in vivo reflectance confocal microscopy of erythematosquamous skin diseases
TRANSCRIPT
In vivo reflectance confocal microscopy oferythematosquamous skin diseases
Silvia Koller1, Armin Gerger2, Verena Ahlgrimm-Siess1, Wolfgang Weger1, Josef Smolle3 and
Rainer Hofmann-Wellenhof1
1Department of Dermatology, Medical University of Graz, Graz, Austria;2Department of Internal Medicine, Division of Oncology, Medical University of Graz, Graz, Austria;3Institute for Medical Informatics, Statistics and Documentation, Medical University of Graz, Graz, Austria
Correspondence: Silvia Koller, MD, Department of Dermatology, Medical University of Graz, Auenbruggerplatz 8, A-8036 Graz, Austria,
Tel.: +43-316-385-2423, Fax: +43-316-385-2466, e-mail: [email protected]
Accepted for publication 29 October 2008
Background: In vivo reflectance confocal microscopy (RCM)
represents a promising imaging tool that allows a non-invasive
examination of skin morphology in real time at nearly histological
resolution, showing microanatomical structures and individual
cells.
Objectives: The aim of our study was to evaluate the diagnostic
accuracy of confocal examination of erythematosquamous skin
diseases, to define typical RCM-features and assess them for their
presence or absence, diagnostic performance and reliability.
Methods: Three independent observers received standardized
instructions about diagnostic RCM-features of
erythematosquamous skin diseases. A total of 1700 RCM images
obtained from 75 patients with psoriasis, contact dermatitis,
mycosis fungoides, chronic discoid lupus erythematosus (CDLE)
or subacute cutaneous lupus erythematosus (SCLE) and from
10 ‘healthy adults’ without any skin disease were evaluated by
each observer.
Results: Overall, sensitivity and specificity values as observed by
three observers were, respectively, 89.13% and 95.41% for
psoriasis; 83.33% and 92.31% for contact dermatitis; 62.96% and
94.53% for SCLE ⁄ CDLE; and 63.33% and 92.89% for mycosis
fungoides.
Conclusions: Reflectance confocal microscopy examination
appears to be a promising method for non-invasive assessment of
erythematosquamous skin diseases. This study provides a set of
well-described morphological criteria with obvious diagnostic
impact, which should be used in further investigations.
Key words: contact dermatitis – lupus erythematosus – mycosis
fungoides – psoriasis – reflectance confocal microscopy
Please cite this paper as: In vivo reflectance confocal microscopy of erythematosquamous skin diseases. Experimental Dermatology 2009; 18: 536–540.
Introduction
Erythematosquamous skin diseases such as psoriasis, contact
dermatitis, mycosis fungoides and chronic discoid lupus
erythematosus (CDLE) ⁄ subacute cutaneous lupus erythe-
matosus (SCLE) may be difficult to distinguish by clinical
assessment. Until now, histopathological assessment is
adjudged as the gold standard for the diagnosis of ery-
thematosquamous skin diseases. In vivo reflectance confocal
microscopy (RCM) represents a promising imaging tool,
which provides a non-invasive window into living skin at
nearly histological resolution viewing microanatomical struc-
tures and individual cells. Initial research has concentrated
on the most clinically relevant cutaneous malignancies. In
the past, especially melanocytic skin tumors and non-mela-
noma skin cancer have been investigated by RCM (1–5). It
has also been shown to be useful as a biopsy guide tool, to
determine in advance the best respectively most significant
location for taking a punch biopsy to avoid taking multiple
and ⁄ or sequential skin biopsies, which are often needed to
establish a diagnosis. Furthermore, RCM was also used to
control the response to treatment in the past. Also topics like
pigmentation, photodamage and ageing were investigated by
RCM (6,7). There exist only a few publications in which
RCM has been used for the evaluation of inflammatory skin
conditions (8–15). RCM has the potential to assess and visu-
alize various histopathological features of erythematosqua-
mous skin diseases in vivo. This study systematically validates
RCM in diagnosing erythematosquamous skin diseases in an
observer-blinded manner. Therefore, we evaluate morpho-
logical features determined by RCM for their presence or
absence, diagnostic performance and reliability.
DOI:10.1111/j.1600-0625.2008.00822.x
www.blackwellpublishing.com/EXDOriginal Article
ª 2009 The Authors
536 Journal compilation ª 2009 Blackwell Munksgaard, Experimental Dermatology, 18, 536–540
Materials and methods
SubjectsEighty-five patients were recruited prospectively from the
Department of Dermatology, Medical University of Graz,
Austria over a period of 8 months. They gave informed
consent for the examination of their skin disease by RCM.
All institutional rules governing clinical investigation of
human subjects were strictly followed. We conformed to
the Helsinki Declaration with respect to human subjects in
biomedical research. Overall, 75 patients with psoriasis (27
lesions), contact dermatitis (20 lesions), mycosis fungoides
(10 lesions) or CDLE ⁄ SCLE (four CDLE and 14 SCLE
lesions) and 10 ‘healthy adults’ without any skin disease
were included in our study. All of them were imaged using
a commercially available in vivo reflectance confocal micro-
scope (Vivascope 1000; Lucid, Rochester, NY). Neither the
erythematosquamous skin diseases selected in any way for
their RCM features, nor any erythematosquamous skin dis-
ease lacking particular RCM characteristics was excluded
from the study set. Most of the patients underwent a histo-
pathological assessment in the past. When the diagnosis
had not been already established, a punch biopsy was per-
formed after clinical and confocal examination and was
subjected to standard histopathological assessment. Some
of the patients, whom we had recruited for our study, had
a positive patch test; therefore, in a few cases, we were able
to avoid taking a punch biopsy for further histopathologi-
cal assessment.
In vivo reflectance confocal microscopyTechnical equipment was used as described previously
(16,17). All images obtained by RCM in this study corre-
spond to sections in the horizontal plane. In each case, the
adapter ring was placed into the centre of the visible skin
disease margins. The whole field of view was imaged in x, y
and z-axis irrespective of diagnostic morphological RCM
features. Overall, about 7000 RCM images were taken and
stored in BMP file format. For each case, four captures (of
stratum corneum, stratum granulosum, stratum spinosum
and of the dermoepidermal junction zone) and one repre-
sentative stack (16 captures at intervals of 5 lm, starting
from stratum corneum to upper dermis) were preselected.
Diagnostic morphological reflectance confocalmicroscopy featuresOverall, 25 morphological RCM features of erythematosqu-
amous skin diseases were selected and assessed according
to recently published studies (8–15). Architectural and
cellular patterns were taken into account for diagnostic
decisions. All morphological features were defined a priori
without reference to the image set of this study.
Training data and study settingThree independent clinical dermatologists with moderate
experience in RCM received a standardized instruction about
diagnostic RCM features of erythematosquamous skin dis-
eases for 1 h by PowerPoint presentation. Overall, 50 image
examples of specific morphological features were demon-
strated for training purposes. For the diagnostic assessment
of the test set, the whole image set of each of the ery-
thematosquamous skin diseases was shown on a computer
screen and evaluated as belonging either to psoriasis, contact
dermatitis, mycosis fungoides, CDLE ⁄ SCLE or healthy skin
by each of the observers. Furthermore, the presence or
absence of each of the morphological features was assessed
by each observer irrespective of the assumed diagnosis in the
whole set of diagnostic images. To ensure strict separation of
learning and test set, none of the specimens used in the
training sample was used in the test set. All of the observers
were blinded in regard to the clinical or histopathological
diagnosis of the erythematosquamous skin diseases.
Statistical analysisStatistical analyses [sensitivity, specificity, positive predic-
tive value (PPV), negative predictive value (NPV), median
value, mean value, standard deviation and j statistic] were
performed by using SPSS statistical software package for
Windows (version 12.0; SPSS Inc, Chicago, IL, USA).
Reliability data (interobserver agreement) were produced in
the form of the j statistic. Kappa (j) takes a value between
0 (no agreement) and 1 (perfect agreement); therefore, it
was assumed that reliability was highly specific when j was
>0.8, excellent when j was >0.6, moderate when j was
>0.4 and poor when j was £0.4. For classification pur-
poses, we used the Classification and Regression Tree
(CART) software (version 4.0; Salford Systems, San Diego,
CA, USA) as described previously (16,17).
Results
General observationsThe test set comprised 75 erythematosquamous skin diseases,
including 27 psoriasis (all histologically verified chronic pla-
que type), 20 contact dermatitis (nine histologically verified
and 11 positive patch tests), 10 mycosis fungoides (all histo-
logically verified, two patch-type, six plaque-type and two
tumor-type), four CDLE and 14 SCLE (all histologically veri-
fied). The remaining 10 samples showed ‘normal’ healthy
skin of patients without any skin disease. Thus, the test set
comprised a variety of erythematous skin diseases.
Qualitative description of RCM criteriaIn general, erythematosquamous skin diseases could be
delineated clearly from healthy skin. In psoriatic lesions,
Confocal examination of erythematosquamous skin diseases
ª 2009 The Authors
Journal compilation ª 2009 Blackwell Munksgaard, Experimental Dermatology, 18, 536–540 537
characteristics such as parakeratosis, elongated and
increased dermal papillae with tortuous, twisted, dilated
capillary loops (Fig. 1a) were found. Inflammatory cells
(Fig. 1b) widespread within the epidermis and upper
dermis, focally forming microabscesses (Fig. 1c) were also
identified. In addition, architectural patterns like disarray
respectively disarrangement of stratum corneum were
detected. Transepidermal migration of inflammatory cells,
epidermal oedema with vesical formation ⁄ disruption
(Fig. 2a,b) and dermal vasodilatation could be observed in
contact dermatitis lesions.
Skin lesions in patients with CDLE or SCLE showed
parakeratosis, interface dermatitis (Fig. 3a), epidermotropic
atypical lymphocytes (Fig. 3b) and blurred intercellular
borders – focally with degeneration of keratinocytes and a
loss of regular epidermal stratification. Even Langerhans
cells were visualized by RCM imaging.
(a)
(b)
(c)
Figure 1. (a) Psoriasis: elongated and increased dermal papillae with
tortuous, twisted, dilated capillary loops. (b) Psoriasis: neutrophils
infiltrating stratum corneum. (c) Psoriasis: inflammatory cells focally
forming microabscesses.
(a)
(b)
Figure 2. (a) Contact dermatitis: intraepidermal vesicle formation.
(b) Contact dermatitis: intraepidermal vesicle formation.
(a)
(b)
Figure 3. (a) SCLE: interface dermatitis. (b) CDLE: epidermotropic
atypical lymphocytes.
Koller et al.
ª 2009 The Authors
538 Journal compilation ª 2009 Blackwell Munksgaard, Experimental Dermatology, 18, 536–540
The examination of mycosis fungoides lesions revealed
an interface dermatitis and an infiltration of the upper
epidermal layers by roundish cells, distributed in nests or
diffusely widespread throughout the epidermis, represent-
ing epidermotropic atypical lymphocytes. In tumor type
lesions, large pleomorph cells were detected (Fig. 4).
Sensitivity and specificityOverall, sensitivity and specificity values as observed by the
three observers were, respectively, 89.13% and 95.41% for
psoriasis (PPV 90.20%, NPV 94.84%); 83.33% and 92.31%
for contact dermatitis (PPV 77.35%, NPV 94.91%); 62.96%
and 94.53% for SCLE ⁄ CDLE (PPV 75%, NPV 90.59%);
and 63.33% and 92.89% for mycosis fungoides (PPV
55.05%, NPV 94.98).
A sensitivity and specificity value of 96.67% and 99.56%
(PPV 96.97%, NPV 99.56%), respectively, could be
achieved for healthy skin.
Diagnostic impact and reliability of morphologicalfeaturesWhen the presence or absence of morphological features
was assessed by the three observers, classification tree soft-
ware (CART; Salford Systems) was applied on the data set
to search for optimal split features, which facilitate an opti-
mal classification of all erythematosquamous skin diseases.
Furthermore, CART automatically performed a ranking of
all features that depended on their diagnostic value. In the
order of the analysis ranking, the results indicated that
mainly increased number of dermal papillae, tortuous,
twisted and dilated capillary loops, and elongated papillae
were taken into account for diagnostic classification by the
software for psoriasis lesions.
Intraepidermal vesical formation in the upper dermis as
well as in the dermoepidermal junction zone presented the
most important RCM feature for contact dermatitis lesions.
Interface dermatitis, architectural epidermal disarray, epi-
dermotropic atypical lymphocytes and dilated blood vessels
were identified as good working features for diagnosing
CDLE ⁄ SCLE lesions.
Epidermotropic atypical lymphocytes, interface dermati-
tis, pleomorph tumor cells and dendritic cells were detected
to be specific for mycosis fungoides lesions. In contrast,
disarray and disarrangement of stratum corneum, loss of
skin folds, prominent nucleoli, intraepidermal disruption,
Kogoj’sche microabscesses, Pautrier’s microabscesses and
spongiosis had less to no diagnostic importance.
Overall, using only the assessed presence or absence of
the top confocal features, the CART software correctly clas-
sified 100% of healthy skin lesions, 82.72% of psoriasis
lesions, 81.67% of contact dermatitis lesions, 88.89% of
CDLE ⁄ SCLE lesions and 90% of mycosis fungoides lesions.
When each feature was measured for its reliability (inter-
observer agreement) by using the j statistic, the results
showed that most of the diagnostic criteria were (moderate
to excellent) reliable, indicating good definitions of the
morphological features (Fig. 5).
Discussion
Reflectance confocal microscopy is a promising, high-reso-
lution imaging technique that opens a window into living
tissue. It is adjudged to be a safe procedure with no
evidence of tissue damage from the low level energy laser
beams. RCM offers the unique opportunity to analyse
skin structures non-invasively at a ‘quasi-histopathologic’
resolution.
In our study, we demonstrate the application of RCM
for diagnostic examination of erythematosquamous skin
diseases. A large number of images of erythematosquamous
Figure 4. Mycosis fungoides: roundish and large pleomorph cells
diffusely widespread throughout the epidermis.
Figure 5. Reliability (interobserver agreement) of morphological
features with the highest diagnostic impact, ranked by CART analysis.
Confocal examination of erythematosquamous skin diseases
ª 2009 The Authors
Journal compilation ª 2009 Blackwell Munksgaard, Experimental Dermatology, 18, 536–540 539
skin diseases were evaluated in an observer-blinded manner
to determine sensitivity and specificity of this method.
Three independent clinical dermatologists with moderate
experience in RCM received a standardized instruction
about diagnostic RCM features of erythematosquamous skin
diseases for 1 h by PowerPoint presentation. It is important
to note that the confocal morphological features that were
used for evaluating the test set, are easy to learn and use,
and this was reflected in the interobserver agreement.
The analysis was sterile and artificial that no clinical or
histological diagnosis was taken into account. Conse-
quently, morphological features could be assessed according
to well-known criteria used in conventional histopathology.
Moreover, confocal and histopathological morphology
seems to correspond well, as conventional microscopy fea-
tures of erythematosquamous skin diseases and normal skin
were applied to confocal image examination.
The main reason for the low sensitivity values of
CDLE ⁄ SCLE and mycosis fungoides is that both diseases
have the same diagnostic features (epidermotropic atypical
lymphocytes, interface dermatitis) in common. Therefore,
sometimes, a mix-up between those two entities occurred.
One limitation in the current state of technological RCM
development that has to be addressed is that the assessment
of microanatomic structures can be performed only to a
depth of 350 lm, which corresponds to the papillary dermis.
Therefore, processes in the reticular dermis cannot be evalu-
ated reliably. Furthermore, we could not conclude from the
results of three independent observers that similar classifica-
tion results would be achieved by the majority of dermatolo-
gists in everyday practice. Finally, in this study, each case
was represented by preselected images. It might be possible
that the evaluation of a larger number of images per case
might not add to the diagnostic accuracy, but on the
contrary, might distract the observers from the correct
diagnosis. This study provides a set of well-described
morphological criteria – with obvious diagnostic impact –
which should be used in future investigations. Until now,
histopathological assessment is adjudged as the gold
standard of diagnosis, but RCM represents – despite the
described limitations of the method – an exciting technology
for rapid imaging, which offers for the first time the unique
opportunity to analyse skin disease structures non-invasively
at a ‘quasi-histopathologic’ resolution at the bedside. Cellu-
lar and architectural details can be examined without having
to excise and process the tissue as in standard histology. At
that time, RCM can be used as a supportive tool to deter-
mine in advance the best, respectively, most significant loca-
tion for taking a punch biopsy, in regards of the patient’s
welldoing and to safe time. Furthermore, RCM represents a
promising tool to control the response to treatment. In con-
sideration of the fact that confocal and histopathological
morphology seemed to correspond well in our study, RCM
may achieve a similar significance in the future similar to
histopathological assessment.
Acknowledgements
This study was supported by the ‘Jubliaumsfond der Osterreichischen
Nationalbank’, project number 12319.
References
1 Pellacani G, Cesinaro A M, Seidenari S. Reflectance-mode confocal microscopyfor the in vivo characterization of pagetoid melanocytosis in melanomas andnevi. J Invest Dermatol 2005: 125: 532–537.
2 Gonzalez S, Tannous Z. Real-time, in vivo confocal reflectance microscopy ofbasal cell carcinoma. J Am Acad Dermatol 2002: 47: 869–874.
3 Agero A L, Busam K J, Benvenuto-Andrade C et al. Reflectance confocalmicroscopy of pigmented basal cell carcinoma. J Am Acad Dermatol 2006: 54:638–643.
4 Curiel-Lewandrowski C, Williams C, Swindells K et al. Use of in vivo con-focal microscopy in malignant melanoma. Arch Dermatol 2004: 140: 1127–1132.
5 Gerger A, Hofmann-Wellenhof R, Langsenlehner U et al. In vivo confocal laserscanning microscopy of melanocytic skin tumours: diagnostic applicabilityusing unselected tumour images. Br J Dermatol 2008: 158: 329–333.
6 Yamashita T, Akita H, Astner S et al. In vivo assessment of pigmentary andvascular compartments changes in UVA exposed skin by reflectance-modeconfocal microscopy. Exp Dermatol 2007: 16: 905–911.
7 Sauermann K, Clemann S, Jaspers S et al. Age related changes of human skininvestigated with histometric measurements by confocal laser scanning micros-copy in vivo. Skin Res Technol 2002: 8: 52–56.
8 Gonzalez S, Rajadhyaksha M, Rubinstein G, et al. Characterization of psoriasisin vivo by reflectance confocal microscopy. J Med 1999: 30: 337–356. NOS.5&6.
9 Astner S, Gonzalez E, Cheung A et al. Pilot study on the sensitivity and speci-ficity of in vivo reflectance confocal microscopy in the diagnosis of allergiccontact dermatitis. J Am Acad Dermatol 2005: 53: 986–992.
10 Gonzalez S, Gonzalez E, White W M et al. Allergic contact dermatitis: correla-tion of in vivo confocal imaging to routine histology. J Am Acad Dermatol1999: 40: 708–713.
11 Hicks S P, Swindells K J, Middelkamp-Hup M A et al. Confocal histopathologyof irritant contact dermatitis in vivo and the impact of skin color (black vswhite). J Am Acad Dermatol 2003: 48: 727–734.
12 Swindells K, Burnett N, Rius-Diaz F et al. Reflectance confocal microscopy maydifferentiate acute allergic and irritant contact dermatitis in vivo. J Am AcadDermatol 2004: 50: 220–228.
13 Astner S, Gonzalez E, Cheung A C et al. Non-invasive evaluation of the kinet-ics of allergic and irritant contact dermatitis. J Invest Dermatol 2005: 124:351–359.
14 Agero A L, Gill M, Ardigo M et al. In vivo reflectance confocal microscopy ofmycosis fungoides: a preliminary study. J Am Acad Dermatol 2007: 57: 435–441.
15 Ardigo M, Maliszewski I, Cota C et al. Preliminary evaluation of in vivo reflec-tance confocal microscopy features of discoid lupus erythematosus. Br J Der-matol 2007: 156: 1196–1203.
16 Gerger A, Koller S, Kern T et al. Diagnostic applicability of in vivo confocallaser scanning microscopy in melanocytic skin tumours. J Invest Dermatol2005: 24: 493–8.
17 Gerger A, Koller S, Weger W et al. Sensitivity and specificity of confocal laser-scanning microscopy for in vivo diagnosis of malignant skin tumours. Cancer2006: 107: 193–200.
Koller et al.
ª 2009 The Authors
540 Journal compilation ª 2009 Blackwell Munksgaard, Experimental Dermatology, 18, 536–540