as coupling
agent between the skin surface and the probe. Minimal
pressure was applied to preserve the thickness and echogenicity of the lesion. The linear probe was held manually,
maintained perpendicular to the skin surface and moved
over the skin surface to provide acquisition of 10 images
per second and screening of the entire lesion. Four examiners participated in the study; two of them had little experience (2 days) of ultrasound examination. Examinations
were performed by a single examiner for any given lesion,
which was scanned in at least two perpendicular directions
(transverse and sagittal). We compared the internal
echogenicity of all lesions using standard ultrasound
terminologies and criteria for isoechoic, hypoechoic or
hyperechoic. We also characterized the pattern of internal
echoes using standard ultrasound terminology and criteria
as homogeneous and heterogenous. The limits of the
lesion and the demarcation from normal skin were studied.
It was possible both to store a sequence of images, and to
visualize them on the screen. Second, the thickness was
measured on a vertical axis perpendicular to the surface,
from the skin surface to the deepest point of the lesion,
using an electronic calliper. At least three sonometric
thicknesses were measured for each lesion and the greatest
depth was used for comparison with the greatest depth
measured on the histology slide (Breslow thickness). No
intra- or interobserver variability study was carried out.
Excision and pathology study
The lesions were removed either with a margin according to the sonographic thickness when the clinical
diagnosis was unambiguous or partial biopsy had
confirmed the diagnosis or with a few millimeters margin
when the diagnosis was not certain or when the surgeon
did not want to remove with wide margins without histologic confirmation of melanoma (Fig. 1). Then, excision
lesions were fixed in formaldehyde and processed
routinely. Immunohistochemical study was performed
with PS-100 and HMB-45 (Dako France SAS, Trappes,
France) when the diagnosis of melanoma needed to be
confirmed with PS-100 or HMB-45 staining or to better
appreciate dermal invasion in melanoma arising on melanocytic nevus. The histology report was standardized and
included Breslow thickness, Clark level, ulceration,
regression, peritumoral infiltrate, vascular invasion and
rate of mitosis. As superficial dermal elastosis is associated with a superficial hypoechoic band localized in the
Preoperative measurement of thickness of cutaneous melanoma d L. MACHET et al.
1413
Fig. 1. Synopsis of study. Sonographic thickness was measured in all patients and surgery was planned as a single intervention without biopsy in 15 cases and with partial biopsy in nine cases. Two patients with metastatic disease had limited
resection. Five patients had a few millimeters in sano resection and further re-excision. Excessive means that in two
patients planned margins were overestimated from sonographic thickness but were correct on Breslow thickness because
of a large in situ lateral component in facial melanoma. Insufficient * means that sonographic thickness was concordant
with Breslow thickness but margins were insufficient because of a wide in situ lateral component in vulvar melanoma.
Finally, 20/24 patients (83%, 95% CI 63–95) were operated on in single interventions with appropriate margins and appropriate margins were predicted by ultrasound thickness for 26/31 patients (84%, 95% CI 66–95).
upper dermis, the presence of elastosis was noted as
absent, moderate or marked. Breslow thickness was
measured at initial diagnosis by a pathologist who was
blinded to the ultrasound thickness measurement. This
measurement was performed with a micrometer on
a histology slide, with the vertical axis perpendicular to
the skin surface. The maximum value indicates the Breslow thickness (Breslow 1970). All pathology slides
were re-examined by a pathologist who was blinded to
the previously measured Breslow thickness and the
sonometric thickness. In the case of marked discordance
(.1 mm) between histometric and sonometric measurements, new slides were requested by the dermatologist.
This second series of measurements was used for the interobserver histometric agreement study.
sonographic and histometric measurements was assessed
by fitting the Bland and Altman graph (Bland and Altman
1986). Such a graph plots individual differences between
two measurements vs. individual mean of these two
measurements. It allows assessing whether a systematic
bias exist, whether measurement errors increases with
means (which then suggests applying a log-transformation
of data prior analysing them) and which finally allows to
assess the ‘‘limits of agreements’’ which have to be
compared with clinically important difference (i.e., in
case limits of agreement are smaller than clinically important differences, the two methods can be used interchangeably). We also estimated a weighted Kappa coefficient.
Agreement between both histometric measurements was
also assessed by the Bland and Altman graph.
Statistics
Thickness measurements of each lesion were
compared on the basis of the Breslow thickness values
of the initial pathology report. Agreement between
Systematic review
A systematic review was carried out on PubMed
from January 1987 to 30 June 2007 to allow comparison
with previous studies using the following key words:
1414
Ultrasound in Medicine and Biology
Volume 35, Number 9, 2009
Fig. 2. Ultrasound imaging of two melanoma lesions. Lesions were generally hypoechoic and well demarcated from the
dermis.
melanoma, ultrasound, preoperative and thickness in the
last 20 years. Studies mentioning at least the correlation
coefficient and using high-frequency (20 MHz and above)
were selected. From these studies, we calculated the
numbers of subjects for whom surgical margins had
been adequately predicted by preoperative ultrasound
imaging, when available.
RESULTS
Subjects
Two hundred seven new patients with melanoma
were referred in this 2-year period. However 176 lesions
had been excised before referral. Thirty-one lesions
from 31 subjects (10 women, 21 men, mean age 64 years,
range 35–85 years) were included. Thirteen lesions were
on the trunk, two on the upper limbs, eight on the lower
limbs, seven on the face and one in the genital area.
Two lesions from two subjects were excluded, corresponding to dysplastic nevus after histologic examination.
Ultrasound imaging
All 31 lesions were examined and measured by one
of the four examiners. Twenty-eight melanomas were
visualized with ultrasound; all were hypoechoic with
a homogeneous structure overall and generally welldelimited from normal skin (Fig. 2). However, we were
not able to characterize the lesion from surrounding
normal skin in three cases and, in two cases, the maximum
depth of the lesion was beyond the maximum depth
explored by the device (7.6 mm).
Pathology
The characteristics of histologic types of melanoma
and locations are shown in Table 1. Histologic thickness
measurements generally showed good interobserver agreement for all the 31 lesions except for a lesion .10 cm wide
on the arm with an initial Breslow thickness measurement
of 3.8 mm and remeasurement of 6.5 mm on new slides,
making a difference of 2.7 mm. For the other lesions, the
differences ranged from –0.34 to 10.20 mm. The mean
Table 1. Main characteristics of population
Subjects
Age (years)
Site
Histologic subtypes
Histologic features
Breslow thickness (mm)
Ultrasound thickness (mm)
N 5 31
Female10
Male 21
Mean 64
Trunk
Limbs
Face
Genitalia
SSM
LMM
ALM
NMM
Unclassified
Inflammatory infiltrate
Elastosis
None
Mean 1.96
Mean 1.95
Range 35–85
13
10
7
1
16
7
4
1
3
17
7
7
SD 2.15
SD 2.62
Mean + 1.96SD
0.2
0.1
0
Mean
-0.1
-0.2
Mean - 1.96SD
-0.3
-0.4
0
2
4
6
8
10
12
Mean (mm)
Fig. 3. Interobserver agreement for measurements of Breslow
thickness. Figure 3 was fitted after removing a discordant case,
the mean difference between histometric thicknesses was
–0.03 mm and the limits of agreement were estimated at –0.30
and 10.23 mm
difference between histometric thicknesses was –0.11 mm
and the limits of agreement were estimated at –1.09 and
10.85 mm. After removing the discordant case reexamined on new slides, the mean difference between
histometric thicknesses was –0.03 mm and the limits of
agreement were estimated at –0.30 and 10.23 mm (Fig. 3).
Ultrasound thickness vs. histologic thickness
In four cases, the measurements were identical.
Ultrasound over-estimated melanoma thickness in 17
cases and under-estimated it in 10 cases. The mean difference between ultrasound thickness and histologic thickness was 10.01 mm and the limits of agreement were
estimated at –1.85 and 11.87mm. After removing the
two subjects with melanomas thicker than 7.6 mm, resulting in a systematic underestimation of thickness measured
with ultrasound, the mean difference between ultrasound
and histometric measurements was 10.16 mm and the
limits of agreement were estimated at –1.36 and 1 1.67
mm. In 18 cases, the absolute difference was lower than
0.31 mm. The Bland and Altman graph was fitted for
raw data (i.e., without any transformation) (Fig. 4).
Because there appears to have a relationship between
measurement difference and measurement mean (i.e.,
the measurement difference increases as the mean
increases), a log-transformation was applied before replotting the Bland and Altman graph, as usually done
(Bland and Altman 1996) (Fig. 5). The latter graph
showed no systematic bias between the two methods.
Limits of agreement were estimated at 21.4 and 11.8,
which corresponds to relative limits of agreement of
–40% to 180%.
Analysis of main differences between sonographic and
histometric measurements in our series
When marked difference between ultrasound and
histologic thickness was evidenced, the histologic slides
were reviewed with the pathologist in order to understand
2.5
2
1.5
1
0.5
0
-0.5
-1
-1.5
-2
-2.5
-3
1415
Mean + 1.96SD
Mean
Mean - 1.96SD
0
2
4
6
8
10
12
Mean (mm)
Fig. 4. Bland and Altman graph showing differences between
sonometric and histometric measurements. The mean difference
is very close to zero but the standard deviation is fairly high.
the discrepancies. Two very thick melanomas (Breslow
thickness 8.7 and 10 mm) were beyond the maximum
depth explored by the device (7.6 mm). In another case
located on the elbow, the ultrasound measurement was
4 mm. On histologic examination, the lesion was ulcerated
and displayed marked signs of regression, with inflammatory infiltrate and fibrosis, and the Breslow thickness was
6.2 mm. One 10 cm-wide melanoma measured 5.6 mm
on ultrasound compared with 3.8 mm for Breslow thickness but the histologic measurement on the second examination revealed a Breslow thickness of 6.5 mm.
Two thin facial melanomas with Breslow thickness
of 0.4 and 1mm measured 1.7 and 2.7 mm on ultrasound
(difference 1.3 and 1.7), respectively. They both displayed
marked dermal elastosis and marked inflammatory infiltrate was present in one. An in situ melanoma located in
the plantar area was erroneously measured at 1.55 mm,
partly because the thick stratum corneum resulted in
a very large entry echo and a thick hypoechoic band
seen beneath the stratum corneum and partly because
there was marked dermal fibrosis with an inflammatory
infiltrate.
Difference between sonometric and
histometric thickness
measurements(mm)
Differencein Breslow
thickness (mm)
0.3
Difference between sonometric
and Breslow thicknesses (mm)
Preoperative measurement of thickness of cutaneous melanoma d L. MACHET et al.
3
2
Mean + 1.96SD
1
Mean
0
-1
Mean - 1.96SD
-2
-3
-2
0
2
4
6
8
10
Mean (ln(mm))
Fig. 5. Bland and Altman graph using logarithmic transformation of sonometric and histometric measurements.
1416
Ultrasound in Medicine and Biology
Volume 35, Number 9, 2009
Table 2. Agreement between sonographic and histometric measurements
Histometric measurements
Sonographic measurements
in situ
0–1 mm
1.01–2 mm
2.01–4mm
.4 mm
Total
in situ
0
1
1
0
0
2
0–1 mm
2
13
2
1
0
18
1.01–2 mm
0
0
3
2
0
5
2.01–4 mm
0
0
0
1
0
1
.4 mm
0
0
0
1
4
5
Total
2
14
6
5
4
31
The well-classified values appear in bold type: 21/31 (68%) subjects were appropriately classified. According to the width of surgical excision, 23
subjects (74%) were well classified because in situ facial melanomas require a 1 cm margin. Two subjects were considered as misclassified but were
exactly at the group borderlines: two subjects with sonographic thickness of 2 mm and Breslow thickness of 2.3 and 2.5 mm.
Peritumoral reactions such as inflammatory infiltrate,
fibrosis, neovascularization and elastosis in the
surrounding dermis are believed to overestimate thickness
when measured with ultrasound since all of these dermal
modifications appear as hypoechoic. However, after
removing the two lesions thicker than 7.6 mm, we
compared the mean difference between ultrasound and
histometric measurements of lesions with marked inflammatory infiltrate, regression, fibrosis or dermal elastosis
with lesions, which displayed none of these histologic
features. The mean difference was 1 0.18 mm SD 0.87
(n 5 23) vs. 1 0.10 mm SD 0.23 (n 5 6).
Recent cicatricial areas, which were hypoechoic,
were not easily distinguishable from the surrounding
melanoma. Partial biopsy was performed before ultrasound examination in six cases, Breslow thickness being
measured in four cases. The examiner who performed
the ultrasound examination was not blinded to the Breslow thicknesses in two cases. Agreement between ultrasound and histometric measurements was good in five
cases (differences ranging from –0.21 to 10.25) and
poor (11.7) in one case.
Surgical margins
Ultrasound measurement alone was able to predict
appropriate surgical margins in 23/31 melanoma lesions
[74%, 95% CI 55–88] (Table 2). The weighted Kappa
score was estimated at 0.71 [95% CI 0.54–0.88]. Surgical
excision was planned according to ultrasound thickness in
29 cases (six of which had had partial biopsy before ultrasound and three after), the two remaining lesions were
intentionally removed with insufficient margins despite
the ultrasound thicknesses because of multiple locoregional metastases in one subject and visceral metastases
in the other. Of the 29 remaining lesions, 24 were removed
surgically according to sonographic thickness and three
with margins greater than might have been necessary
because of overestimation of sonographic thickness.
However, in two of these three subjects, this was not detrimental since the margins proved finally to be appropriate to
the in situ lateral component of the tumor after histologic
examination. Five lesions were removed with margins of
a few millimeters and then proposed for re-excision with
appropriate margins. One subject with a Breslow thickness
of 0.2 mm refused. Four of these five subjects could have
been operated on with appropriate margins according to
preoperative sonographic measurements.
Systematic review
The literature review identified 18 studies focusing
on correlation between sonometric and histometric thickness in melanoma subjects from 1987 to June 2007 using
20 MHz ultrasound apparatus. All studies selected reported a linear correlation between sonographic and histometric measurements (Table 3). Correlation coefficients
ranged from 0.88 to 0.97 (median 0.95, mean 0.94, SD
0.02). To allow comparison of our study with the literature
findings, the thickness measured by ultrasound was
plotted in relation to Breslow thickness (Fig. 6) and the
Pearson correlation coefficient was calculated (0.94),
although we acknowledge that the use of this coefficient
is in fact inappropriate to demonstrate agreement of
measurements with two methods (Lee et al. 1989;
Kraemer and Feinstein 1981). Two studies were not
selected because they reported measurements of melanoma and nevus thickness for the whole group (Hoffman
et al. 1992; Gambichler et al. 2007).
From seven of these studies with a total of 1611
melanoma lesions, it was possible to determine the
number of lesions that were adequately classified according to histologic subgroups (Table 4).
DISCUSSION
The systematic review showed a strong correlation
between sonometric and histometric thickness of melanoma lesions in the 14 studies selected. However, the methodology used was based on the calculation of the
correlation coefficient. As previously reported (Tacke
et al. 1995; Pellacani et al. 2003; Gambichler et al. 2007),
we used the Bland and Altman graph, which is more appropriate to demonstrate agreement of measurements of tumor
thickness with the two techniques. Despite a correlation
Preoperative measurement of thickness of cutaneous melanoma d L. MACHET et al.
1417
Table 3. Results taken from 14 previously published studies correlating high-resolution ultrasound with histometric measurements
First author year
F (MHz)
N
r
Mean difference (mm)
Lassau 2006
Bessoud 2003
Pellacani 2003
Serrone 2002
Lassau 2002
Lassau 1999
Ulrich 1999
Solivetti 1998
Krahn 1998
Partsch 1996
Tacke 1995
Fornage 1993
Hoffmann 1992
Gassenmaier 1990
Present study
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
111
70
88
152
69
27
150
66
39
58
259
4
54
72
31
. 0.93
.0.96
0.89
0.95
.0.96
.0.95
0.94
0.94
0.95
0.92
0.88
0.95
0.94
0.97
0.94
nr
0.06
0.06
0.32
0.06
0.03
nr
0.3
nr
0.47
0.39
nr
0.35
nr
0.01
SD (mm)
nr
nr
0.78
nr
nr
nr
nr
nr
nr
nr
nr
nr
nr
nr
0.95
Predictive value (%)
Nr
Nr
85 (2 groups # and . 1mm)
72
Nr
Nr
76
Nr
84 (2 groups # and . 1mm)
Nr
75
Nr
Nr
88
74
F 5 frequency; N5 number of melanoma lesions; r 5 correlation coefficient; SD 5 standard deviation.
Predictive value refers to the percentage of patients that were appropriately classified according to histologic grouping. Numbers of groups and limits
varied from two to five groups according to studies
coefficient of 0.94 in our series and generally greater than
0.9 in the published studies, marked differences between
the two techniques were evidenced in our series. The design
of our study was very practical, with several examiners,
some of whom were not experienced in ultrasound examination and included subjects with large and thick melanomas. It is, thus, not very surprising that the Bland and
Altman graph showed evidence of greater differences
between sonometric and histometric measurements than
in other studies (Tacke 1995; Pellacani 2003; Gambichler
2007). These differences raised concerns about the accuracy of ultrasound measurement for predicting adequate
surgical margins since limits of agreement were estimated
at –1.4 and 11.8. Nevertheless, in our series it was possible
to predict appropriate margins in 84% of the subjects [95%
CI 66–95] according to the French and British Consensus
reports. This was confirmed by the systematic review
which identified 14 studies, of which seven (total 869
subjects) showed predictive values of adequate margins
in at least 72% of the lesions using preoperative ultrasound
thickness measurement (Table 4).
Breslow thickness (mm)
12
y = 1.15x - 0.30 r = 0.94
10
8
6
4
2
0
0
2
4
6
8
10
Sonographic thickness (mm)
Fig. 6. Linear relationship between sonographic and histometric
measurements.
Causes of errors in measuring thickness
The main aim of the study was to assess the practical
value of preoperative measurement of thickness when
choosing surgical margins. In this consecutive series of
31 subjects, nine of whom had partial biopsy and, after
excluding two subjects with metastases who were operated on with minimal margins, at least 20/24 subjects
who were operated on according to sonographic thickness
had appropriate margins in the initial intervention without
need for re-excision. Of the five lesions excised with
minimal margins, four could have been removed appropriately according to sonographic thickness. However,
overestimation of thickness (average 0.16 mm in this
series after excluding the two melanomas of thickness
.7.6 mm) may have caused problems for the surgeon,
particularly when he had to chose between 1, 2 or 3 cm
margins for lesions with thicknesses measured close to
recommended limits of surgical groups (i.e., close to
1 mm, to 2 mm and to 4 mm). This was not often the
case in our series but may explain some of the ‘‘misclassifications’’. For example, which margin should be chosen
when a melanoma measures 2 mm (two cases) or 4 mm
(one case) on ultrasound? When the location is not difficult, and if the patient agrees, a wider margin is preferable.
In other cases, definite margins can be chosen after limited
resection and measurement of Breslow thickness. The
main problem we encountered was location on plantar
and facial sites. In one case, we were unable to visualize
a thin plantar melanoma correctly with ultrasound (Breslow 0.8 mm), in another case we overestimated an
in situ plantar melanoma, resulting in excessive surgical
margins and finally correctly measured a thick melanoma
(6.4 mm with ultrasound, Breslow thickness 6.6 mm).
Another problem was with lesions on the face. The melanomas were generally thin and associated with actinic
damage resulting in a subepidermal hypoechoic band
1418
Ultrasound in Medicine and Biology
Volume 35, Number 9, 2009
Table 4. Prediction of adequate surgical margins by ultrasound thickness in the literature
Number of lesions well classified by ultrasound
First author
(year of publication)
Gassenmaier (1990)
Tacke (1995)
Subtotal
Dummer (1995)
Ulrich (1999)
Serrone (2002)
Subtotal
Histometric groups (mm)
Number of lesions
73
259
332
108
150
152
410
,0,75
.0,75-1,5
.1,5-4
19
87
106
38
52
43
133
20
53
73
22
35
38
95
24
53
77
21
26
23
70
Number of well
classified lesions
PV %, 95% CI
63
193
256
92
114
110
316
86, 76-93
75, 69-80
77, 74-80
85, 77-91
76, 68-83
72, 65-79
77, 72-81
34
78
112
87, 73-96
89, 80-94
88, 81-93
.4
11
1
6
18
Histometric groups (mm)
,1
Krähn (1998)
Pellacani (2003)
Subtotal
39
88
127
27
51
78
.1
7
27
34
PV: predictive value.
Note that the histometric subgroups were different from actual recommendations (,1 mm, 1.01 to 2 mm, 2.01 to 4 mm, . 4.01 mm).
and thicknesses were mostly overestimated (4/6) with
ultrasound (range 1 0.23 to 11.69). One was underestimated (–0.25) and the other was identical. Serrone et al.
(2002) reported that sonographic measurement was poorly
correlated with histologic measurement for melanomas up
to 0.75 mm in thickness and was responsible for misclassification in 38% of these cases. This is in fact not a real
problem in Europe since the recommended surgical
margin for in situ melanoma of the face is 1 cm, i.e., the
same as for melanomas of Breslow thickness less than
1 mm. However, the recommended margins for in situ
facial melanoma remain 5 mm in the United States,
although some authors have recommended wider margins
because of high rates of incomplete excision and further
recurrence with standard 5 mm margins (Bricca et al.
2005, Jejurikar et al. 2007; McKenna et al. 2006). For
our cases, because of wide in situ lateral extension of
the facial melanomas, overestimation was not detrimental
to subjects and surgical margins were not excessive. This
explains why only 21 lesions were well classified in histometric groups preoperatively (Table 2). If we add the
melanomas of the face that required a 1 cm margin, 23
(74%) lesions were well classified.
Can initial wide excision following sonographic
measurement of thickness affect sentinel lymph node
biopsy (SLNB)?
SLNB is not recommended as standard treatment for
melanoma patients in France or the United Kingdom but is
routinely performed in patients with melanoma thickness
. 1 mm in other countries, including the United States.
The migration of the tracer and identification of the first
node may theoretically be disturbed if a narrow or (even
more so) a wide excision has been performed prior to
SLNB. In a series of 104 patients who underwent wide
excision prior to SLNB, the incidence of positive nodes,
number of positive nodes and the outcome after followup of 51 months were similar to those of 1291 patients
who had SLNB just before wide excision of melanoma
(Gannon et al. 2006). This series included patients with
lesions located on the trunk and near the median line.
Thus, wide excision prior to SLNB does not appear to
be detrimental and can be performed, except in patients
requiring complex flap reconstruction because no study
has demonstrated correct identification of the sentinel
node in this particular setting. Moreover, measuring melanoma thickness with ultrasound may help in planning
SLNB and wide excision in a single operation (Gambichler et al. 2007).
Can ultrasound help to distinguish melanoma
from atypical nevi?
Sonographic measurement of melanoma arising on
a pre-existing nevus, especially when the nevus is thicker
than the melanoma, may overestimate thickness (Hoffmann et al. 1999). This occurred in only one subject in
our series but the nevus was as thick as the melanoma
and there was no disagreement (0.05 mm) between the
measurements. In this case, as in two other excluded
subjects with dysplastic nevus and no melanoma, it was
not possible to differentiate between a benign melanocytic
nevus and melanoma with ultrasound imaging. Although
we agree with some studies, which have demonstrated that
ultrasound imaging may help to differentiate benign pigmented lesions such as seborrheic keratosis pigmented
basal cell carcinoma or dermal nevus from melanoma
Preoperative measurement of thickness of cutaneous melanoma d L. MACHET et al.
(Rallan et al. 2006, 2007), we do not believe from our
experience that 20 MHz ultrasound imaging can yet
help a trained dermatologist to distinguish atypical nevi
from thin melanomas, which is the main problem.
How can diagnostic accuracy be improved?
By increasing transducer frequency and as it
increases axial resolution, technological advances can be
expected to improve the accuracy of sonometric measurement of melanomas thinner than 1 mm. In a recent study
comparing 20 and 100 MHz ultrasound imaging of 37
subjects with 37 melanocytic nevi and 13 melanomas,
Bland and Altman graphs showed greater accuracy of
measurement of sonometric thickness using a 100 MHZ
probe (Gambichler et al. 2007). In this series, 95% limits
of agreement were better even with the 20 MHz probe than
in our own series. However, there were fewer subjects,
and subjects with melanoma larger than 1 cm, thicker
than 1 mm and with histologic signs of ulceration or
regression were excluded, thus, limiting potential errors.
On the other hand, increasing frequency results in
reducing the ability to measure thick melanoma.
Combining ultrasound with Doppler may help to
distinguish melanomas from other pigmented skin lesions.
Abnormal vascularization was evidenced in one third of
melanomas and was absent from all pigmented nevi
(100% specificity, 34% sensitivity , 100% positive predictive value) in a study by Bessoud et al. (2003) . Moreover,
the use of contrast agents may increase the sensitivity of the
method (Lassau et al. 2001) but this has not been specifically demonstrated with primary cutaneous melanomas.
However, this is not routinely available and the clinical
dilemma of distinguishing thin melanomas from atypical
nevi remains a frequent problem. It can be hoped that ultrasound techniques would help in distinguishing benign atypical pigmented lesions from true melanomas (Rallan et al.
2007). The dilemma is currently solved by using dermoscopy (Braun et al. 2005) or histology. Dermoscopy
combined with computer-assisted diagnostic imaging
methods (Perrinaud et al. 2007) or confocal microscopy
is also used in referral centres (Pellacani et al. 2007). These
techniques need experienced examiners and, although they
can estimate the depth of a pigmented lesion, they cannot
really measure them on a vertical axis. However, the combination of videomicroscopy with sonographic measurement
of thickness has been demonstrated to improve the accuracy of predicting appropriate margins in a series of 88
patients with melanoma (Pellacani et al. 2003).
CONCLUSION
High-resolution ultrasound imaging offers the
advantage of real-time examination of the entire lesion
in a few minutes to determine echoic appearance and
1419
maximum depth of a pigmented lesion and to help in the
differential diagnosis between pigmented basal cell carcinoma or seborrheic keratosis and melanoma. However, it
adds little to clinical examination to distinguish an atypical benign pigmented nevus from true melanoma. Ultrasound cannot replace biopsy when it is necessary and we
emphasize the difficulties of measuring melanoma thickness with ultrasound in the plantar area, i.e., the only cause
of excessive margins in our series. However, our prospective study and the systematic review demonstrate that
ultrasound imaging helps in determining surgical margins
without biopsy or diagnostic excision in most cases. We
believe that high-resolution ultrasound imaging is a valuable and noninvasive tool that can be used in preoperative
assessment of cutaneous melanomas, permitting better
choice of surgical margins in a single operation. It can
even be combined with SLNB (Gambichler et al. 2007),
thus, decreasing the number of reinterventions and may,
therefore, prove to be a time- and cost-saving procedure.
Moreover, accurate preoperative measurement of melanoma thickness could make it possible for part of the
primary melanoma not to be routinely processed for
histologic examination but used for research to provide
greater understanding of the mechanisms of metastatic
progression.
REFERENCES
Berson M, Gregoire JM, Gens F, Rateau J, Jamet F, Vaillant L,
Tranquart F, Pourcelot L. High frequency (20 MHz) ultrasonic
devices: Advantages and applications. Eur J Ultrasound 1999;10:
53–63.
Bessoud B, Lassau N, Koscielny S, Longvert C, Avril MF, Duvillard P,
Rouffiac V, Leclere J, Roche A. High-frequency sonography and
color Doppler in the management of pigmented skin lesions. Ultrasound Med Biol 2003;29:875–879.
Bland JM, Altman DG. Statistical methods for assessing agreement
between two methods of clinical measurement. Lancet 1986;1:
307–310.
Bland JM, Altman DG. Measurement error proportional to the mean.
BMJ 1996;313:106.
Braun RP, Rabinovitz HS, Oliviero M, Kopf AW, Saurat JH. Dermoscopy of pigmented skin lesions. J Am Acad Dermatol 2005;52:
109–121.
Breslow A. Thickness, cross-sectional areas and depth of invasion in the
prognosis of cutaneous melanoma. Ann Surg 1970;172:902–908.
Bricca GM, Brodland DJ, Ren D. Cutaneous head and neck melanoma
treated with Mohs micrographic surgery. J Am Acad Dermatol
2005;52:92–100.
Cook J. Surgical margins for resection of primary cutaneous melanoma.
Clin Dermatol 2003;22:228–233.
Dummer W, Blaheta HJ, Bastian BC, Schenk T, Brocker EV, Remy W.
Preoperative characterization of pigmented skin lesions by epiluminescence microscopy and high-frequency ultrasound. Arch Dermatol
1995;131:279–285.
Fornage BD, McGavran MH, Duvic M, Waldron CA. Imaging of the
skin with 20-MHz US. Radiology 1993;189:69–76.
Gambichler T, Moussa G, Bahrenberg K, Vogt M, Ermert H, Weyhe D,
Altmeyer P, Hoffmann K. Preoperative ultrasonic assessment of thin
melanocytic skin lesions using a 100-MHz ultrasound transducer: A
comparative study. Dermatol Surg 2007;33:818–824.
Gannon CJ, Rousseau DL Jr, Ross MI, Johnson MM, Lee JE,
Mansfield PF, Cormier JN, Prieto VG, Gershenwald JE. Accuracy
1420
Ultrasound in Medicine and Biology
of lymphatic mapping and sentinel lymph node biopsy after previous
wide local excision in patients with primary melanoma. Cancer 2006;
107:2647–2652.
Gassenmaier G, Kiesewetter F, Schell H, Zinner M. Wertigkeit der
hochauflösenden sonographie für die bestimmung des vertikalent
beim malignen melanom der haut. Hautarzt 1990;41:360–364.
Hoffmann K, Happe M, Schuller S, Stucker M, Wiesner M, Gottlober P,
Schwarz M, Strahler J, Neubauer H, Jung C, Petereit S, Welzel J,
Brautzsch N, Bohmeyer J, Wohlrab J, Freitag M, Altmeyer P. Stellenwert der 20-MHz-sonographie des malignen melanoms und pigmentierter läsionen in der routinediagnostik. Ultraschall Med 1999;
20:104–109.
Hoffmann K, Jung J, el Gammal S, Altmeyer P. Malignant melanoma in
20-MHz B scan sonography. Dermatology 1992;185:49–55.
Jejurikar SS, Borschel GH, Johnson TM, Lowe L, Brown DL. Immediate, optimal reconstruction of facial lentigo maligna and melanoma
following total peripheral margin control. Plast Reconstr Surg 2007;
120:1249–1255.
Kraemer MS, Feinstein AR. The biostatistics of concordance. Clin Pharmacol Ther 1981;29:111–123.
Krahn G, Gottlober P, Sander C, Peter RU. Dermatoscopy and high
frequency sonography: Two useful noninvasive methods to increase
preoperative diagnostic accuracy in pigmented skin lesions. Pigment
Cell Res 1998;11:151–154.
Lassau N, Koscielny S, Avril MF, Margulis A, Duvillard P, De Baere T,
Roche A, Leclere J. Prognostic value of angiogenesis evaluated with
high-frequency and color Doppler sonography for preoperative
assessment of melanomas. AJR Am J Roentgenol 2002;178:
1547–1551.
Lassau N, Koscielny S, Opolon P, De Baere T, Peronneau P, Leclere J,
Roche A. Evaluation of contrast-enhanced color Doppler ultrasound
for the quantification of angiogenesis in vivo. Invest Radiol 2001;36:
50–55.
Lassau N, Lamuraglia M, Koscielny S, Spatz A, Roche A, Leclere J,
Avril MF. Prognostic value of angiogenesis evaluated with highfrequency and colour Doppler sonography for preoperative assessment of primary cutaneous melanomas: Correlation with recurrence
after a 5-year follow-up period. Cancer Imaging 2006;6:24–29.
Lassau N, Mercier S, Koscielny S, Avril MF, Margulis A, Mamelle G,
Duvillard P, Leclere J. Prognostic value of high-frequency sonography and color Doppler sonography for the preoperative assessment
of melanomas. AJR Am J Roentgenol 1999;172:457–461.
Lee D, Koh D, Ong CN. Statistical evaluation of agreement between two
methods for measuring a quantitative variable. Comput Biol Med
1989;19:61–70.
Lens MB, Dawes M, Goodacre T, Newton-Bishop JA. Excision margins
in the treatment of primary cutaneous melanoma. A systemic review
of randomized controlled trials comparing narrow vs. wide excision.
Arch Surg 2002;137:1101–1105.
Machet L, Ossant F, Bleuzen A, Gregoire JM, Machet MC, Vaillant L.
L’échographie cutanée haute résolution: Utilité pour le diagnostic,
le traitement et la surveillance des maladies dermatologiques. J Radiol 2006;87(12 Pt 2):1946–1961.
Volume 35, Number 9, 2009
McKenna JK, Florell SR, Goldman GD, Bowen GM. Lentigo maligna/
lentigo maligna melanoma: Current state of diagnosis and treatment.
Dermatol Surg 2006;32:493–504.
Negrier S, Saiag P, Guillot B, Verola O, Avril MF, Bailly C, Cupissol D,
Dalac S, Danino A, Dreno B, Grob JJ, Leccia MT, Renaud-Vilmer C,
Bosquet L. National Federation of Cancer Campaign Centers; French
Dermatology Society. Recommandations pour la pratique clinique:
Standards, Options et Recommandations 2005 pour la prise en charge
des patients adultes atteints d’un mélanome cutane M0, rapport
intégral. Fédération nationale des centres de lutte contre le cancer.
Sociéte francaise de dermatologie. Ann Dermatol Venereol 2005;
132(12 Pt 2). 10S3–10S85.
Partsch B, Binder M, Puspok-Schwarz M, Wolff K, Pehamberger H.
Limitations of high frequency ultrasound in determining the invasiveness of cutaneous malignant melanoma. Melanoma Res 1996;
6:395–398.
Pellacani G, Guitera P, Longo C, Avramidis M, Seidenari S, Menzies S.
The impact of in vivo reflectance confocal microscopy for the diagnostic accuracy of melanoma and equivocal melanocytic lesions.
J Invest Dermatol 2007;127:2759–2765.
Pellacani G, Seidenari S. Preoperative melanoma thickness determination by 20-MHz sonography and digital videomicroscopy in combination. Arch Dermatol 2003;139:293–298.
Perrinaud A, Gaide O, French LE, Saurat JH, Marghoob AA, Braun RP.
Can automated dermoscopy image analysis instruments provide
added benefit for the dermatologist? A study comparing the results
of three systems. Br J Dermatol 2007;157:926–933.
Rallan D, Bush NL, Bamber JC, Harland CC. Quantitative discrimination of pigmented lesions using three-dimensional high-resolution
ultrasound reflex transmission imaging. J Invest Dermatol 2007;
127:189–195.
Rallan D, Dickson M, Bush NL, Harland CC, Mortimer P, Bamber JC.
High-resolution ultrasound reflex transmission imaging and digital
photography: Potential tools for the quantitative assessment of pigmented lesions. Skin Res Technol 2006;12:50–59.
Roberts DL, Anstey AV, Barlow RJ, Cox NH, Newton Bishop JA,
Corrie PG, Evans J, Gore ME, Hall PN, Kirkham NUK. guidelines
for the management of cutaneous melanoma. Br J Dermatol 2002;
146:7–17.
Serrone L, Solivetti FM, Thorel MF, Eibenschutz L, Donati P,
Catricala C. High frequency ultrasound in the preoperative staging
of primary melanoma: A statistical analysis. Melanoma Res 2002;
12:287–290.
Solivetti FM, Thorel MF, Di Luca Sidozzi A, Bucher S, Donati P,
Panichelli V. Ruolo dell’ecografia con alta definizione ed elevata frequenza nella determinazione dello spessore tumorale del melanoma
maligno cutaneo. Radiol Med (Torino) 1998;96:558–561.
Tacke J, Haagen G, Hornstein OP, Huettinger G, Kiesewetter F,
Schell H, Diepgen TL. Clinical relevance of sonometry-derived
tumour thickness in malignant melanoma: A statistical analysis.
Br J Dermatol 1995;132:209–214.
Ulrich J, Petereit S, Gollnick H. Präoperative sonographische melanomdiagnostik vergleich von 7,5- und 20-MHz-Sonographie. Ultraschall
Med 1999;20:197–200.