2.1.

Participants

2.2.

In Vivo Fluorescence/Reflectance Confocal Microscopy

Reflectance and fluorescence images were obtained using a commercially available confocal laser scanning microscope [VivaScope 1500 Multilaser (Lucid Inc., Rochester, New York)]. A comprehensive review of the optical principles of the reflectance-only VivaScope has been published elsewhere.¹⁵ The multilaser system differs in that it is equipped with three lasers with wavelengths of 488 nm (blue), 658 nm (red), and 785 nm (near-infrared) and three corresponding filter sets. Each filter set is mounted in a bar. The user inserts the filter bar into the side of the confocal module. After application of a specific fluorescent dye, the compatible laser is used to illuminate the tissue, thereby exciting the appropriate fluorophore within the specimen. Spectral filtering can be selected while imaging the same portion of the skin without stopping the scanning session. The position of the filter bar enables imaging in the different modes. Each filter bar has three positions that allow, respectively, all returning light, only reflectance or only fluorescence, to reach the detector. In one position, reflected or backscattered laser light is eliminated by a fluorescence band-pass filter, whereas in another position, fluorescence is eliminated by a reflectance band-pass filter. In the third position, no filter is present, allowing both reflectance and fluorescence signal to return to the detector. In all filtering configurations, a pinhole attenuates the light from out-of-focus planes. The scanned field of view is 500 μm × 500 μm, producing images of 1000×1000 pixels. Image magnification is dependent on screen size. When displayed on the standard 19-inch monitor, magnification is approximately 600×. Axial resolution (section thickness) is <5μm. By moving the objective lens in the z (vertical) plane with respect to the skin surface, it is possible to image at different horizontal directions (x and y axis) within the tissue. By zeroing the imaging depth at the level of the most superficial layer before going vertically into the tissue, depth measurements can be obtained. A sequence of montage images (“VivaBlock”) can be acquired at constant depth. Each montage corresponds to a horizontal section at the user-selected depth with up to 8×8 mm total mapped field of view and axial resolution (section thickness) of <5 μm.

A 0.5% solution of indocyanine green (ICG-Pulsion, Pulsion Medical Systems, Munich, Germany) was used as a fluorescent dye for the labeling of human skin structures. Approximately 20 μl of the solution were intradermally injected under standardized conditions using a 0.3 ml insulin syringe fitted with a 30-gauge needle. The syringe was placed at an angle of 5 to 10 deg with respect to the skin surface. Minimal pain is associated with administration of the dye. Imaging was performed about 60 min after introduction of the fluorescent contrast agent. This time frame was chosen after observations in a pilot study evincing a homogeneous distribution from that time onward. To study the kinetics of ICG in healthy human skin, images were also taken 24 and 48 h after injection. Between the imaging sessions, the injection site was light-protected using a band-aid (Curapor^®, 7 × 5 cm).

The standard procedure for in vivo scanning with the VivaScope has been previously described.¹⁶ ICG has a maximum absorption rate at a wavelength around 780 nm and the fluorescence emission is peaking at about 805 nm. The tissue was illuminated with near-infrared laser light (785 nm). Automatic image control was active, which allows the VivaScope to automatically optimize the laser power so that the image is displayed with the proper illumination. In reflectance mode (RM), the automatically adjusted laser power ranged between 0.4 and 1.2 milliwatts (mW). In fluorescence mode (FM), the laser power ranged between 12.2 and 14.4 mW. Representative images at different layers of epidermis and dermis were acquired by varying the imaging depth. By changing the corresponding filter position, reflectance and fluorescence images were consecutively obtained from the same area and level within the tissue. We evaluated the depicted representations from both in vivo confocal imaging modes. The color image in Fig. 9 was processed with the public domain software ImageJ (National Institute of Health, Bethesda, Maryland). Green and red color was used to highlight reflectance and fluorescence modes, respectively. There was no processing of any other images with any software before analysis and reproduction for this paper.

Fig. 9

Dermal papilla of a pigmented seborrhoeic keratosis is presented at 50 μm below the skin surface. (a) RM image exhibits a bright rim of highly refractive melanin-containing cells surrounding the dermal area. (b) In FM, the fluorescence signal delineates epidermal keratinocytes and illuminates the relatively acellular dermis. (c) Reflectance (green color) and the fluorescence signal (red color) of (a) and (b) are merged (scale bar, 50 μm).

For quantification of ICG fluorescence in different layers of the skin, stacks of VivaScope images with a step increment of 4.5 μm were analyzed using ImageJ. Measurement areas with homogeneous fluorescence in-between skin grooves were defined and the mean intensity of fluorescence was determined. Data were exported to MS-Excel™ and GraphPad Prism 5.01™ for high resolution graphs.

3.1.

Healthy Human Skin

In the epidermis, a marked outline of keratinocyte cell borders is seen giving a clear impression of the overall morphology (cell architecture, cell size, and shape). At no time after injection of ICG, was a clearly visible fluorescence signal detectable in the cornified layer (data not shown). The applied fluorescent dye remained in the viable epidermis and did not diffuse upward through the barrier of the stratum granulosum-stratum corneum interface. Images were taken 60 min after injection. The results show an RM image and the corresponding fluorescence mode (FM) image from the same skin site and a level of 5 μm below the skin surface [Figs. 1a and 1b]. In both modes, cells of the stratum granulosum measuring 25 to 35 μm in size can be identified. Whereas, in RM, dark oval nuclei and a bright grainy cytoplasm can be observed, in FM the perceptibility has shifted in favor of a better outline of the cell border. Keratinocytes in the underlying layer, the stratum spinosum, are appreciably smaller (15 to 25 μm) and arranged in a tight honeycomb pattern. The cell borders can only partially be detected in RM [Fig. 1c]. In FM, the images clearly and lucidly show every single and polygonal keratinocyte [Fig. 1d]. The extracellular localization of the fluorescent dye encircles the cells with a bright shining contour. As imaging depth is increased toward the basal layer, the RM indicates oval areas containing strongly refractive cells [Fig. 1e]. They correspond to the top of the dermal papillae comprised of melanocytes and melanin-rich keratinocytes. In FM, those “hilltops” of the papillary dermis can be revealed as well, but in that case showing a rather confluent and homogeneous bright signal [Fig. 1f]. In RM, the lucid signal is caused by reflection of melanin in single cells. In FM, however, the signal occurs due to an enriched ICG distribution in the relatively acellular dermis. Therefore, the dermal papillae appear as bright round structures interposed among the deeper epidermal layers. Moving the focal plane a few microns deeper, the dermal papillae are truncated in the horizontal RM image and display a demarcated rim of refractive cells [Fig. 2a], still corresponding to melanin-rich basal cells and melanocytes. Comparing these circumferentially arranged basal cells in the FM, their cytoplasm appears dark, whereas the contrast agent occupies the extracellular space [Fig. 2b]. The reason being that backscattered light from refractive structures in RM is blocked in FM by the fluorescence band-pass filter through which only the fluorescent signal gets passed. The bright contrast within the papillae already corresponds to dermal structures, whereas capillary vessels appear dark. The duct of an eccrine sweat gland is demonstrated. In RM, it is visible as a diffuse whitish structure but in FM the outline of every single cell within the duct can be appreciated [Figs. 2a and 2b]. The RM image of the dermal-epidermal junction at 65 μm shows dermal and epidermal structures [Fig. 2c]. Connective tissue fibers, capillary loops, and interstitial keratinocytes can only be anticipated as opposed to the FM. Therein, fibrous structures that we assume to be collagen fibers and blood vessels are clearly contoured and the morphology of basal cells is still strongly evident [Fig. 2d]. The RM indicates the collagen-rich papillary dermis at 80 μm below the skin surface [Fig. 3a]. In the corresponding FM, these assumingly collagen fibers are sharply demarcated and for the first time, bright roundish structures of variable size are visible that cannot be identified in RM [Fig. 3b]. At present, it is not clear whether these could be fibroblasts and/or mast cells that are surrounded by the dye and/or whether these may be macrophages enclosed by ICG or having absorbed the dye. This finding is of significant interest and the scope of an ongoing investigation. At an imaging depth of 105 μm, the RM image is considerably degraded, whereas in the FM the fibrous structures are, even though less bright, still sharply circumscribed [Figs. 3c and 3d]. Figure 4 demonstrates an intensity profile of ICG one hour after injection showing maximum fluorescence at about 30 μm depth and a detectable signal down to at least 120 μm.

Fig. 1

(a) RM image of the stratum granulosum 5 μm below the skin surface. (b) In FM extracellular spaces are filled with ICG. Remnants of corneocytes (asterisk) appear brighter in the RM. (c) RM opposed to (d) FM of the upper stratum spinosum at a depth of 15 μm. (e) RM of the lower stratum spinosum at 25 μm demonstrates areas of bright melanin-containing cells (arrows). (f) In FM the strong luminosity at the same location (arrows) occurs because of the ICG enriched underlying dermis (scale bar, 50 μm).

Fig. 2

(a) RM versus (b) FM of the dermal-epidermal junction at 40 μm [(bright rim of melanin-rich basal cells and melanocytes (arrows), eccrine sweat gland (asterisk)]. At a depth of 65 μm (c) RM is compared with (d) FM; dermal fibrous structures (asterisk), capillary loops (arrows) (scale bar, 50 μm).

Fig. 3

In the dermis at 80 μm, (a) RM is opposed to (b) FM. In FM, fibers are sharply outlined; bright roundish structures of variable size are only present in the FM. (c) RM versus (d) FM at 105 μm (scale bar, 50 μm).

Fig. 4

Intensity profile of ICG within epidermis and dermis down to 200 μm depth. Image stacks of ICG fluorescence were acquired one hour after injection of the dye with 4.5 μm increments and quantified with ImageJ (n = 9, error bars represent SEM).

The distribution of ICG in the epidermis remained stable for at least 8 h in all of the investigated participants. From then on, the contrast agent slowly started to redistribute from the extracellular space into the cytoplasm of the keratinocytes. This gradual redistribution is substantiated by the fact that areas with distinct intracellular ICG, as well as areas with still only extracellular dye shaped cells, can be identified after 24 h [Fig. 5b]. In comparison to the RM image, the FM still shows the outline of each single cell and, moreover, a clear differentiation between the cell border, the cytoplasm, and the nucleus can be revealed in some areas [Figs. 5a and 5b]. The fluorescence intensity in epidermis and dermis was still comparable with the intensity observed after 60 min (data not shown).

Fig. 5

(a) RM and (b) FM 24 h after ICG application; eccrine gland (asterisk). FM shows a redistribution of the fluorescent dye. (c) RM compared to (d) FM 48 h later. Different representation of an eccrine sweat gland (asterisk) in both modes. All images display the stratum spinosum at 20 μm (scale bar, 50 μm).

Although the fluorescent signal was visibly decreased after 48 h, it was still powerful enough to provide a superior contrast compared to reflectance. To illustrate the difference, we opposed an eccrine duct in both modes [Figs. 5c and 5d]. In RM, it is displayed as a roundish, grainy, and partly whitish structure, whereas in FM the sweat gland is well demarcated and even the inner part of the acrosyringium is sharply contoured and still strongly fluorescent. Imaging of any layer 72 h after introduction of the dye yielded no significant signal anymore (data not shown). The results were reproducible and consistent for all 10 subjects. The application of ICG was safe and well tolerated and no adverse event occurred in any of the participants. A slight green discoloration remained visible at the injection site but completely disappeared a few days after intradermal administration. To show that introduction of ICG did not change the reflectance properties of the skin, we opposed RM images of the injected tissue site to RM images of noninjected untreated skin (Fig. 6). Images of untreated skin were taken during the same scanning session from ICG-free areas surrounding the injected site.

Fig. 6

RM images of ICG-injected skin [(a), (c), (e), (g), (i), (k), (m)] of different skin layers are confronted with RM images of noninjected untreated skin [(b), (d), (f), (h), (j), (l), (n)] of the same patient. No clinically significant alteration can be objectified (scale bar, 50 μm).

3.2.

Skin Pathology

In order to demonstrate the clinical applicability of this technique, we also applied ICG to lesional human skin and subsequently performed RM and FM images. Figure 7 shows a mapped field of the left margin of a nodular basal cell carcinoma at 50 μm below the skin surface in RM and FM in Figs. 7a and 7b, respectively. Already in the overview, typical tumor nests can be identified, which are characterized by polarization of epithelial cells at the periphery of the islands. But not all of these tumor islands can be recognized with the same accuracy in both confocal modes. A detailed section of the left lower part of the lesion demonstrates a lateral tumor branch with palisading tumor cells that can be detected with more precision in FM [Fig. 7d] than in RM [Fig. 7c].

Fig. 7

(a) RM and (b) FM image of the left margin area of a basal cell carcinoma 50 μm below the skin surface (scale bar, 250 μm). The red frame depicts one section with greater magnification in (c) and (d). (c) In RM, several tumor islands on the right image border can be observed. On the left image border, a whitish structure with no clear cell demarcation is displayed (asterisk). (d) In FM, this structure can be clearly identified as lateral tumor branch cohering to the other tumor nests. Moreover, every single tumor cell and especially the polarization of these cells at the periphery of the islands can be appreciated (scale bar, 100 μm). Control images of healthy human skin at 50 μm in [(e),(g)] RM and [(f),(h)] FM are presented.

Key RCM features of actinic keratosis include severe disruption of the epidermal architecture, keratinocyte pleomorphism, and parakeratosis. The extracellular distribution of the fluorescent dye allows visualization of architectural disarray as well as of atypical keratinocytes varying in size and orientation with greater accuracy in FM [Fig. 8b] than in RM [Fig. 8a]. Parakeratosis is a sign of abnormal keratinization marked by the retention of nuclei in the stratum corneum. Highly refractive parakeratotic nuclei can be identified in RM [Fig. 8e], whereas in FM [Fig. 8f] they are barely visible. As mentioned above, the fluorescent dye is hardly able to pass the barrier of the stratum granulosum-stratum corneum interface after intradermal injection, which does not permit cell demarcation or significant uptake of ICG in the cornified layer. The low signal that can be detected in FM at some parakeratotic areas may be more likely caused by autofluorescence or by strong reflecting light that manages to defeat the filter rather than by fluorescence due to dye diffusion. We then used a pigmented seborrhoeic keratosis to exemplify the synergistic effect of RCM and FCM, especially in pigmented skin lesions: Fig. 9 shows a dermal papilla with melanin-rich circumferentially arranged basal keratinocytes and melanocytes. These cells appear bright in RM [Fig. 9a] and dark in FM [Fig. 9b] as backscattered light from refractive structures in RM is blocked in FM by the fluorescence band-pass filter. In Fig. 9c, RM and FM images are merged with two different colors (RM-green, FM-red) and the information of both techniques is combined. This representation now enables a clear distinction between the backscattering signal of melanin-containing cells and fluorescence in dermal and intercellular areas within one image. A further application of confocal microscopy is the evaluation of dermal vessels. Among other characteristics, psoriasis, a chronic inflammatory disease, is marked by uncontrolled angiogenesis and therefore served as a model to assess vascular changes. Figure 10 shows a psoriatic lesion at 65 μm below the skin surface at the dermal-epidermal junction. In RM [Fig. 10a], the capillaries within the papillae can only be identified to some extent, whereas in FM [Fig. 10b], an increase in vascularity can be observed with more precision. Every single capillary can be appreciated and counted, the thickness can be measured, and the course of the vessel can be tracked. Additionally, a concentric formation of keratinocytes encasing the enlarged papillae can be observed in detail.

Fig. 8

(a) RM and (b) FM of an actinic keratosis at 20 μm below the skin surface are displayed. The extracellular uptake of ICG outlines pleomorphic atypic keratinocytes with greater accuracy in FM than in RM. Control images of healthy human skin at 20 μm in (c) RM and (d) FM (scale bar, 50 μm). In the stratum corneum only the (e) RM enables to highlight the backscattering signal of parakeratotic nuclei (arrows). Note that demarcated cells in the right lower area of the corresponding (f) FM image already belong to the stratum granulosum. Control images of healthy human skin in (g) RM and (h) FM (scale bar, 20 μm).

Fig. 10

(a) RM and (b) FM images of a psoriatic lesion at 65 μm below the skin surface are presented. In FM capillary vessels of variable size can be clearly identified and their course can be followed within the papillae. Concentrically arranged keratinocytes are contoured by the extracellular distributed fluorescent dye. Control images of healthy human skin at 65 μm in (c) RM and (d) FM showing single capillary loops (arrows) (scale bar, 50 μm).